University of Southern California Dissertations and Theses

Aging and the use of context and frequency information in ambiguity resolution

(USC Thesis Other)

Aging and the use of context and frequency information in ambiguity resolution

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Aging and the Use of Context and Frequency Information
in Ambiguity Resolution
by
Karen Lee Stevens
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(Psychology)
May 1996
Copyright 1996 Karen Lee Stevens
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 1381616
UMI Microform 1381616
Copyright 19%, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY O F SO U T H E R N C A LIFO R N IA
T H E GRADUATE SC H O O L
U N IV ER SITY PA RK
L O S A N G ELE S. C A LIFO R N IA 0 0 0 0 7
This thesis, w ritten by
K aren L. S t e v e n s __________ __________
under the direction of h ^ x Thesis Com m ittee,
and approved by all its members, has been pre
sented to and accepted by the D ean of The
Graduate School, in partial fulfillm ent of the
requirements for the degree of
Jdas.ter...o.£. .A rts..
THESIS COMMITTEE
^ -", ycrg
( p J j u d C s
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
List of Figures ill
List of Tables iv
Abstract V
Introduction 1
Experiment 1A 7
Experiment IB 25
Experiment 2A 34
Experiment 2B 48
General Discussion 55
References 60
Appendix A 63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Figure 1: Naming times to the visual target us in two age groups as 19
a function of ambiguity and context
Figure 2: Judgments of compatibility between auditory stimulus and 21
the visual target us in two age groups as a function of ambiguity and
context
Figure 3: Naming times to the visual target could in two age groups 43
as a function of context
Figure 4: Judgments of compatibility between auditory stimulus and 45
thevisual target could in two age groups as a function of context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Table 1: Note: The visual target appears in uppercase letters 13
Table 2: Note: The visual target appears in uppercase letters 39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
Abstract
Younger and older adults' use of context and frequency information to aid
ambiguity resolution were investigated. Two experiments presented older and
younger adults with lexical ambiguities ( e .g fires) in a sentence context which
supported either the more Sequent noun (warehouse fires) or less frequent verb
(corporation fires) interpretation of the ambiguity. Experiment 1 resolved the
ambiguity in favor of the low frequency verb meaning and Experiment 2 presented
a disambiguation in favor of the high frequency noun meaning. The two
experiments used an on-line cross modal naming task in which participants heard an
auditory sentence fragment up to an ambiguous word and then named a
disambiguating visual target. Following their naming response on each trial,
participants judged whether the visual target was a good continuation of the
auditory sentence. Finally use of context was tested in an untimed sentence
completion study in which participants read the stimuli up to the ambiguous word
and wrote a continuation. With respect to contextual information, the important
results were that while the younger group's responses showed a reliance on context
in all three tasks, the older group's reliance on context varied with the amount of
time available to use the context to resolve the ambiguities. Thus, older adults
showed no use of context in the cross modal naming task which demanded
immediate use of context, moderate use of context in the judgment task which
demanded less immediate use of context and full use of context, in the untimed
sentence completion task. In addition, indirect evidence suggests that older adults
relied on frequency information in the absence of the ability to use contextual
information. The delayed use of context and subsequent reliance on frequency in
older comprehenders is interpreted as the result of slowed language processing
operations in the elderly and the results are discussed with the respect other research
on aging and context use.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
One question which is at the heart of research on language comprehension
concerns how comprehenders resolve the substantial amount of ambiguity which
exists in natural languages. While comprehenders are extremely successful at this
task, only rarely noticing the existence of ambiguities, the centrality of this issue
to language comprehension can not be overstated as ambiguity resolution
permeates every stage of language comprehension, from low level processes such
as determining the locations of word boundaries in speech to high level
operations, such as identifying referents of pronouns in a discourse. Unfortunately
research in this area has typically focused on the manner and ease with which
younger comprehenders perform ambiguity resolution and very little research has
focused on older comprehenders' performance in ambiguity resolution. While
little previous research has looked into older comprehenders' ability to perform
ambiguity resolution, this issue is a quite interesting one which has great potential
merit in perhaps informing research into language comprehension difficulties in
older adults. Thus, it is extremely plausible, given the pervasive nature of the
ambiguity resolution problem, that failures to resolve ambiguities could be the
source of well documented difficulties for older individuals in language
comprehension (See Kemper, 1992, for review). The purpose of this paper then is
to investigate aging and the process of ambiguity resolution and hopefully open
an interesting dialogue between the two disciplines from which the two will
mutually benefit.
While many different types of ambiguities exist in the English language, the
specific type of ambiguity which our investigation will focus on is lexical
ambiguity. To explain then exactly what lexical ambiguity is, consider the words
"bank" and "watch." In the absence of any contextual information, the words
"bank" and "watch" are ambiguous as to the meaning which one wishes to convey
with them. Thus the word "bank" for example, while at first glance probably
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
invokes the "financial" sense of the word, the word may also invoke the "side of
the river" sense of "bank." Additionally, lexical ambiguities may also be
ambiguous as to which lexical category they belong. While this does not occur
with the word "bank," as both senses of the word are nouns, this does occur with
the word "watch" which may be either a noun, as denoted by the "jewelry" sense
of the word or a verb, as denoted by the "observation" sense of the word.
Of interest then to psycholinguistic researchers is the process by which
comprehenders resolve such ambiguities and invoke the correct sense of lexically
ambiguous words. The most recent research on lexical ambiguity resolution has
suggested that such ambiguities are resolved via the joint effects of frequency and
contextual information (Carpenter & Daneman, 1981; Seidenberg, Tanenhaus,
Leiman & Bienkowski, 1982; Simpson, 1981; 1984; Simpson & Burgess, 1985;
Tabossi, 1988). To explore just how frequency and contextual information
contribute to the resolution of lexical ambiguities, let us consider a computational
model implemented by Kawamoto (1993; Cottrell, 1989 offers a related account).
The connectionist model implemented by Kawamoto represents the alternative
meanings of each lexical ambiguity (in addition to other aspects of lexical
representation that are not relevant to the present discussion) as differential
patterns of activation over sets of units. The model is dynamic, such that the
activation of alternative interpretations of ambiguities builds up over time, in
contrast to strictly feed-forward models, which complete all computations in a
single timestep. The dynamic property of the model allows the effects of context
use during ambiguity resolution to be observed at different points in time. In
Kawamoto's computational model, the process of lexical ambiguity resolution
begins with the partial activation of the alternative meanings of the lexical
ambiguity as a function of their relative frequency of usage in the English
language. For example then, with the lexical ambiguity "bank," the "financial"
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sense of "bank" would receive greater activation relative to the "side of the river"
sense of "bank," as the financial usage of "bank" is far more frequent in the
English language (Francis and Kucera, 1982). As this rapid process is occurring,
contextual information also contributes to the resolution process by additionally
influencing the relative patterns of activation of the alternative senses of the
lexical ambiguity. Thus in relation to the lexical ambiguity "bank," contextual
information concerning either someone depositing money for instance, or
someone going fishing would increase the relative activation of one particular
interpretation of the lexical ambiguity. According to Kawamoto's model then, the
process of lexical ambiguity resolution ends when one particular interpretation of
the lexical ambiguity accrues greater activation than the alternative interpretation,
thereby causing the winning interpretation to be adopted for the input.
As evidenced in various empirical studies and depicted in Kawamoto's model,
lexical ambiguity resolution is easiest when contextual and frequency information
do not conflict with one another (Carpenter & Daneman, 1981; Galbraith &
Taschman, 1969). In the case of the lexical ambiguity "bank" for example, given
the high frequency advantage for the financial meaning of bank, when the lexical
ambiguity is embedded in a financial context or not preceded by any context,
invoking the financial interpretation of the lexical ambiguity should be a relatively
easy task. When however contextual and frequency information conflict with one
another, lexical ambiguity resolution is quite difficult (Carpenter & Daneman,
1981; Duffy, Morris & Rayner, 1988; Kawamoto, 1993). Thus using the lexical
ambiguity "bank" as an example once again, when the lexical ambiguity bank is
embedded in a river context, given the high frequency advantage for the financial
meaning of bank, frequency information serves to promote the financial sense
while context increases the activation of the river sense. As a result, it takes a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
longer period of time for one interpretation to achieve sufficiently greater
activation over the other, and ambiguity resolution is a more difficult task.
It is in these situations, in which frequency and contextual information
conflict with one another, that the process of lexical ambiguity resolution
becomes potentially interesting from an aging perspective. The reason that lexical
ambiguity resolution becomes of potential interest is that it is in this situation, in
which the processing system is maximally taxed, that breakdowns in lexical
ambiguity resolution as a result of aging are most likely to occur and the most
amenable to investigation. There is then some outside research which does admit
the possibility that such a breakdown to the lexical ambiguity resolution system
could potentially occur as a function of aging. This research, conducted by
Wingfield, Poon, Lombardi & Lowe (1985) evidenced that older adults had
slower speech/language processing operations relative to younger adults, which
may be part of the general cognitive slowing that some have hypothesized to be a
consequence of aging (Salthouse, 1982; 1985; 1988). This processing deficit,
independent of whether it is part of a more general slowing, when considered
within a computational framework such as Kawamotos, makes some interesting
predictions regarding differences in older adults' use of contextual and frequency
information relative to that of younger adults. Let us then explore what exactly
this two processing deficit means within a computational framework such as
Kawamotos and the specific nature of the predictions which such a computational
framework makes regarding the differences between older and younger adults' use
of contextual and frequency information in lexical ambiguity resolution as a result
of such a processing deficit.
Thus when considered within a computational framework such as
Kawamoto's, slower speech/language processing operations in older adults would
have the effect of slowing older adults' access to contextual and frequency
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
information during the lexical ambiguity resolution process. However, while
access to both frequency and contextual information would be relatively slowed,
older adults* access to contextual information would be the most slowed relative
to their access of frequency information. The reason for the differential slowing
of access between frequency and contextual information rests in the
computational demands that each requires. Frequency information is widely held
to be embedded in the mapping between a word’s phonological or orthographic
code and its semantic codes, and thus this information becomes automatically
available during the word recognition process (Kawamoto, 1993; Seidenberg &
McClelland, 1989). By contrast, contextual information is computed by
integration of the semantic representations of several words or sentences. In the
most minimal case, in which only one word provides context for an ambiguous
word, as in the sequence river bank, computation of the contextually appropriate
meaning requires recognition of river and activation of its semantic
representations, recognition of the word bank and activation of its semantic
representations, and the operation of combining the two sets of representations.
Slowed processing in this case would delay access to each piece of information
and would collectively slow the entire process of contextual integration. On this
view, slowed language processing operations would have a greater detrimental
effect on computation of the more complex contextual information than on
activation of simple frequency information.
On the basis of this account then, one might predict that the timecourse over
which older adults access contextual and frequency information would differ
relative to the timecourse with which younger adults access contextual and
frequency information. Thus one would predict that early in the process of
lexical ambiguity resolution, as language processing operations slow in older
adults, the easier to access frequency information would tend to dominate the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
more difficult to access contextual information and frequency information would
therefore act as the dominant contributor to the lexical ambiguity resolution
process. There has been some research conducted within memory paradigms
which does lend support to this idea. The results from that research evidenced
that older adults persisted in encoding words in their high frequency
interpretation, even when the context promoted a lower frequency interpretation
(Hess & Higgins, 1983; Micco & Masson, 1991). In younger adults however,
which do not experience such detrimental slowing to their language processing
system, both frequency and contextual information would contribute to the early
stages of the lexical ambiguity resolution process. In later stages of the lexical
ambiguity resolution process however, when more time is available for context
effects to develop in older adults, one would expect that older and younger adults
would have comparable access and use of contextual information to aid the
ambiguity resolution process.
In order to assess whether the timecourse over which contextual and
frequency information becomes available during the process of ambiguity
resolution differs between older and younger adults, two experiments were
implemented. These two experiments presented older and younger adults with
lexical category ambiguities ( e .g fires) in a sentence context which supported
either the noun (warehouse fires) or verb (corporation fires) interpretation of the
ambiguity. All ambiguous words were more frequent in their noun than verb
interpretations. Experiment 1 resolved the ambiguity in favor of the low
frequency verb meaning of the ambiguity and Experiment 2 presented a
disambiguation in favor of the high frequency noun interpretation. Both
Experiments 1 & 2 examined the process of lexical ambiguity resolution at three
different timepoints.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
EXPERIMENT 1A
The purpose of this experiment was to determine whether older and younger
adults could use context to resolve lexical ambiguities during on-line language
comprehension. The method used to assess subjects' interpretation of the lexical
ambiguities was a cross-modal naming task, which has previously been shown to
be sensitive to the time course of on-line processing that occurs during ambiguity
resolution (Tyler & Marslen-Wilson, 1977). In this task, subjects are presented
with an auditory sentence fragment. A lexically ambiguous word is embedded at
the very end of this auditory sentence fragment. At the offset of the auditory
presentation of the ambiguous word, a visual target word is presented on a screen
for subjects to name aloud. This visual target then continues the auditory
sentence fragment and disambiguates the lexical ambiguity to one particular
interpretation. A number of previous studies have evidenced that subjects'
naming times to the visual target reflect the degree to which the target sensibly
continues the auditory sentence fragment (Tyler & Marslen-Wilson, 1977; 1982a;
1982b). In the case of lexical ambiguity resolution, subjects' naming times were
short when the visual target disambiguated the lexical ambiguity in a way which
was consistent with subjects' interpretation of the lexical ambiguity and the visual
target was thus interpreted as a plausible continuation of the auditory sentence
fragment Naming times were long however, when the visual target
disambiguated the lexical ambiguity in a way which was inconsistent with
subjects' interpretation of the lexical ambiguity and the visual target was thus
interpreted as an implausible continuation of the auditory sentence fragment. To
assess subjects' interpretations of the lexical ambiguities, naming times to the
visual target were compared to a control condition in which interpretations of the
lexical ambiguities were unambiguous and the visual target was always
interpreted as a plausible continuation of the auditory sentence fragment.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To ensure however that the visual target truly disambiguated the lexical
ambiguity to a single interpretation, ambiguities were chosen in which the two
alternative interpretations of the lexical ambiguity came from different lexical
categories. The ambiguities which were chosen then all had a high frequency
noun interpretation and a low frequency verb interpretation (Francis and Kucera,
1982), and were words such as fires, guards, judges, etc. A single interpretation
of these lexical ambiguities can clearly be disambiguated by the syntactic
properties of the visual target. For example, the ambiguous word fires in the
sentence fragment The union told the reporters that the warehouse fires can be
disambiguated to the noun interpretation with the visual target could, because the
syntactic sequence fires could must be a noun + verb in English. Similarly, the
visual target us would disambiguate fires in favor of the verb interpretation,
because the syntactic sequence fires us is grammatical in English only as a verb +
pronoun.
In the experimental trials, the visual target word us always disambiguated the
ambiguity in favor of the low frequency verb interpretation. In the verb-
supporting context condition, a single context word within the sentence promoted
the ultimately correct verb interpretation of the lexical ambiguity prior to
presentation of the disambiguating visual target us. In the noun-supporting
context condition, a single context word within the sentence promoted the
incorrect noun interpretation of the lexical ambiguity prior to presentation of the
disambiguating visual target us. If comprehenders use context to aid ambiguity
resolution, then one would predict that an Ambiguity x Context Interaction should
occur. The nature of this interaction would be that when the context supports the
correct low frequency verb interpretation (verb-supporting context condition), the
visual target would disambiguate the lexical ambiguity in a way which was
consistent with subjects' interpretation of the lexical ambiguity and the visual
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
target us would thus form a plausible continuation of the auditory sentence. In
this case then, subjects' naming times would be expected to match naming dmes
for an unambiguous control, in which the visual target would similarly be
interpreted as a plausible continuation. When however, the context supports the
incorrect noun interpretation (noun-supporting context condition), the visual
target would disambiguate the lexical ambiguity in a way which was inconsistent
with subjects’ interpretation of the lexical ambiguity and the visual target us
would thus form an implausible continuation of the auditory sentence. In this
case then, subjects' naming times should be longer than naming times in the
unambiguous control condition and the verb-supporting context condition, as the
visual targets for the unambiguous control and verb-supporting context conditions
would in contrast form plausible continuations of the auditory sentence.
Alternatively, if comprehenders do not use context to aid ambiguity
resolution, then a Main Effect of Ambiguity should occur. The nature of this
main effect of Ambiguity would be that naming times in ambiguous conditions
(where frequency information always leads to the incorrect interpretation of the
ambiguity) should be longer than naming times in unambiguous conditions,
independent of context The pattern of effects across age would then reveal
whether older and younger comprehenders utilize context to the same degree in
on-line ambiguity resolution.
It should be noted that only strong contexts could be expected to influence
subjects’ naming times (Simpson, 1981; Simpson & Burgess, 1985; Simpson &
Krueger, 1991). In order to ensure sufficiently strong contexts, a stimulus set was
used in Experiment 1 which had previously evidenced an Ambiguity x Context
interaction for young comprehenders in a reading time paradigm. Thus in this
paradigm, in the noun-supporting context condition, subjects had long reading
times in the disambiguation region of ambiguous sentences compared to reading
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
times for unambiguous sentences, but in the verb-supporting context condition,
there was no effect of ambiguity on reading times (MacDonald, 1993, Experiment
2). Given then that these contexts appear to be of sufficient strength, one would
expect a similar type of interaction in the naming times of the young group using
the cross-modal naming task. Additionally, one would expect the same type of
interaction in the naming times of the older group if they are in fact able to use
context in on-line language comprehension. If however, the older group can not
use context in such a task, then as was aforementioned, one would expect a main
effect of Ambiguity to occur.
In addition to the aforementioned Anova analyses, a correlation analysis will
assess the extent to which the older group's naming times correlate with the
naming times of the younger group.
Serendipitously, the need for a second task in the cross-modal paradigm
provided an opportunity to collect more data on context effects in addition to that
provided by the naming time data. This second task, entitled Compatibility
Judgments, was implemented in the experiment to ensure that subjects were
paying attention to the auditory sentence fragment information and were treating
the visual target as a continuation of the auditory sentence fragment. Subjects in
this task were asked to judge whether or not the visual target made sense as a
continuation of the auditory sentence fragment e.g. "Does the word make sense in
the sentence?." This compatibility judgment was presented to subjects on the
computer screen just after they named the visual target and thus the data on the
compatibility judgments was collected a few seconds after subjects' naming
response. One might reason then that if the older adults have slowly-developing
context effects, these effects might be absent from naming data but might
evidence themselves in this slightly delayed compatibility judgment task. Thus
again, if context is used in ambiguity resolution at the point that the compatibility
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
judgments are made, then one would expect an Ambiguity x Context interaction
to occur. One would expect that when context supported the correct verb
interpretation (verb-supporting context condition), the percentage of visual targets
judged by subjects as compatible with the auditory sentence would be the same as
the percentage of visual targets judged by subjects as compatible with the auditory
sentence in the unambiguous control condition in which the visual targets always
form plausible continuations. When the context however supported the incorrect
noun interpretation (noun-supporting context condition), the percentage of visual
targets which subjects judged as compatible with the auditory sentence would be
significantly smaller then the percentage of visual targets which subjects judged
as compatible in the unambiguous control condition and the verb-supporting
context condition. Again, if context is not used in ambiguity resolution at the
point that the compatibility judgments are made, then one would expect a Main
Effect of Ambiguity, such that the percentage of visual targets judged as
compatible in the Ambiguous Condition should be significantly smaller than the
percentage of visual targets judged as compatible in the Unambiguous Condition,
regardless of context.
In addition to the aforementioned Anova analyses, a correlation analysis will
assess the extent to which the older group's compatibility judgments correlate with
the compatibility judgments of the younger group. Also, the extent to which the
older group's compatibility judgments correlate with their naming times and the
extent to which the younger group's compatibility judgments correlate with their
naming times will be assessed.
Method
Participants. The younger participant group was composed of thirty two USC
undergraduates (23 female, 9 males) who were either paid or received extra credit
points in an introductory psychology course for their participation. An additional
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ten participants were tested but were rejected due either to machine error or to the
fact that they were bilingual. All participants were native English speakers and
were naive to the purpose of the experiment. Due to an oversight, the exact age
and education level of the younger participants were not recorded; the
characteristic age of the younger participants was 20 years and the characteristic
number of years of formal education which the younger participants had received
was 13.5. As seen below, the older group had higher educational levels than the
younger group.
The older participant group was composed of thirty two USC alumni (17
female, 15 male) who were affiliated with S.A.G.E.. S.A.G.E. is composed of
over 1,000 USC alumni from the Los Angeles area who voluntarily donate their
time to participate in studies on aging research. The older participants were
compensated for parking and travel related expenses. An additional five
participants were tested but were rejected because of large numbers of missing
data points from failures of the participants’ voices to trigger the microphone. All
participants were native English speakers, naive to the purpose of the
psycholinguistic experiment, and in reportedly good health. The mean age of the
older participants was 72.3 years, SD = 4.6, and the older participants reported a
mean 16.9 years of formal education, SD = 1.9.
Materials. The stimuli used in the cross modal naming task were taken from
materials used in MacDonald (1993, Experiment 2) and can be found in Appendix
B of that paper. An example may be seen in Table 1. In the task, the stimuli were
constructed around 16 lexical category ambiguous words with both a plural noun
and a third person singular verb meaning (fires, benefits, guards, etc.). All of the
ambiguous words had a higher frequency noun interpretation than verb
interpretation; the mean was 6% verb interpretations in Francis and Kucera
(1982).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
Table 1
Example Stimuli, Experiment 1
Auditory Fragments, Verb-Supporting Context
Ambiguous The union told the reporters that
the corporation fires US
Unambiguous The union told the reporters that
the corporation could fire US
Auditory Fragments, Noun-Supporting Context
Ambiguous The union told the reporters that
the warehouse fires US
Unambiguous The union told the reporters that
the warehouse could fire US
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
As shown in Table 1, the prior context for the ambiguous word was manipulated
at two levels, verb-supporting and noun-supporting. These contexts differed by
only one word, the word immediately preceding the ambiguous word, e.g.
corporation/warehouse in the example in Table 1. The verb-supporting context
condition promoted the verb interpretation of the ambiguous word; this context is
consistent with the interpretation which is forced by the visual target us. For
example, the verb-supporting context word corporation in the phrase corporation
fires supports the verb interpretation of fire s, because it seems more plausible for
a corporation to fire people than for a corporation to have a fire. In contrast, the
noun-supporting context supports the noun interpretation of the ambiguous word;
this context is inconsistent with the interpretation which is forced by the visual
target us. For example, warehouse fires supports the noun interpretation of fires
because a fire in a warehouse seem more plausible than a warehouse firing people.
With the exception of the context word, the verb-supporting and noun
supporting contexts were identical. The context word immediately preceded the
ambiguous word, so that the auditory sentences ended with the sequence context
word, ambiguous word, followed immediately by the disambiguating visual
target. This sequence of adjacent context, ambiguous, and disambiguating words
created the smallest possible interval in which context effects could develop and
influence ambiguity resolution.
The unambiguous condition differed from the ambiguous condition in two
respects. The first way in which the two conditions differed was that the final 's'
on the ambiguous word was deleted, thus yielding/Ire, guard, etc. The second
way in which the two conditions differed was that the word could was added to
the critical pair of words, so that the ambiguous corporation/warehouse fires
became corporation could fire/warehouse could fire in the unambiguous
condition. The final word still has both noun and verb interpretations but
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
following the word could, the only grammatical interpretation is the verb
interpretation.
All experimental stimuli were identical to those in MacDonald (1993) up to
the ambiguous word, with one exception. One item in the original stimuli,
agency/bank accounts, is not grammatical with the us continuation. This item
was replaced by the ambiguous word judges and the verb supporting and noun
supporting context words lawyer and court.
In addition to the stimulus sentences constructed around the sixteen lexical
category ambiguities, 64 filler and 5 practice sentences were constructed; 24 of
the fillers were from an unrelated experiment. All the filler items used visual
target items other than the word us (as used by the lexical category ambiguity
stimuli), so that participants would not expect to see the visual target word us on
every trial. Two other visual targets {their and street) were repeated across 10
filler items each, so that repetitions of visual targets were not limited to us.
Following each stimuli sentence, participants were asked one of two
questions. Participants were either asked (1) a comprehension question
concerning information in the auditory sentence or (2) whether the visual target
word was a good continuation of the sentence in the form of "Does the word make
sense in the sentence". The latter question was always asked in the case of the
experimental items and the two types of questions were counterbalanced across
the filler and experimental items.
Stimuli were recorded by the first author and digitized at 22KHz using
MacRecorder software. Each soundedit file was edited so that it terminated at the
offset of the ambiguous word (or some other word in the filler items). Special
care was taken to record the ambiguous phrases with a neutral intonation, as
intonation could promote the noun or verb interpretation of the ambiguous word.
To check the intonation, one judge who was naive to the purpose of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experiment examined the sound waveforms and measured the duration of the final
word of the experimental items (e.g ..fires/fire), because duration is a major
component of intonation at the boundaries between noun and verb phrases
(Cooper & Paccia-Cooper, 1980). For the ambiguous items, the ambiguous words
had identical lengths of 729 msec in both contexts (SD = 103 ms in the verb-
supporting context condition and 136 ms in the noun-supporting context
condition). In the unambiguous conditions, the final words contained one fewer
phoneme than their ambiguous counterparts (e.g.,/ire vs. fires) and were slightly
shorter, with 694 ms (SD = 163 ms) in the verb-supporting context condition and
671 ms (SD =137 ms) in the noun-supporting condition. There were no
differences in final word duration across these four conditions, F < 1.
Procedure. The cross modal task was conducted on a Macintosh IISE
computer. Auditory sentences were presented over an external speaker which
participants voluntarily adjusted to a comfortable sound level. Participants were
seated in front of the computer at a distance from which they could comfortably
read the computer screen. A table microphone and button box were situated
nearby in front of the computer screen. The microphone was linked to a voice-
activated relay connected to the computer via the button box.
To begin the cross-modal experiment, participants read instructions on the
computer screen explaining how to proceed through the task. Once participants
had completed the instructions, participants were told that they could advance to
five practice trials which would enable them to become familiar with the task.
Upon completion of the practice trials, participants then advanced to the
experimental trials which consisted of 80 experimental and filler trials in random
order. In each presentation trial, participants were presented a single auditory
sentence fragment, at the end of which a visual target word appeared on the
computer screen in front of the participants. Participants named the visual target
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
word aloud as quickly and accurately as possible. Participants' naming responses
to the visual target word triggered a voice activated relay, and the visual target
word was removed from the computer screen. The experimenter recorded
whether (1) participants correctly named the visual target word, (2) participants
incorrectly named the visual target word, or (3) a machine error such as a
background noise, a cough, etc. occurred during participants' naming responses
causing participants' naming latencies to be incorrectly timed. A question was
then presented on the computer screen concerning either information in the
auditory sentence or asking whether the visual word target was a good
continuation of the auditorily presented sentence (always the latter type of
question for the experimental items). Participants pressed a key marked Yes or
No to answer the question and did not receive any feedback. Participants then,
when ready, pressed a key on the button box which allowed them to advance to
the next presentation trial.
Upon completion of the cross modal task, participants were thanked for their
participation and were debriefed as to what the experimental task was evaluating.
Each experimental session lasted 45 min.
Analyses. All Anova analyses conducted in Experiment 1A on naming times
and compatibility judgments treated participants as the random factor. The
correlational analyses conducted in Experiment 1A on naming times and
compatibility judgments in contrast treated items rather than participants as the
random factor. The reason that the correlational analyses treated items as the
random factor was that correlational analyses in the following Experiment IB
required, as a means of maintaining statistical balance, that items rather than
participants be treated as the random factor (This issue is discussed in Experiment
IB). As a means then of maintaining the comparability among the correlational
comparisons, all correlational analyses were conducted with items rather than
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
participants as the random factor. Additionally, the treatment of items rather than
participants as the random factor, reflects the growing trend in the
psycholinguistics literature for the treatment of strength of semantic context as a
continuous variable across stimulus items (MacDonald, Pearlmutter &
Seidenberg, 1994; Trueswell, Tanenhaus & Gamsey, 1994).
RESULTS
Naming Times. Prior to statistical analysis, naming times which were coded
as errors were removed. These errors comprised 2.8% of the younger group's
naming times and 8.5% of the older group's naming times. The naming data were
then trimmed in a two-step procedure. First, all naming times which were greater
than 5000 ms. were removed and second, all times more than 3 SD above or
below a participant's mean naming time were removed. This trimming affected an
additional 1.6% of the data for the younger participants and 3.3% of the data for
the older participants. Finally, a z score transformation was conducted on
younger and older participants' naming times in order to control for differences in
variability between the two groups.
Anova Analyses. The naming times for younger and older participants
are shown in Figure 1. It should be noted that the naming times presented in
Figure 1 are those calculated prior to the z score transformation. An analysis of
the effect of Ambiguity at each level of context for younger and older groups
evidenced a significant three-way interaction between Ambiguity, Context, and
Age, F (l, 62) = 4.04, MSe = 1.29, p < .05. The nature of this interaction was that
the younger participants' naming times were more influenced by context than
were times of the older participants. Thus younger adults' naming times as
aforementioned, yielded a reliable Ambiguity x Context Interaction, F(l,31) =
9.49, MSe = 2.13,/? < .01. The nature of this interaction was that when context
was verb-supporting, naming times in the ambiguous and unambiguous conditions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 1
Younger Group Older Group
■ Unambiguous
B Ambiguous
Verb-Supporting Noun-Supporting Verb-Supporting Noun-Supporting
Context Context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were not significantly different from one another, F < 1, however when the
context was noun-supporting, naming times in the ambiguous condition were
significantly longer than naming times in the unambiguous condition, F (l, 31) =
31.67, MSe = 8.17,p < .001 and also longer than in the ambiguous verb-
supporting condition, F (l, 31) = 9.49, MSe = 2.68, p < .01.
In contrast however, the older participants' naming times did not yield a
significant Ambiguity x Context Interaction, F < 1. Instead, older participants'
naming times yielded a significant Main Effect of Ambiguity, F(l,31) = 12.53,
MSe = 3.73, p < .001. The nature of this main effect was that naming times in the
ambiguous condition were significantly longer than naming times in the
unambiguous condition, irregardless of whether the context was verb-supporting
or noun-supporting. There was no main effect of Context, F < 1.5.
Correlation Analysis. A correlation analysis conducted over all
ambiguous experimental items (across both verb-supporting and noun-supporting
context conditions) evidenced that older participants' naming times were not
significantly correlated with younger participants naming times, r(31) = .28, p >
.10. These results then suggest that older adults differed in their ability to use
contextual information relative to younger adults during on-line cross modal
naming.
Compatibility Judgments. The analyses of the auditory fragment-visual
target compatibility judgments were conducted for all experimental trials,
including those naming trials which had initially been trimmed.
Anova Analyses. The compatibility judgments for younger and older
participants may be seen in Figure 2. At first glance, the number of auditory
fragments and visual targets judged as compatible with one another may appear
somewhat low, with fewer than 70% compatible judgments present in either of the
unambiguous conditions for both younger and older groups. The low number of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Percent Compatible
Figure 2
2 1
Younger Group
100' ■ Unambiguous
B Ambiguous
80-
Verb-Supporting Noun-Supporting
Context
Older Group
100-1
80-
Verb-Supporting Noun-Supporting
Context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
visual targets judged compatible however may in large part be due to the
somewhat odd use of the visual target "us" in the continuation of the auditory
sentences.
In contrast to the naming task, an omnibus analysis conducted on the
compatibility judgment data did not reveal an Ambiguity x Context x Age
Interaction, nor was there a main effect of Age, F's < 1. There was a significant
Ambiguity x Context Interaction, however, F(l,62) = 5.91, MSe = .35, p < .05.
Pairwise comparisons investigating this ambiguity showed that when context
promoted the incorrect noun interpretation, the percentage of targets judged as
compatible was significantly lower than in both the unambiguous condition,
F(l,62) = 31.21, MSe = 2.55, p < .001 and the verb-supporting ambiguous
condition, F(l,63) = 12.65, MSe = .74, p < .01. This pattern reflects good use of
context, and is similar to the pattern of younger group's naming data, although the
verb-supporting context did not remove all effects of ambiguity here, in that the
percentage of targets judged as compatible in this condition was lower than those
judged as compatible in the unambiguous condition, F(l, 63) = 7.20, MSe = .59, p
< .01. In sum, the absence of an interaction with Age suggests that both groups
were using contextual information similarly to aid in ambiguity resolution in this
task, but the benefit that younger and older adults received from the verb-
supporting context did not remove all effects of ambiguity. It is possible that the
muted effect of context might be due to the awkwardness of the first person visual
target "us."
Correlation Analyses. In a correlation analysis conducted over all
ambiguous experimental items (across both verb-supporting and noun-supporting
context conditions), a strong positive correlation between younger and older
participants' compatibility judgments was evidenced, r(31) = .64, p < .0001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These results suggest that older and younger adults use contextual information
similarly to resolve ambiguities in the compatibility judgment task.
In addition, the younger participants' compatibility judgments evidenced a
trend toward a negative correlation with their naming time data, r(31) = -.33, p >
.06. The nature of this negative correlation was such that the greater the
percentage of visual targets judged as compatible in the younger participants'
compatibility judgments, the shorter younger participants' naming times in on-line
cross modal naming. In contrast, older participants’ compatibility judgments did
not evidence a significant correlation of any type with their naming time data,
r(31) = -.17, p > .10. Combined, these results suggest that while older adults were
able to use contextual information in the compatibility judgment task, such
contextual information did not have any effect on the older participants' naming
times.
Discussion
Taken in totality, these results suggest that the older group in contrast to the
younger group was slower to use contextual information to aid ambiguity
resolution during on-line language processing. In the cross modal naming task,
whereas the verb-supporting context condition removed all effects of ambiguity
for the younger participants, the older participants were affected by ambiguity
regardless of whether the ambiguity was situated in a verb-supporting or noun-
supporting context. In contrast however, in the compatibility judgment task,
context effects did begin to exert some influence on older participants' ambiguity
resolution. While this task still required rapid processing of the auditory signal in
order to perceive the context and ambiguity, the delayed presentation of the
compatibility judgment task may have allowed older participants the time
necessary to compute the essential contextual information.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
One must now explain the finding that older adults were not able to use
contextual information to aid ambiguity resolution during the naming task. It was
suggested previously that older adults' inability to use contextual information in
ambiguity resolution is the result of slowed language processing. There is an
alternative hypothesis however which is that older participants' inability to use
contextual information is the result of a loss of the requisite and real-world
knowledge which makes the verb-supporting contexts more effective than the
noun-supporting contexts. Thus according to this view, older participants have
lost knowledge indicating that a fire in a warehouse is more plausible than a
warehouse firing someone. This hypothesis does not seem very plausible, first
because the older group's use of context in the compatibility judgment task closely
mimicked that of the younger group, and second because there exists an abundant
amount of previous research which has shown that older adults do retain such
information (Light, Valencia-Lavar & Zavis, 1991; Burke & Harrold, 1988; Burke
& Yee, 1984; Cohen & Faulkner, 1983; Madden, 1986,1988,1989; Nebes, Boiler
& Holland, 1986). These studies did not examine the exact contexts used here,
however, and thus it is important to demonstrate that younger and older adults do
have comparable knowledge relevant to the specific contexts used in the present
experiment.
Experiment IB explores these alternative hypotheses in a study in which
elderly participants read the experimental items up to the ambiguous word and
wrote completions for the sentences. Sentence completions reflect which
interpretation of the ambiguity was adopted and thus whether elderly adults used
contextual information in interpreting the ambiguities. If elderly participants'
interpretations of the ambiguities in this untimed task match those of a group of
young adults who did show strong context effects in sentence completion
(MacDonald, 1993), this result would suggest that the young and the elderly have
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
comparable knowledge about the contextual information in the stimuli. In this
case, the differences in Experiment 1A results would implicate differences in
speed of processing in the two age groups rather than knowledge differences.
However if older adults in the sentence completion task again show less reliance
on contextual information than do younger adults, the Experiment 1A results
would point to differences in knowledge between the two age groups.
EXPERIMENT IB
Experiment IB assessed older comprehenders' ability to use context in a
questionnaire task in which the older comprehenders had an unlimited amount of
time to read sentence fragments ending at an ambiguous word and write
completions for the fragments. Because the alternative interpretations of the
ambiguous words are from different lexical categories (noun/verb), respondents'
completions of the sentence fragments clearly indicate how they interpreted each
ambiguous word. Thus for example, for one experimental questionnaire item
ending at the ambiguous word "houses,” one 83 year old respondent completed
the sentence fragment, The prospective students were informed that the fraternity
houses with were. kcatedAgorprivate. homes and thus odd sounds should, he.
limited. The nature of the older respondents' completion then clearly indicated
that the respondent had interpreted the ambiguous word "houses" as a noun.
The results for the older respondents in Experiment IB will be compared to
data previously collected for younger adults in MacDonald, 1993 in the following
ways:
Old vs. Young Sentence Completions. The extent to which the older group’s
sentence fragment completions compare with those sentence fragment
completions of the younger adults in MacDonald, 1993 will be determined.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sentence Completions and Naming Times. The extent to which the younger
and older group's naming times from Experiment 1A correlate with context
strength as assessed by the younger and older respondents' sentence fragment
completions respectively, will be determined. There is a clear prediction that the
younger group’s naming times should be reliably correlated with the sentence
completion data, because MacDonald, 1993 found robust correlations between
sentence completions and reading times, and the younger group’s naming time
data in Experiment 1A replicated the reading time data from MacDonald, 1993.
The prediction for the older group is somewhat more complicated than that for the
younger group and must be considered in concert with the nature of the
correlation between the older and younger group's sentence fragment completions.
To explain then, if the older group's sentence completion data does not correlate
with the younger group's sentence completion data (which evidenced strong
context effects in MacDonald 1993), this then would point to a difference between
the two groups in their knowledge of contextual information. In this case, the
older group’s inability to use context in Experiment 1A would appear to be due to
the loss of linguistic and real-world knowledge which makes contextual
information effective and one would therefore predict that there would be a strong
correlation between the older group's naming times in Experiment 1A (which
evidence a lack of use of contextual information), and their sentence completion
data (which would also evidence a lack of use of contextual information). If
however, the older group's sentence completion data does correlate with the
younger group’s sentence completion data, this then would suggest that the older
and younger groups have comparable knowledge about contextual information in
stimuli. In this case, the older group's inability to use contextual information in
Experiment 1A would appear to be due to their slower processing speed and thus
one would therefore predict that there would be an insignificant correlation
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
between the older group’ s naming rimes in Experiment 1A (which evidence a lack
of use of contextual information) and their sentence completion data (which
would evidence use of contextual information).
Sentence Completions and Compatibility Judgments. And finally, the
extent to which the younger and older group’s compatibility judgments (for which
context effects began to emerge for the older group) from Experiment 1A
correlate with context strength as assessed by the younger and older adults'
sentence fragment completions respectively, will be determined.
Method
Respondents. For the older respondents, eighty elderly adults from the USC
S.A.G.E. alumni pool utilized in Experiment 1A, but who had not participated in
Experiment 1 A, were mailed questionnaires. Forty Four completed
questionnaires were returned, five of which were not included because the
respondents reported that their native language was a language other than English.
The mean age of the older respondents was 74.5, SD = 7.0 and 69% of the older
respondents were women.
The younger respondent group was composed of 96 native English-speaking
students in an introductory psychology course who as part of a course assignment
completed a questionnaire similar to that of the older respondents, as reported in
MacDonald (1993). Age and sex data for the younger respondents were not
recorded.
Materials. The eight different versions of the questionnaire employed by
MacDonald (1993) were used in this experiment. The questionnaires in the
current experiment however were slightly shortened and presented in larger type
than the questionnaires in the previous study. Each of the eight versions of the
questionnaire consisted of a series of sentence fragments for which respondents
were instructed to write a completion. Two examples of sentence fragments and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
plausible completions for each of the sentence fragments were illustrated in the
instructions. Respondents were encouraged to complete the sentence fragments
with the first ending that occurred to them and it was stressed that there were no
right or wrong answers.
The critical items on each of the eight different versions of the questionnaire
were the ambiguous experimental items. The ambiguous experimental items
consisted of sentence fragments which ended at the ambiguous words, e.g. "The
union told the reporters that the warehouse fires___________________________
followed by respondents' subjective completions of the fragments. A large
number of unambiguous fillers were included in each version of the questionnaire
to limit the chance that the respondents would become aware of the ambiguities;
each version contained 4 verb-supporting context and 4 noun-supporting context
ambiguous experimental fragments, intermixed with 19 filler fragments of the
same word length as the experimental items. The experimental items were always
separated by at least two filler items.
Analyses. All the statistical analyses conducted in Experiment IB treated
items as the random factor, rather than respondents. The comparison of younger
and older respondents' sentence completions was conducted in this manner
because younger and older respondents gave completions for only a subset of the
ambiguous experimental items, due to a design that placed only a few ambiguous
experimental items and a large number of fillers on each of the eight versions of
the questionnaire. Statistical analyses which treated respondents as the random
factor would therefore have been unbalanced. Additionally, the correlational
analyses which investigated a) the relationship between sentence completions and
naming times and b) the relationship between sentence completions and
compatibility judgments were also conducted with items as the random factor as a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
means of maintaining the comparability of correlation comparisons between
naming time, compatibility judgment, and sentence completion data.
Results
Old vs. Young Sentence Completions. The older respondents' sentence
fragment completions were scored for whether each completion reflected either a
verb or noun interpretation of the ambiguous word. The older respondents'
completions evidenced a strong effect of context, such that when the context
supported the verb interpretation of the ambiguous word, older respondents
completed sentence fragments with the verb interpretation of the ambiguity 60.4
% of the time. In contrast however, when the context did not support the verb
interpretation of the ambiguous word but rather supported the noun interpretation
of the ambiguous word, older respondents completed sentence fragments with the
verb interpretation of the ambiguity only 8.4 % of the time. These completions
then yielded a reliable effect of context in a matched t-test with items as the
random factor, t(15) = 7.73, p < .001. The completions of the older respondents
very closely paralleled those completions of the younger respondents in
MacDonald, 1993. In that study, younger respondents completed sentence
fragments with the verb interpretation of the ambiguity 55.3 % of the time when
the context supported the verb interpretation of the ambiguous word. When
however the context did not support the verb interpretation of the ambiguous
word, younger respondents completed sentence fragments with the verb
interpretation only 11.9 % of the time. Similar to the completions of the older
respondents, the completions of the younger respondents yielded a reliable effect
of context, t(15) = 8.87, p < .001.1 In a correlation analysis conducted over all
1 These and other values reported here for the younger group's sentence completions are slightly different
than those reported in MacDonald (1993). The reason they differ is because that study reported ratings for
the bank/agency accounts item useid there, whereas here we omitted that item and report ratings for the
lawyerlcourt judges item that was substituted for the accounts item in Experiment 1 A. The judges item
was assessed in the sentence completion task but not used in the reading task in MacDonald (1993).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ambiguous experimental items (across both verb-supporting and noun-supporting
context conditions), a significant correlation between older and younger
respondents' sentence fragment completions was evidenced, r(31) = -87, p <
.0001. These results then suggest that in the sentence completion task, younger
and older adults use contextual information similarly to resolve ambiguities.
Sentence Completions and Naming Times. As was predicted, the younger
participants' naming times for Experiment 1A evidenced a strong negative
correlation with the younger respondents' sentence completions reported in
MacDonald 1993, r(31) = -.59, p < .001. The nature of this negative correlation
was such that the greater the number of ambiguous experimental items given the
verb interpretation in younger respondents' sentence completions, the shorter
younger participants' naming times in on-line cross modal naming. In contrast,
the older participants' naming times for Experiment 1A did not evidence a
significant correlation of any type with the older respondents' sentence fragment
completions, r(31) = -.03, p > .10. These results suggest that while the older
adults were able to use contextual information in the sentence completion task,
such contextual information did not have any effect on older subjects' naming
times in Experiment 1A.
Sentence Completions and Compatibility Judgments. Younger participants'
compatibility judgments for Experiment 1A evidenced a trend towards a positive
correlation with younger respondents' sentence completion data, r(31) = .315, p <
.07. The nature of this positive correlation was such that the greater the number
of ambiguous experimental items given the verb interpretation in younger
respondents’ sentence completions, the greater the percentage of visual targets
judged as compatible in younger participants’ compatibility judgments. In
addition, older participants' compatibility judgments for Experiment 1A also
evidenced a positive correlation with older respondents' sentence completion data,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
r(31) = .4, p < .05. These results then suggest that like younger adults, older
adults were able to use context in the on-line compatibility judgment task as well
as the off-line sentence completion task.
Discussion
In Experiment IB, the significant correlation found between the older and
younger respondents' sentence fragment completions evidences that older and
younger adults use contextual information similarly to resolve ambiguities in the
off-line sentence completion task. These results then contradict the idea that older
adults in contrast to younger adults do not have the requisite linguistic and real-
world knowledge which makes contextual information effective. The older adults'
inability to use contextual information in the Experiment 1A on-line cross modal
naming task in contrast to the younger adults’ ability to use contextual information
then appears to be due to the older adults' slower speech processing speed relative
to that of younger adults.
This line of reasoning is further supported by the significant correlation found
between older adults' compatibility judgments and sentence completions
evidencing that older adults, when given enough time, can use contextual
information similar to younger adults in the compatibility judgment task in
addition to the sentence completion task. These results then illustrate that older
comprehenders can use context in on-line language comprehension but that their
use of context is delayed compared to that of younger comprehenders as a result
of the older adults' slower language processing operations relative to that of
younger adults.
Experiment 1 General Discussion
This research tested older and younger adults' use of context across the
timecourse of lexical ambiguity resolution. The important result was that whereas
the younger group's responses showed good use of context at all three timepoints,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the older group's use of context was a function of the timepoint at which
ambiguity resolution was assessed. Thus older adults showed no use of context
when ambiguity resolution was assessed at an early point in the cross modal
naming task, moderate use of context when ambiguity resolution was assessed at a
later point in the compatibility judgment task, and full use of context, exactly
comparable to that of younger adults, when ambiguity resolution was assessed in
an untimed sentence completion task in which respondents were allowed an
unlimited amount of time to resolve the lexical ambiguities. In addition, the older
group's responses on the compatibility judgment task and the sentence completion
task correlated with the responses of the younger group, demonstrating good
agreement between the two group's interpretation of the contextual biases. The
one exception to the pattern of agreement across the two age groups is the older
adults' naming data, which did not correlate with either the younger group's
naming data or with the older group's compatibility judgment or sentence
completion data.
These results then suggest that older adults have delayed use of contextual
information in comparison to younger adults in the resolution of lexical
ambiguities. An alternative hypothesis, that older adults have lost knowledge
relevant to the contextual information used in the present experiment can be
rejected in view of the good sensitivity to context which the older adults showed
in the compatibility judgment and sentence completion tasks.
What these results then leave one to question is what does guide older
comprehenders' ambiguity resolution when contextual information is delayed?
While Experiment 1 does not explicitly address this question, it is quite likely that
frequency information guides older adults' ambiguity resolution while contextual
information is delayed. As pointed out in the introduction, there are
computational reasons to expect that early in the process of ambiguity resolution,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
frequency effects will tend to dominate contextual effects more in older adults
than younger adults. As elucidated earlier, slower speech/language processing
operations in older adults would have less of an impact at an early time point in
the ambiguity resolution process on simple frequency information relative to
combinatorial contextual information, which requires integration of multiple
words, each of which would suffer from slowed language processing. There is in
fact research which has been conducted within memory paradigms which lends
support to this idea. The results from that research evidenced that older adults
persisted in encoding words in their high frequency interpretation even when the
context promoted a lower frequency interpretation (Hess & Higgins, 1983; Micco
& Masson, 1991).
Further indirect support for this hypothesis comes from the older group's
ambiguity effect in naming times in Experiment 1 A. In this experiment, the
ambiguous words were all more frequent in their noun interpretation than verb
interpretation, but the visual target always forced the lower frequency verb
interpretation. If the older group was insensitive to frequency information, it is
difficult to explain why their naming times in both ambiguous conditions were
longer than in the unambiguous conditions. One might hypothesize that any
ambiguity slows older comprehenders' responses, but this hypothesis is
incompatible with the abundant research showing the substantial role of frequency
in lexical ambiguity resolution (Simpson, 1981; 1984; Seidenberg et al., 1982;
Simpson & Burgess, 1985; Tabossi, 1988; Simpson & Krueger, 1991) and
research showing that in neutral contexts, ambiguous words with a strong
frequency bias toward one interpretation are no more difficult than unambiguous
words (Duffy et al., 1988).
The purpose of Experiment 2 is twofold. First Experiment 2 will attempt to
replicate the results of Experiment 1 with respect to younger and older adults' use
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of contextual information across the dmecourse of lexical ambiguity resolution
and second, Experiment 2 will assess indirectly older and younger adults' use of
frequency information across the dmecourse of ambiguity resolution.
EXPERIMENT 2A
The purpose of this experiment was to replicate older and younger adults' use
of contextual information and to indirectly assess younger and older adults' use of
frequency information in the resolution of lexical ambiguities during on-line
language processing. To accomplish this task, an experiment was constructed
which was quite similar in many respects to Experiment 1 A. Thus like
Experiment 1A, this experiment utilized the same cross modal paradigm as a
means of assessing participants' interpretations of lexical ambiguities. Similar
also to Experiment 1 A, this experiment used as stimuli, the lexical ambiguities
present in Experiment 1A in which the two alternative interpretations of the
ambiguities came from different lexical categories and had a high frequency noun
interpretation and low frequency verb interpretation (Francis and Kucera, 1982).
The point at which the two experiments differed was in the nature of the
experimental conditions. Specifically, unlike Experiment 1 A, in the current
experiment, the visual target word could always disambiguated the ambiguity in
favor of the high frequency noun interpretation. Thus unlike Experiment 1A, in
the ambiguous noun-supporting context condition, a single context word within
the sentence promoted the ultimately correct noun interpretation of the ambiguity
to be revealed at the disambiguation. Similarly, in the ambiguous verb-supporting
context condition, a single context word within the sentence promoted the
incorrect verb interpretation of the lexical ambiguity. Also in contrast to
Experiment 1A, the current experiment as a result of restrictions in the English
language, only implemented a single unambiguous control condition in which the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lexical ambiguity was disambiguated to the noun interpretation prior to the
presentation of the disambiguating visual target
If then during early on-line processing, comprehenders do in fact rely on
contextual information to aid ambiguity resolution, then one would expect that
when context supported the correct high frequency noun interpretation
(ambiguous noun-supporting context condition), the visual target would
disambiguate the lexical ambiguity in a way which was consistent with
participants' interpretation of the lexical ambiguity and the visual target could
would thus form a plausible continuation of the auditory sentence. In this case
then, participants’ naming times in the ambiguous noun-supporting context
condition would be expected to match naming times for the unambiguous control
condition in which the visual target would similarly be interpreted as a plausible
continuation. When however context supports the incorrect verb interpretation
(ambiguous verb-supporting context condition), the visual target would
disambiguate the lexical ambiguity in a way which was inconsistent with
participants' interpretation of the lexical ambiguity and the visual target could
would thus form an implausible continuation of the auditory sentence. In this
case then, participants' naming times in the verb-supporting context condition
should be significantly longer than their naming times for the unambiguous
control condition and for the noun-supporting context condition, as the visual
targets for the unambiguous control and noun-supporting context conditions
would in contrast form plausible continuations of the auditory sentences.
If however, comprehenders rely on frequency information to aid ambiguity
resolution during early on-line language processing, then one would expect that
participants' naming times in both the ambiguous noun-supporting and verb-
supporting context conditions (where frequency always leads to the correct
interpretation of the ambiguity regardless of context) should not significantly
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
differ their naming times for the unambiguous control condition in which the
correct high frequency interpretation of the ambiguity is similarly disambiguated
prior to the presentation of the disambiguating visual target The pattern of
effects across age would then reveal whether older and younger comprehenders
differ in their reliance on contextual in contrast to frequency information during
early on-line processing.
In addition to the aforementioned Anova analyses, a correlation analysis will
assess the extent to which the older group's naming times correlate with the
naming times of the younger group.
If at a later point during the ambiguity resolution process e.g. at the point the
Compatibility Judgments are made, comprehenders rely on contexmal information
to aid ambiguity resolution, then one would expect that when context supported
the correct high frequency noun interpretation (ambiguous noun-supporting
context condition), the percentage of visual targets judged by participants as
compatible with the auditory sentence would not significandy differ from the
percentage of visual targets judged by participants as compatible with the auditory
sentence in the unambiguous control condition in which the visual targets always
form plausible continuations. When the context however supported the incorrect
low frequency verb interpretation (ambiguous verb-supporting context condition),
the percentage of visual targets which participants judged as compatible with the
auditory sentence would be significantly smaller then the percentage of visual
targets which participants judged as compatible in the unambiguous control and
noun-supporting context conditions.
If however comprehenders rely on frequency information to aid ambiguity
resolution at a later point in on-line processing e.g. at the point that Compatibility
Judgments are made, then one would expect that in both the ambiguous noun-
supporting and verb-supporting context conditions (where frequency always
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
makes the visual target a plausible continuation), the percentage of visual targets
judged by participants as compatible with the auditory sentence would not
significantly differ from the percentage of visual targets judged by participants as
compatible with the auditory sentence in the unambiguous control condition in
which the visual target always forms a plausible continuation. Like the cross
modal naming task, the pattern of effects across age would then reveal whether
older and younger comprehenders differ in their reliance on contextual in contrast
to frequency information in lexical ambiguity resolution at a later point in on-line
processing.
In addition to the aforementioned Anova analyses, a correlation analysis will
assess the extent to which the older group's compatibility judgments correlate with
the compatibility judgments of the younger group. Also, the extent to which the
older group’ s compatibility judgments correlate with their naming times and the
extent to which the younger group's compatibility judgments correlate with their
naming times will be assessed.
Method
Participants. The younger participant group was composed of thirty two USC
undergraduates (25 female, 7 male) who were paid for their participation. These
participants were randomly selected from a larger group of forty seven
participants who had been tested in the experiment. All participants were Native
English speakers and were naive to the purpose of the experiment. The mean age
of the younger participants was 22.5 years, SD = 5.5 and the younger participants
reported a mean 14.9 years of formal education, SD = 2.6.
The older participant group was composed of thirty two USC alumni (17
male, 15 female) who were affiliated with S.A.G.E. The older participants were
compensated for parking and travel related expenses. An additional three
participants were tested but were rejected due to probable dementia (Parkinson's
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Disease/Alzheimer's Disease). All participants were native English speakers,
naive to the purpose of the psycholinguistic experiment, and in reportedly good
health. The mean age of the older participants was 72.6 years, SD = 5.0 and the
older participants reported a mean 16.0 years of formal education, SD = 2.0.
Materials. The experimental stimuli used in the current cross modal naming
task consisted of the 16 ambiguous experimental items implemented in
Experiment 1A supplemented by eight additional ambiguous experimental items
constructed by the first author, and 24 unambiguous experimental items which
differed from those used in Experiment 1A, also constructed by the first author.
These eight additional items were added as a means of increasing power and
balancing the items across the three cells in the present experiment. These
experimental items may be found in Appendix A. An example may be seen in
Table 2. Similar to Experiment 1 A, the stimuli in this task were constructed
around 24 lexical category ambiguous words with both a plural noun and third
person singular verb meaning (fires, benefits, guards, etc.). Again, all of the
ambiguous words had a higher frequency noun interpretation than verb
interpretation; the mean was 6% verb interpretations in Francis and Kucera
(1982).
As shown in Table 2, in the ambiguous conditions, prior context for the
ambiguous word was manipulated at two levels, noun-supporting and verb-
supporting. These contexts differed by only one word, the word immediately
preceding the ambiguous word, e.g. warehouse/corporation in the example in
Table 2. The noun-supporting context promoted the noun interpretation of the
ambiguous word; this context is consistent with the interpretation which is forced
by the visual target could. For example, the noun-supporting context word
warehouse in the phrase warehouse fires promotes the noun interpretation of fires
because it seems more plausible for a fire to occur in a warehouse than for a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
Table 2
Example Stimuli, Experiment 2
Auditory Fragments, Ambiguous Conditions
Noun-Supporting Context The union told the reporters that the
warehouse fires COULD
Verb-Supporting Context The union told the reporters that the
corporation fires COULD
Auditory Fragment, Unambiguous Condition
Adjective Context The union told the reporters that
the dangerous fires COULD
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
warehouse to fire people. In contrast, the verb-supporting context promoted the
verb interpretation of the ambiguous word; this context is inconsistent with the
interpretation which is forced by the visual target could. For example, the verb-
supporting context word corporation in the phrase corporation fires promotes the
verb interpretation of fires because it seems more plausible for a corporation to
fire people than for a corporation to have a fire.
As may be seen in Table 2, the unambiguous condition in contrast to the
ambiguous conditions did not manipulate context. This contrasts with Experiment
1A in which context was manipulated at both the verb-supporting and noun
supporting levels in the unambiguous condition. The reason that the current
experiment did not attempt to manipulate context in the unambiguous condition
was due primarily to constraints imposed by the English language. Thus there
was no syntactic context which could force phrases such as
warehouse/corporation fires to be interpreted unambiguously as noun phrases.
The unambiguous condition then did not differ from the ambiguous noun
supporting and verb-supporting context conditions in any other way except for the
nature of the word immediately preceding the lexical ambiguity. Thus while the
words employed in the ambiguous noun-supporting and verb-supporting context
conditions to immediately precede the ambiguous word and serve as context were
nouns, the words employed in the unambiguous condition to immediately precede
the ambiguous word were adjectives. The significance of this difference is that
when an ambiguous word such as fires is preceded by an adjective as opposed to a
noun, although the ambiguity still has both the noun and verb interpretations,
there is no other grammatical interpretation of the word fires than that of the noun
interpretation.
In addition to the stimulus sentences constructed around the twenty four
lexical category ambiguities, 38 filler items and 5 practice sentences were
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
constructed. All filler items used either the visual target "could" or "by." The
visual target items "could" and "by," from both the experimental and filler items,
were counterbalanced across the experiment so participants would not build up a
preferential expectation for either one of the visual target items.
Following each stimuli sentence, participants were asked whether the visual
target word was a good continuation of the sentence with the question "Could the
word follow what you just heard"?.
The stimuli were recorded by the first author and digitized at 22KHz using
MacRecorder software. Each soundedit file was edited so that it terminated at the
offset of the ambiguous word (or some other word in the filler items). Again,
special care was taken to record the ambiguous phrases with neutral intonation, as
intonation could promote the noun or verb interpretation. In order to ensure that
the intonation was neutral, an analysis similar to that implemented in Experiment
1A was conducted in which the duration of the final word of the experimental
items was measured. For the ambiguous conditions, items with verb-supporting
contexts had average lengths of 637 ms, SD = 105 and items with noun-
supporting contexts had average lengths of 641 ms, SD = 100. For the
unambiguous condition, items had average lengths of 629 ms, SD = 106. There
were no differences in final word duration across these three conditions, F < 1.
Procedure. There were no differences between the experimental procedures
used in the current experiment and those procedures used in Experiment 1A.
Analyses. Similar to Experiment 1A, all Anova analyses conducted in this
experiment on naming times and compatibility judgments treated participants as
the random factor. Additionally, like Experiment 1 A, the correlational analyses
conducted in this experiment on naming times and compatibility judgments
treated items rather than participants as the random factor.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
Results.
Naming Times. Prior to statistical analysis, incorrect naming responses and
naming times that were coded as microphone errors were removed, comprising
0.8% of the younger participants' naming times and 6.7% of the older participants'
naming times. The naming data were then trimmed as in Experiment 1 A,
affecting an additional 1.0% of the data for the younger participants and 1.7% of
the data for the older participants. Finally, a z-score transformation was
conducted on younger and older participants' naming times in order to control for
differences in variability between the two groups.
Anova Analyses. The naming times (prior to z-score transformation) for
younger and older participants in the three experimental conditions are shown in
Figure 3. The Age x Context Interaction was marginal, F(2,124) = 2.22, MSe =
.51, p < .12. Additional planned one-way Anovas investigated the effect of
context at each age group. For the younger group, there was a reliable difference
across the three levels of context, F(2,62) = 5.05, MSe = .83, p < .01. The pattern
of naming times reflected good use of context for this group: When the context
supported the noun interpretation of the ambiguity, naming times were no
different than times in the unambiguous condition, F < 1, but naming times in the
verb-supporting context were significantly longer than in the unambiguous
condition, F (l, 31) = 6.02, MSe = 1.43, p < .05 and also longer than in the noun-
supporting condition, F (l, 31) = 6.71, MSe = 1.02, p < .05. By contrast, there
were no differences in the older group's naming times across the three
experimental conditions, F < 1.
Correlation Analysis. A correlation analysis conducted over all
ambiguous experimental items (across both noun-supporting and verb-supporting
contexts) evidenced that older participants' naming times were not significantly
correlated with younger participants' naming times, r(31) = .045, p > .75. These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Naming times (msec)
Figure 3
43
Younger Group
1800
1600
1400
1200
1000
800
600
400
V«rb-Support!iNoun-Supportfn(Unam bifluou»
Context
Older Group
mmmm
M S S H ft
V erb-SupportiiN oun-Supportinf U nam biguous
Context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
results then lend greater support to the idea that older and younger adults differ in
their reliance on contextual in contrast to frequency information in the resolution
of lexical ambiguities during on-line cross modal naming.
Compatibility Judgments. The analyses of the auditory fragment-visual
target compatibility judgments were conducted for all experimental trials,
including those for which the corresponding naming trials had been initially
trimmed. These data may be seen in Figure 4. In contrast to Experiment 1A, the
overall percentage of targets judged compatible was quite high, suggesting that
participants found the could visual target generally more acceptable than the us
target used in the previous experiment.
Anova Analysis. With the exception of the overall higher rates of judged
acceptability, the pattern of responses for both age groups was very similar to that
found in Experiment 1A. An omnibus analysis revealed neither an Age x Context
interaction nor a main effect of Age, F's < 1. There was, however, a significant
main effect of context, F(2, 124) = 13.88, MSe = .08, p < .001. Pairwise
comparisons across age showed that when the context supported the noun
interpretation, the percentage of visual targets judged as compatible was no
different from that in the unambiguous condition, F < 1, but the context that
promoted the verb interpretation yielded a significantly lower percentage of
compatible judgments than in both the unambiguous condition (F(l,63) = 19.15,
MSe = .13, p < .001) and the noun-supporting context condition (F(l,63) = 12.92,
MSe = .10,p<.001).
Correlation Analyses. In a correlation analysis conducted over all
ambiguous experimental items (across both noun-supporting and verb-supporting
contexts), a strong positive correlation between younger and older participants1
compatibility judgments was evidenced, r(31) = .647, p < .0001. These results
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Percent Compatible
Figure 4
Younger Group
Context
Older Group
Vcr*»Sueportlftf N ««ft-Sw pporitaf U n e w b lfw e e
Context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
suggest that older and younger adults similarly rely on contextual information in
the resolution of lexical ambiguities in the compatibility judgment task.
In a correlation analysis conducted over all ambiguous experimental items
(across both noun-supporting and verb-supporting contexts), a negative
correlation which was not quite reliable was found between younger participants'
compatibility judgments and younger participants’ naming times, r(31) = -.226, p
> .12. The nature of this negative correlation was such that the greater the
percentage of visual targets judged as compatible in the younger participants'
compatibility judgments, the shorter younger participants’ naming times in on-line
cross modal naming. It should be noted that it is quite possible that the
percentages of visual targets judged as compatible across both the ambiguous
noun-supporting and verb-supporting context conditions were so close to ceiling
that the power to find a relationship between the younger participants'
compatibility judgments and naming times was obscured. While the correlation
between younger participants' naming times and compatibility judgments was not
statistically significant, it is clear, as evidenced by the previous Anova analyses,
that younger participants are relying on contextual information to resolve lexical
ambiguities in both the on-line cross modal naming task and compatibility
judgment task. In contrast, older participants' compatibility judgments did not
evidence a significant correlation with older participants' naming time data, r(31)
= -.188, p > .2. Taken in totality, these results suggest that older adults, while
relying on contextual information in the compatibility judgment task to resolve
lexical ambiguities, do not use such a strategy in the cross modal naming task,
relying instead on frequency information to resolve the lexical ambiguities.
Discussion
Experiment 1, having evidenced that older adults are delayed but not
qualitatively different from younger adults in their use of contextual information,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
motivated the inquiry of the present experiment which was what guides older
comprehenders' ambiguity resolution when contextual information is delayed.
Taken in totality, the results of the present experiment confirm the on-line
findings of Experiment 1 and suggest that older adults, in the absence of the
ability to use contextual information, rely on frequency information.
Thus like Experiment 1A findings on cross modal naming, in the current
experiment, younger participants' evidenced use of contextual information in the
resolution of lexical ambiguities, with younger participants' naming times to
visual targets in the ambiguous noun-supporting and verb supporting context
conditions and the unambiguous control condition varying roughly as a function
of whether or not the visual targets in the three experimental conditions were
consistent with the interpretation of the lexical ambiguity invoked by the
preceding context. The older participants, in contrast to the younger participants,
did not evidence use of contextual information in cross modal naming. What the
older participants did appear to rely on in the absence of the ability to use
contextual information, was frequency information. Thus no statistically
significant differences were evidenced between older participants' naming times
to visual targets in the ambiguous noun-supporting and verb supporting context
conditions and unambiguous control condition in which the visual targets for all
three experimental conditions were consistent with the most frequent
interpretation of the lexical ambiguity.
In contrast, in the compatibility judgment task, which was presented several
seconds after the cross modal naming task, our results, like those found in
Experiment 1A (although more statistically significant in the current experiment),
evidenced that older adults in addition to younger adults, relied on contextual
information to resolve lexical ambiguities. Thus the percentage of visual targets
judged as compatible with the auditory sentence by both older and younger adults
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in the ambiguous noun-supporting and verb-supporting context conditions and
unambiguous control condition varied roughly as a function of whether or not the
visual targets in the three experimental conditions were consistent with the
interpretation of the lexical ambiguity invoked by the preceding context rather
than as a function of the most frequent interpretation of the lexical ambiguity.
Thus older adults, when given additional time to integrate contextual information
during on-line processing, rely similarly to younger adults on contextual
information in contrast to frequency information.
EXPERIMENT 2B
The purpose of this experiment was to assess younger and older adults' use of
contextual information when given an unlimited amount of time to resolve lexical
ambiguities. Although the results from the off-line sentence completion data in
Experiment IB speak to this issue, evidencing that both younger and older adults
rely on contextual in contrast to frequency information when given an unlimited
amount of time to resolve lexical ambiguities, differences between Experiment 1
and Experiment 2 in the nature of the contextual conditions and experimental
items used, necessitated that this issue be reassessed.
To reassess this issue then, a sentence completion task was implemented
which was essentially identical to the sentence completion task implemented in
Experiment IB (with the exception of a few additional experimental items and the
fact that the present experiment was contextually supportive/unsupportive of the
noun in contrast to the verb interpretation of the lexical ambiguity). Thus similar
to Experiment IB, younger and older comprehenders in this sentence completion
task had an unlimited amount of time to read the experimental sentence fragment
items up to the ambiguous word and then write completions for the fragments.
Like Experiment IB, the sentence completions reflected which interpretation of
the lexical ambiguity was adopted and thus were indicative of younger and older
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
respondents' use of contextual/frequency information in interpreting the
ambiguities. The results for the younger and older respondents in Experiment 2B,
like those for Experiment IB, will be compared in the following ways:
Young vs. Old Sentence Completions. The extent to which the younger and
older respondents' sentence fragment completions reflect a reliance on contextual
as opposed to frequency information will be assessed. There is a clear prediction
that both the younger and older respondents' sentence completions will evidence a
significant effect of context in contrast to an effect of frequency as both the
younger and older respondents’ sentence fragment completions in Experiment IB
evidenced a significant effect of context.
Sentence Completions and Naming Times. The extent to which the younger
and older group's naming times from Experiment 2B correlate with context
strength as assessed by the younger and older respondents' sentence fragment
completions respectively, will be determined. There is a clear prediction that
while the younger participants' naming times from Experiment 2A should be
reliably correlated with the younger respondents' sentence fragment completions,
the older participants’ naming times from Experiment 2A should in contrast not be
correlated with the older respondents' sentence fragment completions. This
prediction follows both because Experiment IB evidenced a robust correlation
between younger participants' naming times and younger respondents’ sentence
fragment completions, and no significant correlation between older participants'
naming times and older respondents’ sentence fragment completions and the
younger and older participants’ naming time data in Experiment 2A replicated the
naming time patterns for younger and older participants in Experiment 1 A.
Sentence Completions and Compatibility Judgments. And finally, the
extent to which the younger and older group's compatibility judgments from
Experiment 2A correlate with context strength as assessed by the younger and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
older respondents' sentence fragment completions respectively, will be
determined. There is a clear prediction that younger and older participants'
compatibility judgments from Experiment 2A should be significantly correlated
with younger and older respondents' sentence fragment completions, respectively.
This prediction follows both because Experiment IB evidenced robust
correlations between younger participants' compatibility judgments and younger
respondents' sentence completions and older participants' compatibility judgments
and older respondents' sentence completions and the younger and older
participants' compatibility judgment data in Experiment 2A replicated (although
somewhat more strongly) the younger and older participants' compatibility
judgment data in Experiment 1 A.
Method
Respondents. The younger group of respondents was composed of 163
native English-speaking students who completed the questionnaire as a part of a
course assignment in an introductory psychology course. An additional 46
respondents completed the questionnaire but their responses were not used
because they either reported that they were bilingual or that their native language
was a language other than English. The mean age of the younger respondents was
18.8, SD = 1.2, and 64% of the younger respondents were women. Although the
education level of the younger respondents was not collected, the characteristic
number of years of formal education which the younger respondents had received
was 13.5 years.
For the older group of respondents, forty elderly adults from the USC
S.A.G.E. alumni pool utilized in the previous experiments, but who had not
participated in any of the previous experiments, were mailed questionnaires.
Twenty Six completed questionnaires were returned, all of which were usable.
The mean age of the older respondents was 73.1, SD = 4.5 and the older
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
respondents reported a mean 17.6 years of formal education, SD = 2.1. 42% of
the older respondents were women.
Materials. The eight different versions of the questionnaire employed in
Experiment IB were slightly expanded to contain the eight additional
experimental items implemented in Experiment 2A. In contrast to Experiment
IB, the eight different versions of the questionnaire in the present experiment
contained 6 ambiguous noun-supporting context and 6 ambiguous verb-
supporting context experimental fragments, intermixed with 23 filler fragments of
the same word length as the experimental items. Like Experiment IB, the
experimental items were always separated by at least two filler items.
Analyses. Similar to Experiment IB, all statistical analyses in Experiment 2B
treated items, rather than respondents, as the random factor.
Results
Young vs. Old Sentence Completions. The younger and older respondents'
sentence fragment completions were scored for whether each completion reflected
either the noun or verb interpretation of the ambiguity. It should be noted that in
contrast to Experiment IB, the effect of context in the present Experiment was
evaluated with respect to the percentage of respondents' sentence fragments
completed with the noun interpretation of the lexical ambiguity in the ambiguous
noun-supporting context condition vs. the ambiguous verb-supporting context
condition rather than with respect to the percentage of respondents' sentence
fragments completed with the verb interpretation of the lexical ambiguity in the
ambiguous verb-supporting context condition vs. the ambiguous noun-supporting
context condition. An analysis of the younger respondents’ sentence completions
evidenced a reliable effect of context in a matched t-test with items as a random
factor, t(23) = 8.84, p < .0001. The nature of this effect was such that when
context supported the noun interpretation of the ambiguous word, younger
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
respondents completed sentence fragments with the noun interpretation of the
ambiguity 89.0 % of the time. When however the context did not support the
noun interpretation but rather supported the verb interpretation of the ambiguous
word, younger respondents completed sentence fragments with the noun
interpretation of the ambiguity only 44.4 % of the time. Similar to the younger
respondents' sentence completions, an analysis of the older respondents' sentence
completions evidenced a reliable effect of context in a matched t-test with items
as the random factor, t(23) = 8.6, p < .0001. The nature of this effect of context
for the older respondents' sentence completions was identical to that of the
younger respondents' sentence completions. Thus the older respondents
completed sentence fragments with the noun interpretation of the ambiguity 86.5
% of the time when the context supported the noun interpretation of the
ambiguity. When however the context did not support the noun interpretation of
the ambiguity, older respondents completed sentence fragments with the noun
interpretation only 33.3 % of the time. In addition, a correlation analysis
conducted over all ambiguous experimental items (across both noun-supporting
and verb-supporting contexts), evidenced a significant correlation between
younger and older respondents' sentence fragment completions, r(31) = .83, p <
.0001. These results then suggest that both younger and older adults rely on
contextual information in contrast to frequency information when given an
unlimited amount of time to resolve lexical ambiguities.
Sentence Completions and Naming Times. As was predicted previously,
the younger respondents' sentence fragment completions evidenced a strong
negative correlation with the younger participants' naming time data for
Experiment 2A, r(31) = -.37, p < .01. The nature of this negative correlation was
such that the more supportive the context of the noun interpretation of the
ambiguous word in younger respondents’ sentence completion data, the shorter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
younger participants naming times in on-line cross modal naming. In contrast, the
older respondents' sentence fragment completions did not evidence a significant
correlation of any type with the older participants’ naming time data for
Experiment 2A, r (31) = -.189, p > .10. These results then support the contention
that while older adults rely on contextual in contrast to frequency information to
resolve ambiguities in the sentence completion task, they do not rely on such
contextual information to resolve ambiguities during on-line cross modal naming.
Sentence Completions and Compatibility Judgments. As was also
predicted, the younger respondents' sentence completion data evidenced a strong
positive correlation with the younger participants' compatibility judgment data for
Experiment 2A, r(31) = .47, p < .001. In addition, the older respondents'
sentence completion data also evidenced a strong positive correlation with the
older participants' compatibility judgment data for Experiment 2A, r(31) = .34, p
< .05. The nature of these positive correlations was such that the more supportive
the context of the noun interpretation of the ambiguous word in younger and older
respondents’ sentence completion data, the greater the percentage of visual targets
judged as compatible in younger and older participants' compatibility judgments.
These results then support the idea that both younger and older adults rely on
contextual in contrast to frequency information to resolve lexical ambiguities
during the on-line compatibility judgment task in addition to the off-line sentence
completion task.
Discussion
The results of Experiment 2B thus replicate the findings of Experiment IB
and evidence that both younger and older adults rely on contextual information in
contrast to frequency information when given an unlimited amount of time to
resolve lexical ambiguities. These results are also in line with the findings from
the compatibility judgment task in Experiment 2A which evidenced that both
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
younger and older adults rely on contextual information when given additional
time during on-line processing. These results then suggest that older adults, when
given enough time to integrate and utilize contextual information, will similar to
younger adults, rely on contextual in contrast to frequency information to resolve
lexical ambiguities.
Experiment 2 General Discussion
This experiment assessed younger and older adults' use of contextual and
frequency information across the timecourse of lexical ambiguity resolution.
Importantly, the results of the present experiment replicated the pattern of
contextual effects found for both younger and older adults at all three timepoints
within the previous experiment and demonstrated that younger and older adults
differed in their use of contextual and frequency information as a function of the
timepoint at which ambiguity resolution was assessed. Thus at an early timepoint
in the ambiguity resolution process, as assessed by the cross modal naming task,
while the younger adults relied on contextual information to resolve lexical
ambiguities, the older adults in contrast to relying on contextual information
appeared to rely on frequency information to resolve lexical ambiguities. At
timepoints later in the ambiguity resolution process however, as assessed by the
on-line compatibility judgment task and the off-line sentence completion task,
both younger and older adults relied on contextual information to resolve lexical
ambiguities. These results then lend support to the idea that for older adults, as
result of slower speech/language processing operations, access to contextual
information is slowed relative to access to frequency information.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
GENERAL DISCUSSION
This research assessed older and younger adults' use of context and frequency
information in three tasks. With respect to contextual information, the important
results were that while the younger group’s responses showed a reliance on
contextual information in all three tasks, the older group's reliance on contextual
information varied with the amount of time available to use the contextual
information to resolve the ambiguities. Thus, older adults showed no use of
contextual information in the cross modal naming task which demanded
immediate use of context, moderate use of contextual information in the judgment
task which demanded less immediate use of context and full use of contextual
information, exactly comparable to younger adults, in an untimed sentence
completion task. In addition, indirect evidence suggests that older adults rely on
frequency information in the absence of the ability to use contextual information
to resolve the lexical ambiguities. These results then support the account of age
related changes in the nature of ambiguity resolution predicted by Kawamoto's
computational model. Below these findings are related to work in aging and
language comprehension in three respects. First the question of what does guide
ambiguity resolution in older comprehenders while contextual information is
delayed is considered. Second, the findings of the present study are contrasted
with those of other studies that suggest that older comprehenders have similar or
even greater use of context compared to younger comprehenders and finally, the
role which task differences might have played in the results is considered.
Context and Frequency. It was argued that what guides older adults’
ambiguity resolution in the absence of the ability to use contextual information to
resolve lexical ambiguities is frequency information. Support for this argument
rests on the cross modal naming times of older adults in Experiments 1 & 2. Thus
in Experiments 1 & 2, the lexical ambiguities were all more frequent in their noun
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interpretation than verb interpretation. In Experiment 1, the visual target always
forced the lower-frequency verb interpretation and in Experiment 2, the visual
target always forced the higher-frequency noun interpretation. Consequently, in
Experiment 1, older adults' naming times were longer in the ambiguous condition
than the unambiguous condition. This pattern of naming times did not obtain in
Experiment 2 however, where older subjects were disambiguated to the high
frequency interpretation. In this Experiment, older adults' naming times in the
ambiguous conditions became equivalent to those in the unambiguous condition.
This is exactly the pattern of results which one would predict were the older
adults relying on frequency information.
It must be pointed out however that the present experiments only provide
indirect evidence with regard to older adults' reliance on frequency information.
Thus the present experiments in their focus on the role of context in ambiguity
resolution, simply controlled frequency information and did not manipulate it.
Thus the extent to which older and younger adults rely on frequency information
during the process of ambiguity resolution has not be explicitly addressed.
Clearly this is an issue which merits further investigation. At the present time,
this issue is being addressed in an experiment which specifically looks at the
relative activation of the two alternative interpretations of lexical ambiguities
across the process of ambiguity resolution in younger and older adults.
O ther Context Research. Although a number of studies have investigated
context use during language comprehension with older adults, the present
experiments are unique in combining on-line measures with a context that
immediately precedes the ambiguity. Several studies that show similar use of
contextual information in older and younger adults have used off-line tasks
(Burke & Harrold, 1988; Light et al., 1991). These results are entirely consistent
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
with the present data which showed comparable use of context by older and
younger adults in the off-line tasks of Experiments 1 & 2.
Other research has shown good context use by older adults in on-line tasks. In
an investigation of on-line processing using a cross modal naming task similar to
the one used here, Burke and Yee (1984) found that older and younger adults
made similar inferences about instruments used in actions, as evidenced by
speeded response times to visual targets (e.g., knife) following sentences
containing a related verb (e.g., cut), compared to a control condition. This result
suggests that older and younger adults both used the context of cut rapidly, and to
the same degree. This finding is consistent with the present results, in that the key
contextual information in Burke and Yee's study was a verb that was in the middle
of the stimulus sentence, so that more time elapsed between the context word and
the sentence-final visual target in Burke and Yee’ s study than in the present study.
Similarly, Hopkins, Kellas and Paul (1995) in an investigation of the
activation of the alternative meanings of lexical ambiguities during on-line
processing found that both younger and older adults exhibited selective activation
of the contextually biased senses of homographs. Thus both younger and older
adults evidenced a facilitation in responses to contextually appropriate targets
relative to unrelated controls. Responses to contextually inappropriate targets
however were no different than unrelated targets. A closer inspection of the
Hopkins et al. data reveals that these results are also entirely consistent with the
present results. Thus in the Hopkins et al. study, the key contextual information
was syntactic in contrast to semantic in nature. So it was the syntax immediately
preceding the ambiguity which limited the ambiguity to a single grammatical
interpretation. This is quite similar to the unambiguous conditions in the present
experiments in which a modal "could" in Experiment 1 and various adjectives in
Experiment 2 constrained the lexical ambiguities to a single grammatical
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interpretation. Recall that although the syntactic contexts immediately preceded
the lexical ambiguity, the older comprehenders were able to use the contextual
information. This is apparent in Experiment 1 in which the older comprehenders
were able to use the syntactic "could" context in order to adopt the verb
interpretation of the ambiguous word as evidenced by their shorter reaction times
in the unambiguous condition as compared to the ambiguous condition. Perhaps
then it is better to distinguish context as either being semantic or syntactic in
nature. Thus whereas the more subtle semantic context such as
(warehouse/corporation fires) in the present experiments was only useful to the
younger group in the naming task, the stronger syntactic context was useful to
both the younger and older groups in the naming task within the present
experiments and within the Hopkins et al. investigation of word meaning
activation during on-line processing. Taken in totality, these results suggest that
the extent to which comprehenders use context in ambiguity resolution can be
viewed as a function of both speed of processing and context strength. Clearly
this is an issue that merits additional investigation.
Task Differences. One potential criticism of the present series of
experiments is that the three tasks differ not only in the time which one has to
integrate contextual information to resolve the lexical ambiguities but also with
respect to the nature of the dependent variables, task demands, the role of strategy,
etc. Thus one might argue that older adults differed from younger adults in the
cross modal naming task not as a result of the time required to integrate
contextual information but as a result of differences between the two groups in
their ability to negotiate some fundamental aspect of the cross modal naming task
itself. This seems unlikely given that older adults did in fact evidence an effect of
syntactic context in the cross modal naming task of Experiment 1, with older
adults evidencing longer naming times in the ambiguous conditions which
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
contained the more subtle semantic contexts than the unambiguous conditions
which contained the stronger syntactic contexts. Older adults do seem to be able
to negotiate the cross modal naming task at some fundamental level. One could
however more definitively alleviate such criticism by demonstrating that older
adults like the younger adults were able to use the more subtle semantic contexts
if they were given additional time to allow for processing the semantic context.
There are a number of ways in which one could vary the distance between the
context and the ambiguity and/or between the ambiguity and disambiguation to
pursue this option. The avenue which is currently being pursued is increasing the
ISI between the offset of the ambiguity and the appearance of the visual target (In
the current experiment, the ISI was 0 between the offset of the ambiguous word
and the appearance of the visual target "us") to allow older adults more time to
integrate the subtle semantic context.
In sum, a more definitive account of the relationship of frequency effects,
context effects and processing speed awaits additional investigation. The initial
foray into aging and ambiguity resolution however appears to be quite promising
to both of the respective fields. Thus it is likely that delayed use of context and
subsequent reliance on frequency information in older comprehenders contributes
at least partially to differences between younger and older adults' language
comprehension performance. In addition, the investigation of context and
frequency effects in aging can contribute to the active debate concerning the
extent to which contextual information can modulate frequency and other
"bottom-up" effects (Duffy et al. 1987; Just & Carpenter, 1992; MacDonald et al.
1993; Marslen-Wilson, 1987; McClelland, 1987). The present results show that
in addition to strength of context and frequency, researchers must also look to
comprehenders’ age and/or processing speed in order to understand the
context/frequency interactions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 0
REFERENCES
Burke, D. M., & Harrold, R. M. (1988). Automatic and effortful semantic
processes in old age: Experimental and naturalistic approaches. In L. L. Light
& D. M. Burke (Eds.), Language. Memory, and Aging (pp. 100-116).
Cambridge: Cambridge University Press.
Burke, D. M., & Yee, P. L. (1984). Semantic priming during sentence processing
by young and older Adults. Developmental Psychology. 2 0 .903-910.
Carpenter, P. A., & Daneman, M. (1981). Lexical retrieval and error recovery in
reading: A model based on eye fixations. Journal of Verbal Learning and
Verbal Behavior. 20. 137-160.
Cohen, G., & Faulkner, D. (1983). Word recognition: Age differences in
contextual facilitation effects. British Journal of Psychology. 7 4 .238-251.
Cooper, W. E., & Paccia-Cooper, J. (1980). Syntax and Speech. Cambridge,
MA: Harvard University Press.
Cottrell, G. W. (1989). A connectionist approach to word sense disambiguation.
San Mateo, CA: Morgan Kaufmann.
Daneman, M., & Carpenter, P. (1980). Individual differences in working memory
and reading. Journal of Verbal Learning and Verbal Behavior. 19. 450-466.
Duffy, S. A., Morris, R. K., Rayner, K. (1988). Lexical ambiguity and fixation
times in reading. Journal of Memory and Language. 27.429-446.
Francis, W. N., & Kucera, H. (1982). Frequency analysis of English usage:
Lexicon and grammer. Boston: Houghton-Mifflin.
Galbraith, G. G., & Taschman, C. S. (1969). Homophone units: A normative and
methodological investigation of the strength of component elements. Journal
of Verbal Learning and Verbal Behavior. 8 .737-744.
Hess, T. M., & Higgins, J. N. (1983). Context utilization in young and old adults.
Journal of Gerontology. 38. 65-71.
Hopkins, K. A., Kellas, G., & Paul, S. T. (1995). Scope of word meaning
activation during sentence processing by younger and older adults.
Experimental Aging Research. 21. 123-137.
Just, M. A., & Carpenter, P. A. (1992). A capacity theory of comprehension:
Individual differences in working memory. Psychological Review. 9 8 .122-
147.
Kawamoto, A. (1993). Nonlinear dynamics in the resolution of lexical ambiguity:
A parallel distributed processing account. Journal of Memory and Language.
22*474-516.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 1
Kemper, S. (1992). Language & Aging. In F. I. M. Craik & T. A. Salthouse
(Eds.), The Handbook of Aging and Cognition (pp. 213-270). Hillsdale, NJ:
Erlbaum.
Light, L. L., Valencia-Lavar, D., & Zavis, D. (1991). Instantiation of general
terms in young and older adults. Psychology and Aging. & 337-351.
MacDonald, M. C. (1993). The interaction of lexical and syntactic ambiguity.
Journal of Memory and Language. 32.692-715.
MacDonald, M. C., Pearlmutter, NJ. & Seidenberg, M.S. (1994). The lexical
nature of syntactic ambiguity resolution. Psychological Review. 101.676-
703.
Madden, D. J. (1986). Adult age differences in visual word recognition: Semantic
encoding and episodic retention. Experimental Aging Research. 8 .91-98.
Madden, D. J. (1988). Adult age differences in the effects of sentence context and
stimulus degradation during visual word recognition. Psychology and Aging.
167-172.
Madden, D. J. (1989). Visual word identification and age-related slowing.
Cognitive Development. 4 . 1-29.
Marslen-Wilson, W. D. (1987). Functional parallelism in spoken word-
recognition. Cognition. 2 5.71-102.
McClelland, J. L. (1987). The case for interationism in language processing. In
M. Coltheart (Eds.), Attention & Performance XII: Reading. London:
Erlbaum.
Micco, A., & Masson, M. J. (1992). Age-related differences in the specificity of
verbal encoding. Memory and Cognition. 20. 244-253.
Nebes, R. D., Boiler, F., & Holland, A. (1986). Use of semantic context by
patients with Alzheimer's disease. Psychology and Aging. 1. 261-269.
Salthouse, T. A. (1982). Adult cognition: An experimental psychology of human
aging. New York: Springer-Verlag.
Salthouse, T. A. (1985). A theory of cognitive aging. Amsterdam: North Holland
Salthouse, T. A. (1988). Resource reduction interpretations of cognitive aging.
Developmental Review. 8 .238-272.
Seidenberg, M. S., Tanenhaus, M. K., Leiman, J. L., & Bienkowski, M. (1982).
Automatic access of the meanings of ambiguous words in context: Some
limitations of knowledge-based processing. Cognitive Psychology. 14.489-
537.
Seidenberg, M. S., & McClelland, J. L. (1989). A distributed, developmental
model of word recognition and naming. Psychological Review. 9 6.447-452.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Simpson, G. B. (1981). Meaning dominance and semantic dontext in the
processing of lexical ambiguity. Journal of Verbal Learning and Verbal
Behavior. 20. 316-340.
Simpson, G. B. (1984). Lexical ambiguity and its role in models of word
recognition. Psychological Bulletin. 96.316-340.
Simpson, G. B., & Burgess, C. (1985). Activation and selection processes in the
recognition of ambiguous words. Journal of Experimental Psychology:
Human Perception and Performance. 11.28-39.
Simpson, G. B., & Krueger, M. A. (1991). Selective access of homograph
meaning in sentence context Journal of Memory and Language. 3 0 .627-643.
Tabossi, P. (1988). Accessing lexical ambiguity in different types of sentential
context Journal of Memory and Language. 2 7 .324-340.
Trueswell, J. C., Tanenhaus, M. K., & Gamsey, S. M. (1994). Semantic
influences on parsing: Use of thematic role information in syntactic ambiguity
resolution. Journal of Memory and Language. 33.285-318.
Tyler, L. K., & Marslen-Wilson, W. D. (1977). The on-line effects of semantic
context on syntactic processing. Journal of Verbal Learning and Verbal
Behavior. 16. 683- 692.
Tyler, L. K., & Marslen-Wilson, W. D. (1982a). Processing utterances in
discourse contexts. Journal of Semantics. 1 .297-314.
Tyler, L. K., & Marslen-Wilson, W. D. (1982b). The resolution of discourse
anaphors: Some on-line studies. Text. 2 .263-291.
Wingfield, A., Poon, L. W., Lombardi, L., & Lowe, D. (1985). Speed of
processing in normal aging: Effects of speech rate, linguistic structure, and
processing time. Journal of Gerontology. 40. 579-585.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix A
a = ambiguous verb-supporting context condition
b = ambiguous noun-supporting context condition
c = unambiguous condition
la. The townspeople were pleased that the new sheriff guards could
lb. The townspeople were pleased that the new prison guards could
lc. The townspeople were pleased that the watchful guards could
2a. The doctor refused to believe that the shrine cures could
2b. The doctor refused to believe that the miracle cures could
2c. The doctor refused to believe that the medical cures could
3a. The prospective students were informed that the university houses could
3b. The prospective students were informed that the fraternity houses could
3c. The prospective students were informed that the spacious houses could
4a. It was evident to everyone that the lawyer judges could
4b. It was evident to everyone that the court judges could
4c. It was evident to every one that the critical judges could
5a. The efficiency expert recently reported that the company supplies could
5b. The efficiency expert recently reported that the office supplies could
5c. The efficiency expert recently reported that the expensive supplies could
6a. It says in the manual that the technician programs could
6b. It says in the manual that the computer programs could
6c. It says in the manual that the difficult programs could
7a. Bob explained to the police that the report documents could
7b. Bob explained to the police that the official documents could
7c. Bob explained to the police that the controversial documents could
8a. Ellen was pleased to see how the curtain frames could
8b. Ellen was pleased to see how the window frames could
8c. Ellen was pleased to see how the pretty frames could
9a. Since the company’s policy was revised, the community benefits could
9b. Since the company's policy was revised, the employee benefits could
9c. Since the company's policy was revised, the maternity benefits could
10a. At the beginning of every summer, the coach loans could
10b. At the beginning of every summer, the college loans could
10c. At the beginning of every summer, the small loans could
11a. Fred wanted to be sure that the client contacts could
lib . Fred wanted to be sure that the business contacts could
1 lc. Fred wanted to be sure that the electrical contacts could
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12a. The seamstress may have problems because the ruler measures could
12b. The seamstress may have problems because the tape measures could
12c. The seamstress may have problems because the best measures could
13a. In developing the new promotion policy, the department bases could
13b. In developing the new promotion policy, the army bases could
13c. In developing the new promotion policy, the industrial bases could
14a. The young inspector soon learned that the grocer stores could
14b. The young inspector soon learned that the grocery stores could
14c. The young inspector soon learned that the new stores could
15a. The union told the reporters that the corporation fires could
15b. The union told the reporters that the warehouse fires could
15c. The union told the reporters that the dangerous fires could
16a. The residents were all pleased that the fund returns could
16b. The residents were all pleased that the tax returns could
16c. The residents were all pleased that the profitable returns could
17a. The most recent election illustrated that the citizen votes could
17b. The most recent election illustrated that the campaign votes could
17c. The most recent election illustrated that the earliest votes could
18a. The company chairman was informed that the supervisor talks could
18b. The company chairman was informed that the labor talks could
18c. The company chairman was informed that the lengthy talks could
19a. It's reported in the article that the stove bums could
19b. It's reported in the article that the cigarette bums could
19c. It's reported in the article that the severe bums could
20a. The new employees were advised that the boss orders could
20b. The new employees were advised that the customer orders could
20c. The new employees were advised that the important orders could
21a. The television spokesperson recently announced that the weather controls
could
21b. The television spokesperson recently announced that the climate controls
could
21c. The television spokesperson recently announced that the inadequate controls
could
22a. Lauren was surprised to find that the cat traps could
22b. Lauren was surprised to find that the mouse traps could
22c. Lauren was surprised to find that the big traps could
23a. It was apparent to all that the plaque honors could
23b. It was apparent to all that the student honors could
23c. It was apparent to all that the literary honors could
24a. The newly appointed president stated that the accountant values could
24b. The newly appointed president stated that the family values could
24c. The newly appointed president stated that the fundamental values could
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Explanations, the availability heuristic, and biases of subjective likelihood in economics

PDF

Depression in offspring following severe prenatal stress

PDF

Affection and conflict in family relationships

PDF

History of depression, antidepressant treatment, and other psychiatric illness as risk factors for Alzheimer's disease in a twin sample

PDF

Effects of bilateral stimulation and stimulus redundancy on performance in processing nonword letter trigrams

PDF

Effects of threat and self-focus on consensual bias in majorities

PDF

Intragroup evaluations, attitude source, and in-group member derogation

PDF

Alcohol expectancies and consumption: Age and sex differences

PDF

Effects of personal resource sufficiency on perceived difficulty and desirability of earthquake preparedness

PDF

Asymmetries in the bidirectional associative strengths between events in cue competition for causes and effects

PDF

Effects of a corneal anesthetic on the extinction of the classically conditioned response in the rabbit

PDF

Depression and suicidality in Latino adolescents: A study of acculturation and gender role beliefs

PDF

Depressed children and the social and behavioral attributes of their best friend

PDF

Adolescents' social attitudes: Genes and culture?

PDF

Applying the theory of reasoned action to condom use: The effect of immediate consequences on intention-behavior consistency

PDF

Head injury and dementia: A co-twin control study of Swedish twins

PDF

Insights into the nature of phonological and surface dyslexia: Evidence from a novel word learning task

PDF

Ethnic identity, acculturation, self-esteem and perceived discrimination: A comparison study of Asian American adolescents

PDF

Expectancies for alternative behaviors predict drinking problems: A test of a cognitive reformulation of the matching law

PDF

Hispanic culture, acculturation, and distress among caregivers of dementia patients

Asset Metadata

Creator Stevens, Karen Lee (author)

Core Title Aging and the use of context and frequency information in ambiguity resolution

School Graduate School

Degree Master of Arts

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,Psychology, clinical,psychology, experimental

Language English

Contributor Digitized by ProQuest (provenance)

Advisor [illegible] (committee chair), [illegible] (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-7259

Unique identifier UC11336921

Identifier 1381616.pdf (filename),usctheses-c16-7259 (legacy record id)

Legacy Identifier 1381616.pdf

Dmrecord 7259

Document Type Thesis

Rights Stevens, Karen Lee

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA