Phonological Priming Effects on Word
Retrieval and Tip-of-the-Tongue Experiences
in Young and Older Adults
Lori E. James Deborah
M. Burke
Claremont Graduate
University Pomona
College
James, L.E. &, Burke, D.M. (2000). Phonological priming effects on word retrieval and tip-of-the-tongue experiences in young and older adults. Journal of Experimental Psychology: Learning, Memory and Cognition. 26, 1378-1391.
In a repetition priming paradigm, young and older participants read aloud prime words which sometimes shared phonological components with a target word that answered a general knowledge question. In Experiment 1, prior processing of phonologically related words decreased tip-of-the-tongue states (TOTs) and increased correct responses to subsequent questions. In Experiment 2, the priming task occurred only when the participant could not answer the question. Processing phonologically related words increased correct recall, but only when the participant was in a TOT state. Phonological priming effects were age invariant, although older adults produced relatively more TOTs. Results support the Transmission Deficit model that the weak connections among phonological representations that cause TOTs are strengthened by production of phonologically related words. There was no evidence that phonologically related words block TOT targets.
Phonological Priming
Effects on Word Retrieval and Tip-of-the-Tongue Experiences
in Young and Older Adults
The tip of the tongue state (TOT) is a relatively common type of speech error in which a word retrieval failure is coupled with a strong feeling of knowing, and often with a considerable sense of frustration at the inaccessibility of the desired word. Typically, a person can access semantic and syntactic properties of the TOT word and partial phonological properties such as initial sound or number of syllables, although the complete phonology remains inaccessible (e.g., A. Brown, 1991; R. Brown & McNeill, 1966; Koriat & Lieblich, 1974; Miozzo & Caramazza, 1997; Vigliocco, Antonini & Garrett, 1997). Resolution of a TOT is as compelling as the TOT onset when it occurs spontaneously, with the target word popping into mind at a time when retrieval attempts have been abandoned (Burke, MacKay, Worthley & Wade, 1991; A. Brown, 1991; Reason & Lucas, 1984). In this paper we investigate phonological encoding processes that are fundamental to speech production and are implicated in the cause of TOTs and their spontaneous resolution. We also investigate the role of phonological encoding processes in aging effects: TOTs are a hallmark of old age, increasing in frequency with normal aging in both experimental and naturalistic studies (A. Brown & Nix, 1996; Burke et al., 1991; Cohen & Faulkner, 1986; Heine, Ober & Shenaut, 1999; Maylor, 1990b; Rastle & Burke, 1996) and ranking as older adults' most annoying cognitive failure (Lovelace & Twohig, 1990).
TOTs are a valuable source of information about the nature of the processes and architecture of the speech production system. They require production models, for example, to account for successful selection of semantic and lexical information but failed selection of phonological information during word retrieval (e.g., Bock & Levelt, 1994; Caramazza, 1997). Models of speech production generally agree that semantic, lexical and phonological information are represented in independent systems and that production begins with conceptualization processes involving activation of semantic information and then selection of a syntactically specified lexical representation, followed by retrieval of corresponding phonological components of the word (e.g., Bock & Levelt, 1994; Burke et al., 1991; Caramazza, 1997; Dell, 1986; Harley & Bown, 1998; Levelt, Schriefers, Vorberg, Meyer, Pechmann & Havinga, 1991; MacKay, 1987; Martin, Weisberg & Saffran, 1989). The vulnerability of phonological retrieval to failure has been explained in the Transmission Deficit model of TOTs in terms of the strength of connections that transmit priming1 to phonological representations (Burke et al., 1991; MacKay & Burke, 1990). This model is a detailed version of an insufficient activation model of TOTs wherein retrieval fails because of incomplete activation of representations (e.g., Bowles, Obler & Poon, 1989; A. Brown & Nix, 1996; R. Brown & McNeill, 1966; Cohen & Faulkner, 1986; Harley & Bown, 1998; Meyer & Bock, 1992; Yaniv & Meyer, 1987; see A. Brown, 1991 for a review).
Under the Transmission
Deficit (TD) model, TOTs occur when the strength of connections among
phonological nodes is too weak to transmit sufficient priming for activation of
the complete phonology of the TOT target word. The TD model is based on a
general interactive activation model of language production and perception,
Node Structure Theory (NST), in which priming is a form of subthreshold
excitation that prepares a node for activation or retrieval. The strength of
connections to
nodes determines the rate and amount of priming transmitted to them and is an
important determinant of what information in memory becomes available (MacKay,
1987). Recent and frequent activation of nodes, in production of a word for
example, strengthens connections increasing priming transmission, whereas aging
weakens connections, reducing priming transmission (Burke et al., 1991; Burke & MacKay, 1997; MacKay &
Burke, 1990). Because connection strength declines over time in the absence of
activation, the TD account explains why TOTs, although involving familiar words
(see A. Brown, 1991), are more likely for low than high frequency words and for
words that have not been used recently (Burke et al. 1991; Harley & Bown,
1998; Rastle & Burke, 1996). The architecture of the phonological system
renders it more vulnerable to transmission deficits than the semantic system
because phonological nodes are generally linked by single connections rather
than the multiple converging connections that characterize the semantic system
(see Burke et al., 1991; Laver & Burke, 1993; MacKay & Abrams,
1996). This explains why access to
phonology declines with aging (as indicated by increased TOTs) but access to
the meaning of words, as measured by vocabulary tests (e.g., Schaie, 1996) or
experimental semantic retrieval tasks (e.g., Light,1992), does not decline, at
least until very old age.
We test the TD model of
TOTs using a repetition priming paradigm in which participants read prime words
sharing phonological components with a target word that answers a general
knowledge question. We constrain repetition effects to the phonological level
by eliminating overlap between prime and target at the perceptual2,
lexical and semantic levels. For example, representations for the prime word pellet
and the target word velcro shown in Figure 1 overlap only at the
phonological level, in this case at the nodes for /el/ . Darker lines
indicate phonological connections strengthened by production of pellet.
Inasmuch as a TOT for velcro is caused by transmission deficits among
phonological connections for /el/
, strengthening these connections by saying pellet should reduce the
likelihood of a TOT for velcro, or increase the likelihood of resolution
if a TOT for velcro has already occurred. Indeed, these predicted results would support the proposal
that ñspontaneousî resolution of TOTs in everyday life is triggered by phonological
priming. It has been suggested
that a TOT target word pops into mind "spontaneously" when
phonological components of the word occur inadvertently during conversation and
provide a boost in priming to the phonological nodes suffering from transmission
deficits (Burke et al., 1991; Seifert, Meyer, Davidson, Patalano & Yaniv,
1994; Yaniv & Meyer, 1987).
Non-perceptual repetition priming effects on word production are well established, but only when repetition at a lexical level is involved. For example, young adults' picture naming was facilitated by prior production of the picture name in response to a printed word or definition (A. Brown, Neblett, Jones & Mitchell, 1991; Durso & Johnson, 1979; Park & Gabrieli, 1995; Wheeldon & Monsell, 1992.). Similarly, correct responses to general knowledge questions increased with prior processing of the responses (e.g., Blaxton, 1989; Kelley & Lindsay, 1993; Yaniv, Meyer & Davidson, 1995) and this priming effect was invariant with aging (Rastle & Burke, 1996; Small, Hultsch & Masson, 1995). The type of prior processing of target words influenced priming effects for responses to general knowledge questions and provided evidence for the phonological locus of the retrieval failure in TOTs. Prior semantic processing increased correct production of targets more than prior phonological processing, but the two types of processing were equally effective in reducing TOTs, presumably because both required pronunciation which strengthened phonological connections (Rastle & Burke, 1996).
Only a few studies,
however, have investigated phonological priming effects on production while
eliminating prime-target overlap at the lexical level. These studies provide little evidence
of priming effects. Prior
production of a homophone in response to a definition (e.g., One grows at
the end of each finger) speeded naming of a subsequent picture (e.g., nail
[e.g., hammer]), but the effect was marginally significant and held only for
same spelling homophones, not for differently spelled homophones (e.g.,
son/sun) (Wheeldon & Monsell, 1992).
Valentine, Moore and Bredart (1995) found that prior production of
homophones of the surnames of the people (e.g., bush [e.g., shrub]) produced no
significant priming effect on naming latency of pictures of famous people
(e.g., Bush).
The duration of the
prime to target-generation interval in these studies is standard for repetition
priming paradigms, ranging from several minutes to 20 minutes or more. The
mechanism responsible for a priming effect, were it to be observed, is a
relatively long-term change in connection strength, the mechanism of interest
in the present research (MacKay, 1987; Valentine et al., 1995; Wheeldon &
Monsell, 1992). In contrast,
researchers using a very brief prime to target-generation interval of a few
hundred milliseconds, reported that phonologically related prime words
facilitated picture naming and they attributed this to a different mechanism: a
very short-lived, residual priming in nodes common to the prime and target or
that spreads from prime to target (Collins & Ellis, 1992; McEvoy, 1988;
Meyer & Schriefers, 1991; Schriefers, Meyer & Levelt, 1990). Repetition
priming paradigms, including the present research, are generally designed to
eliminate the contribution of this mechanism via the delay between prime word
processing and the target production task (see Wheeldon & Monsell, 1994 for
discussion of this issue).
Another
approach to investigating phonological priming effects on production has
yielded quite different results. In this approach, participants are presented
definitions of low frequency words, followed by a word cue related to the
target word. In studies where
cuing occurs before the participant answers the questions, Meyer and Bock
(1992) reported that phonologically related cue words sharing the initial
sound, number of syllables and stress pattern with the target word increased
production of targets more than unrelated or semantically related cue words.
Using similarly constructed phonological cues, Perfect and Hanley (1992) also
obtained a beneficial effect on target production, although smaller and only
compared to unrelated cues. These studies, however, reported little effect of
phonologically related cues on TOTs.
Meyer and Bock (1992) found a marginally significant reduction in TOTs
with phonologically related compared to unrelated cues, and Perfect and Hanley
(1992) reported no reduction compared to an unrelated or no cue condition. In
studies where cuing occurred after a participant had failed to produce the
correct reponse to a definition, there was no difference in the number of TOTs
remaining after phonologically related and unrelated cues (Meyer & Bock,
1992). Correct responses, however,
increased with presentation of a phonologically similar word (Kozlowski, 1977;
Meyer & Bock, 1992), the initials for proper names (e.g., Brennen, Baguley,
Bright & Bruce, 1990), or the initial letters of the target word (e.g.,
Freedman & Landauer, 1966; Gruneberg & Monks, 1974; Hanley & Cowell,
1988; Heine et al., 1999).
As Meyer and Bock
pointed out, facilitatory effects in these cuing paradigms may reflect
strategic retrieval processes, not processes involved in normal word
retrieval. A single prime or cue
was presented with each question and participants knew that the prime was
sometimes related to the target and that it might aid recall of the
target. Participants may have used
this information to improve performance, for example, by generating word
candidates sharing phonology with the prime or by rejecting responses
dissimilar in sound to the prime.
The present research aims to determine whether phonological priming
effects on responses to general knowledge questions are obtained in a procedure
that minimizes strategic processes and is more similar to repetition priming
paradigms such as Wheeldon and Monsell (1992).
Different
predictions for the effects of phonologically related words on production are
made by an alternative, inhibition explanation of TOTs in which TOTs are caused
by ñblockingî of the intended word by a more accessible alternate word that
comes to mind first (e.g., A. Brown, 1979; Jones, 1989; Jones & Langford,
1987; Reason & Lucas, 1984; Roediger, 1974; Woodworth, 1929; see A. Brown,
1991 for a review). These
alternate words usually share initial phonemes and number of syllables with the
intended word (Burke et al., 1991; Kohn, Wingfield, Menn, Goodglass, Berko
Gleason & Hyde, 1987). Under
the Inhibition hypothesis, processing a word phonologically related to the
intended word blocks its retrieval, increasing the probability of a TOT. Jones (1989; Jones & Langford,
1987; see also Maylor, 1990a) provided support for this hypothesis, reporting
that TOT inducing questions presented with a word phonologically related to the
answer produced more TOTs than questions presented with a word not
phonologically related. Two
studies, however, provide evidence that this influential result occurred
because specific questions were assigned to only one cue word condition and the
questions assigned to the phonologically related word condition produced more
TOTs than questions assigned to other conditions, independent of prime word
(Meyer & Bock, 1992; Perfect & Hanley, 1992). These two studies provide no evidence that phonologically
related words impair retrieval.
Despite the absence of supporting evidence and the availability of
counter evidence, the ñblockingîexplanation of TOTs has a tenacious hold and is
widely cited (e.g., Schacter, 1999).
There
is, however, evidence that once a TOT occurs, an alternate word that comes to
mind spontaneously will delay resolution.
Burke et al. (1991) reported that young and older adults' took
considerably longer to resolve naturally occurring TOTs that were accompanied by
persistent alternate words, most of which shared phonological features with the
target, than to resolve TOTs without alternates. Similarly, laboratory induced TOTs were much less likely to
be resolved during the experiment if they occurred with alternates than if they
did not. Burke et al. argued that
alternates are a consequence not a cause of TOTs and are
activated when the target word cannot be retrieved. Once the alternate word is
activated, its lexical node is more accessible than the target's until the
effects of the recent activation of the alternate subside (see Burke et al.,
1991). It is an interesting
paradox that spontaneously occuring alternate words delayed TOT resolution
whereas in another study phonologically related words presented by the
experimenter facilitated TOT resolution (e.g., Meyer & Bock, 1992). In Experiment 2, we examine further the
effect of related words on TOT resolution.
The present research was
designed to test the prediction of the TD model that production of words
phonologically related to a target word will reduce TOTs for the target or
increase resolution if a TOT for the target has already occurred. Within this framework, the mechanism
responsible for improved target retrieval is a relatively long-term change in
connection strength that occurs without the participant's awareness. To investigate this process, it is
essential to eliminate retrieval or guessing strategies based on prime words
because these might mimic positive priming effects. Previous studies of the effects of phonologically related
words on production of answers to questions have presented a single cue word
which is phonologically related on 33% or more of the trials. We designed our experiment with the
paramount goal of measuring priming effects without the participant's awareness
of the relation between prime words and targets, thereby eliminating retrieval
strategies. Several aspects of our
methodology were critical for this goal.
First, for a given
target word, processing of prime words was accomplished in a pronunciation
rating task that was introduced as a task completely separate from the question
answering. Second, the priming stimulus for a single question was not a single
word but 10 words which participants read aloud and rated on pronunciation
difficulty. On half of the trials,
the 10 words were all phonologically unrelated to the target; on the other half
of the trials, five of the words were unrelated and five were phonologically
related to the target. This
relatively large number of words for a single target disguised prime target
relatedness and made quite onerous any attempt at strategic use of the words,
if they could be remembered.
Moreover, there were no clues to which list contained related words, no
clues to which words, if any, were related, and no indication that any word was
ever related to the target word.
Third, the relatedness between target and prime words was difficult to
notice even if the target was produced because a related prime word and the
target often overlapped only in one or two phonemes, and most were not initial
phonemes. The related words could not be distinguished from the unrelated words
because they were not related to each other any more than the unrelated words
were related to each other: Phonological segments repeated among related words
on a list as little as possible and any overlap in phonemes among list words
occurred equally often on related and unrelated lists. For example, in the related list shown
in Figure 2 for the target abdicate, related words indigent and tradition
overlap with each other, but so do unrelated words tappet and velvet. Related words truncate and locate
overlap, but truncate and tradition also overlap in components
that are not shared with the target.
Thus repeated phonological components within a list were not a clue to
the phonology of the target nor did they identify the related words. The
unrelated list for abdicate was: vector, rejoicing, serious,
tappet, liniment, velvet, procedure, reread,
publish, older.
These unrelated words overlapped, eliminating overlap as a clue to
relatedness, e.g., tappet and velvet, vector and velvet,
rejoicing and reread, procedure and publish.
Finally, the presentation rate for prime words was brisk, allowing little time
for strategies.
Experiment 1
In this experiment, we
presented the pronunciation task before each question to examine the effect of
phonologically related prime words on correct responses and TOTs in both young
and older adults. While processing prior to the production task is standard in
repetition priming paradigms, this procedure has not been used in studies of
phonological effects on TOTs, with one exception (Jones, 1989). The TD model
predicts an increase in correct recall via a decrease in TOTs because the prime
words strengthen phonological connections that are vulnerable to transmission
deficits.
As for aging and priming
effects, no age differences were obtained in two studies showing repetition
priming effects on responses to general knowledge questions (Rastle &
Burke, 1996; Small et al., 1995). Repetition
priming effects, in general, show little age difference in individual studies,
although slightly smaller effects for older adults emerge in meta-analyses
(e.g., LaVoie & Light, 1994). Under the TD model, repetition at the lexical
or phonological level improves retrieval by strengthening existing connections,
a process that is age invariant (MacKay & Burke, 1990). Thus, although
older adults are more susceptible to TOTs because they are more likely to
suffer transmission deficits, the benefit from phonological priming should be
equivalent in young and older adults and any age difference in TOTs should
remain.
The results are also
relevant to the Inhibition hypothesis which predicts a decrease in correct
recall and an increase in TOTs following phonologically similar prime
words. Although this hypothesis
has received little support in previous studies, strategic use of
phonologically related cue words may have offset any negative effect. It has
been proposed that inhibitory processes are less efficient in old age (Hasher
& Zacks, 1988); thus any blocking effect by related prime words should be
smaller for older than young adults within this framework.
Method
Participants. Thirty six young (M=19.4 years, SD=1.3)
and 36 healthy older adults (M=72.0 years, SD=5.1) participated
in the experiment. Young adults
participated for credit in introductory psychology courses and older adults
were paid for their participation.
Mean score on the Nelson-Denny vocabulary test was lower for young (M=17.4,
SD=2.8) than older participants (M=21.6, SD=3.0), t(67)=6.04,
p<.001. Mean years of
education were 13.4 (SD=1.2) and 16.5 (SD=3.6) for young and
older adults, respectively. All
participants were native English speakers.
Materials. One hundred fourteen questions each answered by a single, low-frequency word, were selected from a pool of questions used in previous studies to induce TOT states (e.g., Burke et al., 1991; Yaniv & Meyer, 1987). When relevant data were available, questions were selected that elicited comparable numbers of TOTs in young and older adults so as to maximize the number of TOTs elicited in the experiment. For each of the 114 target words, two lists of ten words were compiled. A related list contained five words phonologically related to the target and five words phonologically unrelated to the target, and an unrelated list contained ten words unrelated to the target. All list words were semantically unrelated to the target. The phonologically related words were selected from the dictionary on the basis of sharing one or more phonemes with the target word. Syntactic class of prime words was not controlled, but on average 3.14 out of the 5 phonologically related prime words shared syntactic class with the target word. The Appendix presents the target words and phonologically related prime words used in Experiments 1 and 2, and indicates overlap in phonemes and grammatical class.
Certain constraints were placed on selection of phonologically related words to minimize participants' awareness of their phonological similarity to target words. For two-syllable targets, only one of the five related words shared a whole syllable with the target word; the other four related words shared only parts of syllables (e.g., the initial or final phoneme of the first syllable). For targets with three or more syllables, most related list words shared only one syllable with the target. The five unrelated words in the related list were matched with the phonologically related words on number of syllables and frequency (Francis & Kucera, 1982), but shared no phonology with the target. The unrelated list contained the five unrelated words from the related list and five new unrelated words matched on number of syllables and frequency. For each participant, half the questions were associated with a related list and half with an unrelated list with the assignment of questions to list type counterbalanced over participants.
A
multiple-choice recognition test included the question and four possible
answers: the target, a phonologically related word (not a prime word), a
semantically related word, and an unrelated word.
Procedure. A Macintosh computer presented
instructions and stimuli, and also recorded responses. The instructions informed participants
that: "An answer is on the 'tip of your tongue' (TOT) if you know you know
the answer, but you cannot produce the answer at this moment. You are only in a TOT state if you are
SURE you know the correct answer.
Being able to recognize the correct answer if it were presented to you
is NOT the same as being in a TOT state." The instructions also indicated that the list pronunciation
task was unrelated to the general knowledge questions.
The
events on a trial are shown in Figure 2. A related or unrelated list of words
was presented one word at a time on the monitor. Participants pronounced each
word aloud and rated it on pronunciation difficulty, with 1 being "very
easy" and 5 being "very difficult." Each word remained on the screen for four seconds or until
the response, whichever came first. Immediately after the tenth word, a
question appeared on the computer monitor, and participants responded by typing
on the keyboard 1, 2, or 3 to indicate TOT, Don't Know or Know, respectively.
After a Know response, participants typed in the word and then immediately the
next trial began with presentation of the pronunciation list for the next
question. After a TOT or Don't
Know response, the next trial began immediately with presentation of the pronunciation
list. Each question was presented
only once and no backtracking or changing previous responses was possible.
After two practice trials and the first 60 experimental trials, participants took a break during which they completed a background questionnaire and the vocabulary test. Then they performed the remaining 54 trials. Following all trials, questions that had elicited a TOT response were presented in the paper and pencil multiple choice recognition test. Participants were told to circle the word for which they were having a TOT, to verify that their TOT state was for the intended target word.
Table 1 presents the percentages of correct Know, TOT, and Don't Know responses following related (primed) and unrelated (unprimed) lists by age group. The analysis of Know responses excluded trials on which a participant responded Know and typed an incorrect answer. Such incorrect Know responses occurred on 9.2% and 15.7% of trials for young and older adults, respectively. The analysis of TOT responses excluded trials on which a participant responded TOT but did not select the target word on the multiple choice test. Such incorrect TOTs occurred on 5.0% and 3.9% of trials for young and older adults, respectively. An ANOVA with age and prime condition as factors was performed on number of responses, separately for Know and TOT responses. Because the different types of responses are not independent, we limit the analyses to these primary measures. We report analyses by participant (F1) and by item (F2).
More correct Know responses followed related than unrelated lists, F1(1,70) = 12.19, MSE = 21.23, p<.01, F2 (1,113) = 25.20, MSE = 3.38, p < .01, and older adults produced more correct Know responses than young adults, F1(1,70) = 4.19, MSE = 112.87, p<.05, F2(1,113) = 9.72, MSE = 15.85, p < .01. There was no prime condition by age interaction, F1 (1,70) = .003, MSE = 21.23 and F2 = .01, MSE = 3.36. As can be seen in Table 1, the priming effect (the difference between primed and unprimed Know responses) was virtually identical for young and older adults, 4.6 and 4.8 respectively. The unbiased effect size for the age difference in priming effects, calculated following the procedures of Wolf (1986) and Hedges (1984), was essentially zero, d = .006, suggesting that the absence of a significant age interaction with prime condition was not a power problem.
Fewer TOTs followed
related than unrelated lists, F1(1,70) = 6.34, MSE = 7.18, p<.05,
F2 (1,113) = 6.78, MSE = 2.17, p < .05. This effect is
more apparent in absolute number of TOTs than in percentage, with the young M
= 5.86 and 6.69 and the old M = 6.44 and 7.86 for related and unrelated
lists, respectively. The effect of age on number of TOTs was not significant, F1(1,70)
= 1.04, MSE = 26.43, F2 (1,113) = 2.61, MSE =3.23, both p's
> .10, and there was no age by prime condition interaction, F1(1,70) = .43, MSE =
3.06, F2 (1,113) = .49, MSE = .88, both p's > .40. As
can be seen in Table 1, the priming effect was slightly smaller for young than
older adults, 1.4 and 2.5, respectively.
The unbiased effect size for the age difference in priming effects, d
= .155, is a small effect size (Cohen, 1992). To obtain significance with a small effect size at p = .05
and .80 level of power requires 393 participants per group (Cohen, 1992).
Although TOTs usually increase in old
age, the lack of age differences here is unsurprising because we selected
questions that produced the most TOTs and smallest age difference in TOTs in
previous studies. Nonetheless, TOTs can only occur when a word is not
immediately recalled and older adults produced more correct answers than young
adults. To adjust for any possible age differences in opportunity for TOTs, we
calculated TOTs as a proportion of unsuccessful retrievals, e.g., incorrect
Know, incorrect TOT, TOT (correct) and Don't Know responses (A. Brown &
Nix, 1996; Burke et al., 1991 ). The proportion of TOTs (converted in Table 1
to percentage) was lower following related than unrelated lists, F1(1,70)
= 5.09, MSE = .00, p<.05, F2 (1,113) = 6.43, MSE
= .00, p < .05. This
relative proportion of TOTs was greater for older than young adults, F1(1,70)=5.23,
MSE=.02, p<.05, F2 (1,113) = 3.53, MSE = .00, p
= .06.
To summarize these
results, prior processing of phonologically related words increased correct
retrieval of target words and reduced TOTs. The pattern of priming effects is
consistent with the TD model that production strengthens connections in the
phonological system, reducing deficits in transmission of priming and thereby
reducing TOTs for words that share the same connections (see Figure 1). There
was no age difference in phonological priming effects on production of correct
responses or TOTs, and the age difference in TOTs as a proportion of incorrect
responses remained in the primed condition. Under the TD model, the baseline
strength of connections is reduced by aging, making older adults more
susceptible to retrieval failures. Within this framework, the age invariance in
priming effects indicates that increments in connection strength caused by
activation are similar across age. The age invariance in priming effects also
is inconsistent with a strategic basis for priming because episodic recall
declines with age (e.g., Burke & Light, 1981) so young adults would be more
successful than older adults in remembering list words after the question was
presented, and this would yield greater cuing effects, were such a strategy
used. Finally, both the age invariance in priming effects and the positive
nature of priming effects provide no support for inhibitory effects of
phonologically related words.
Experiment 2
In Experiment 2, the
pronunciation task was presented after the general knowledge question and only
when there was no word produced in response to the question: When participants
responded TOT or Don't Know, they performed the pronunciation rating task and
then again attempted to answer the question. Under the TD model, processing
phonologically related words should increase the strength of weak phonological
connections causing TOTs and thereby promote retrieval and resolution of the
TOT. In addition to the theoretical implications, observing this predicted
effect would provide insight into a possible mechanism producing ñspontaneousî
resolution of naturally occurring TOTs, namely that they are triggered by
phonological priming from language input.
On the other hand, if phonologically related words block target
retrieval either through inhibition (Jones, 1989) or through their increased
accessibility (Burke et el., 1991) then they should reduce TOT resolution.
Participants.
Eighteen young adults (M=18.6
years, SD=1.4) and 18 healthy older adults (M=71.8 years, SD=5.1)
participated in the experiment.
Young adults received credit in Introductory Psychology courses and
older adults were paid for their participation. Mean score on the Nelson-Denny vocabulary measure was lower
for young (M=17.4, SD=2.0) than older participants (M=22.5,
SD=1.3), t(33)= 8.78, p<.001. Mean years of education were 13.1 (SD=1.1) and 17.1 (SD=2.3)
for young and older adults, respectively.
All participants were native English speakers.
Materials. Eighty general knowledge questions were selected from the same sources as in Experiment 1 and in some cases were the same questions used in Experiment 1. The phonologically related and unrelated lists and the multiple-choice recognition test were constructed as in Experiment 1 (see Appendix). For each participant, half the questions were associated with a related list and half with an unrelated list with the assignment of questions to list type counterbalanced over participants. This assignment was done in advance of the participants' responses but it determined which prime condition was presented when a TOT or Don't Know response was given.
Procedure.
The computer program designed for Experiment 1 was modified so that each trial
began with the the general knowledge question. Participants indicated on the
keyboard one of three responses: Know, Don't Know or TOT as in Experiment 1. After a Know response, participants
typed in the word. After a TOT or
Don't Know response, a related or unrelated list of words was presented one
word at a time. Participants
pronounced each word aloud and rated it on pronunciation difficulty, with 1
being "very easy" and 5 being "very difficult." Then the question was
re-presented. Participants who now
knew the answer typed in the word and proceeded to the next trial, and
participants who still had a TOT or did not know the answer went immediately to
the next trial. No backtracking or
changing previous responses was possible.
Participants
were given two practice trials followed by 80 experimental trials. After completing a background
questionnaire and the vocabulary measure, participants were re-presented the
questions that had elicited a TOT response on their second presentation.
Questions that were not answered on this post-session presentation were
presented in a multiple choice recognition test. All other aspects of the
procedure followed Experiment 1.
Table
2 presents the percentage of correct Know, TOT, and Don't Know responses on the
initial presentation of the questions by age group. The analysis of Know
responses excluded trials on which a participant responded Know and typed an
incorrect answer, 10.4% and 16.3% of trials for young and older adults,
respectively. The analysis of TOT responses excluded trials on which a
participant responded TOT initially, but produced an incorrect answer on a
subsequent presentation of the question or did not select the target word on
the multiple choice test, 6.9% and 8.0% of trials for young and older adults,
respectively .
One way ANOVAs on number of responses were performed separately for Know and TOT responses with age as a factor using both participants (F1) and items (F2) as unit. Older adults produced more Know responses than young adults but this difference was smaller than in Experiment 1 and was not significant, F1 (1,34) = 1.12, MSE = 98.78, p = .30, F2 (1,79) = 2.18, MSE = 11.40, p = .14. Older adults produced more TOTs than young adults, although the effect reached significance only in the item analysis, F1 (1,34) = 2.75, MSE = 25.29 p < .11, F2 (1,79) = 6.63, MSE = 1.91, p < .02. As in Experiment 1, the proportion of unsuccessful retrievals (Incorrect Know, Don't Know, TOT (correct) and incorrect TOT responses) that were correct TOTs was greater for older than young adults, F1 (1,34) = 4.59, MSE=.009, p<.05, F2 (1,79) = 8.41, MSE = .03, p < .01. These proportions are shown as percentages in Table 2.
The effect of priming on
resolution of TOT and Don't Know responses is shown in Table 3 for each age
group. These data combine responses on the second and the final (end of
session) presentations of the questions. Three young participants were excluded
from this analysis because they had no TOTs in one of the prime conditions. An
ANOVA with age, prime condition (related, unrelated), and original response
(Don't Know, TOT) as variables was performed on proportion of correct
responses. The analysis was performed with participants (F1) but not
question (F2) as the unit because so few questions elicited TOTs for
both young and older adults in both primed and unprimed conditions (see R.
Brown & McNeill, 1966 and Burke et al., 1991 for discussion of the
fragmentary data problem characteristic of TOTs). More resolutions followed TOT than Don't Know responses F(1,31)=165.51,
MSE =.064, p<.001, and related than unrelated lists, F(1,31)=
7.75, MSE =.037, p<.01. As can be seen in Table 3, the priming
effect depended on original response, occurring for TOTs but not for Don't Know
responses, F(1,31) = 7.19, MSE =.042, p <.02. There was
no significant age by prime condition interaction, F(1,31) = .39, MSE
=.042, and no other significant main effects or interactions. As can be seen
in Table 3, the priming effect was slightly larger for young than older
adults. The unbiased effect size
for the age difference in priming effects, d = .219, is a small effect
size (Cohen, 1992) which requires approximately 393 participants per group to
reach significance at p = .05 and a .80 level of power (Cohen , 1992).
Note however, that the direction of the age difference in priming effects is in
the opposite direction from Experiment 1.
Because of the small number of TOTs in
the primed and unprimed condition in Table 3, we also conducted a sign test to
determine if the number of participants who recalled more targets in the primed
condition than in the unprimed condition was greater than chance. Nine out of
13 young adults (excluding ties) recalled more targets in the primed than
unprimed condition, p = .13 and 12 out of 15 older adults did so, p
< .05. For both ages combined,
21 out of 28 participants showed the priming effect, p < .01.
There is no evidence
that the related prime condition improved retrieval following Don't Know
responses. We evaluated the
possibility that the related prime condition increased the frequency that Don't
Know responses were changed to correct TOT responses on the second or third
question presentation. If the related prime condition increased priming at the
lexical level in cases where participants believed they did not know the target
word, then an increase in TOTs would occur (cf., Meyer & Bock, 1992). This
occurred very infrequently, however, on less than 3.5% of trials in any
condition or age, and there was no significant effect of age or condition. The
lack of benefit from the related prime condition following a Don't Know
response contrasts with Kozlowski's (1977) finding that a rhyming cue word
increased target recall by the same amount both when participants were in a TOT
state and when the answer was unknown. Kozlowski told participants that the cue
word rhymed with the target sometimes and a rhyme cue was presented on half of
the trials. Thus it seems likely that participants were able to use the cue in
guessing strategies and to guess the correct answer regardless of whether or
not they were in a TOT state. The absence of any effect of the related prime
condition following Don't Know responses in the present study is consistent
with participants being unaware of the the relation of list words to the
target.
Anecdotal
comments during participant de-briefing were also consistent with participants'
lack of awareness. Some subjects reported that they thought the purpose of the
list words was to distract them from
trying to answer the question and make retrieval more difficult, e.g., "I
thought the list words were to confuse me." There was nothing in their spontaneous comments to suggest
that participants had gleaned the structure of the lists or had used a cuing
strategy. When de-briefed about
the phonological relatedness, one participant replied, "I noticed that
once after I remembered the word."
Not one subject reported using list words as cues.
Given
the absence of any indication that the benefit from the related prime condition
was a consequence of strategies, the results support the TD hypothesis that
TOTs are caused by weak phonological connections that can be strengthened by
activation of the relevant phonological nodes, enabling production of the
target. These findings support the
argument that ñspontaneousî resolution of TOTs may be triggered by language
input involving critical phonological components of the TOT target. Older
adults reported more TOTs, but the effect of phonological processing was
invariant across age suggesting that phonological activation produced a
comparable change in connection strength in young and older adults. The data provide no support for the
view that phonologically related words block target retrieval.
General Discussion
The
results demonstrate that processing phonologically related words reduces TOTs
and increases retrieval of answers to general knowledge questions, under
conditions that minimize the opportunity for strategic processes. In Experiment
1, prior production of words phonologically related to a target word increased
production of the target in response to a general knowledge question and
decreased the probability of TOT. Older adults produced relatively more TOTs
than young adults, but because priming effects were age-invariant, the age
difference remained in the primed condition. In Experiment 2, production of
phonologically related words after the onset of a TOT increased the probability
of resolution of the TOT. There was no comparable positive priming effect on
retrieval following a Don't Know response. Older adults again produced more
TOTs and priming effects on resolution were age invariant. These results
contribute to our understanding of the mechanism producing TOTs and age-related
changes in TOTs, and highlight the significance of priming effects in speech
production.
Interpretation of these results in terms of general language processes would be precluded by evidence that the results were influenced by strategies that are not part of normal language processing. The evidence suggests, however, that the experimental paradigm was successful in avoiding contamination by strategies. First, if related lists increased correct responses in Experiments 1 and 2 because participants compared possible responses to memorized list words, or guessed the target based on memorized list words, this effect should certainly be smaller in older than young adults, contrary to the findings. Older adults would be less able to remember list words because of age-related declines in episodic recall (e.g., Burke & Light, 1981). Second, the lack of priming effects on target retrieval following Don't Know responses in Experiment 2 provides counter evidence to the use of strategies to guess the target. Finally, anecdotal comments during participant de-briefing were consistent with participants being unaware of the relation of lists to targets.
Transmission Deficits versus Blocking as a
Cause of TOTs
The results
support the TD hypothesis that TOTs are caused by weak phonological connections
that transmit insufficient priming to retrieve the phonology corresponding to
an activated lexical representation for the target word. Processing phonologically related words
counteracts these transmission deficits by activating phonological
representations shared with the target, thereby strengthening connections and
enabling target retrieval. The
representations for prime and target words overlap only at the phonological
level (see Figure 1). Thus, the
faciliatory priming effect locates the TOT deficit in these phonological
connections under the TD model. The present research does not identify which
aspects of the phonological overlap between prime and target words are critical
for priming effects, but recent research in a similar paradigm has shown larger
priming effects when words overlap in the initial rather than middle or final
syllable (White & Abrams, 1999).
Would this increase in
phonological connection strength be expected to facilitate lexical selection,
for example, by improving target retrieval following a Don't Know response
which occurs when no lexical node has been activated? Lexical selection is
driven primarily by top-down semantic processes triggered by the question;
phonological processes may play a role in lexical selection under interactive
models of production, but not sequential stage models. In stage models of
production (e.g., Levelt, 1989; Levelt, Schriefers, Vorberg, Meyer, Pechmann
& Havinga, 1991), lexical and phonological retrieval are modular,
sequential stages; lexical access is accomplished on the basis of semantic
information without input from the phonological system, so changes in
phonological connections would have no impact. In
NST and other interactive activation models (e.g., Cutting & Ferreira,
1999; Dell, 1986; Dell & O'Seaghdha, 1991; MacKay, 1987; Martin, Weisberg
& Saffran, 1989), priming spreads bi-directionally between lexical and
phonological levels during lexical selection. Strengthening phonological connections
relevant to the target word would increase transmission of priming back up to
the target lexical node during lexical selection. In a Don't Know state where
no word is selected, however, there is no basis for assuming that the target
lexical node was even competing for lexical selection. Thus, it is unsurprising
that phonological priming produced no improvement in correct retrieval
following Don't Know responses in Experiment 2.
The
results provide no support for an inhibitory effect of phonologically related
words on target retrieval. Some
investigators have suggested that phonologically related alternate words that
come to mind during a TOT are "blockers" that cause the TOT by
inhibiting the intended word (e.g., Jones, 1989; Reason & Lucas, 1984). The
only direct evidence that phonologically related words increase TOTs
(Jones, 1989; Jones & Langford, 1987) appears to be a consequence of the
specific questions assigned to this condition (see Meyer & Bock, 1992;
Perfect & Hanley, 1992). Thus,
despite the strong phenomenological experience that alternate words
"block" target retrieval, direct experimental tests of an inhibition
model of TOTs have yielded negative results. Nonetheless, further research on inhibitory effects may be
motivated by findings that semantic similarity between prime and target
words under certain conditions slowed naming latency (Wheeldon &
Monsell, 1994), increased naming errors (e.g., Vitkovitch & Humphreys,
1991), and increased TOTs but without reducing correct recall (Meyer &
Bock, 1992). Moreover, resolution
time was longer for naturally occurring TOTs with, than without, a persistent
alternate word that came repeatedly to mind. Alternate words shared initial
phoneme, number of syllables and syntactic class with target words at greater
than chance probability (Burke et al, 1991), but their semantic overlap has not
been evaluated. No research has yet constrained overlap between a
phonologically related prime word and target on all these dimensions shown to
be important in naturally occurring alternates. It remains for future research
to determine under what conditions phonologically related words, shown to
facilitate TOT resolution in the present research, will impede TOT resolution
as do spontaneous persistent alternates.
Locus of Priming Effects
Wheeldon
and Monsell (1992) reported faster picture naming latency with prior production
of a picture name, (e.g., sun), but not a homophone of the picture name
(e.g., son). Based on these findings, they suggested that priming effects
on word production require prime-target overlap at the lexical level, which
does not occur with homophone primes.
The present results, however, demonstrate priming effects that depend on
overlap at the phonological, not the lexical level. The discrepancy between the
Wheeldon and Monsell results and the present results is likely because of the
difference in the frequency of occurrence of the target words in the two
studies. That is, priming at the
phonological level may only be observed in production tasks involving words
with weak phonological connections, for example, low frequency words with
relatively rare phonological components.
TOTs are more likely for low than high frequency words and for words with
few rather than many phonological neighbors (Harley & Bown, 1998) because
the infrequent activation of phonological components of these words makes them
vulnerable to transmission deficits. Such words would be expected to benefit
from activation of phonological components via pronounciation of a phonologically
related prime word.
In contrast, Wheeldon
and Monsell (1992) used words that name common objects (e.g., sun) and
are unlikely candidates for weak phonological connections because of their
relatively high frequency of occurrence. Such words may show no measurable
long-term priming effect from phonologically related words because their
connection strength at the phonological level is very high. Note that this argument holds for
longer-term priming effects based on changes in connection strength; it does
not apply to short-lived priming effects occurring with brief prime-target
intervals and resulting from a spread of priming from prime to target
nodes. There is considerable
evidence that the latter mechanism produces phonological priming effects on
high frequency words (e.g., Collins & Ellis, 1992; Schriefers, Meyer &
Levelt, 1990).
Priming, Aging, and Word Retrieval
Evidence for older
adults' word retrieval problems extends beyond studies of TOTs: In their spoken discourse, older adults
produced more pronouns and ambiguous references compared to young adults
(Cooper, 1990; Heller & Dobbs, 1993; Pratt, Boyes, Robins & Manchester,
1989) and more disfluencies such as filled pauses, repetitions and hesitations,
phenomena that appear to be linked to word retrieval difficulties (Cooper,
1990; Kemper, 1992). Under the TD
model, word retrieval failures increase in old age because aging reduces the
strength of connections in memory. Diverging one-to-one connections, such as
those involved in top-down phonological encoding (see Figure 1) are especially
sensitive to transmission deficits caused by weakened connections because there
are not additional sources of priming to compensate for reduced transmission
(e.g., Burke et al., 1991; MacKay & Abrams, 1996; MacKay & Burke,
1990). We consider now what the age invariance in priming effects indicates
about the nature of age differences in transmission deficits3.
Repetition priming
effects rarely vary with age in individual studies, although small declines in
older adults' priming effects emerge in meta-analyses. Some of the declines can
be attributed to contamination by episodic memory which declines with age, but
this factor alone cannot explain the extant results which are not well
understood (Fleischman & Gabrieli, 1998; LaVoie & Light, 1994). In the present research, the unbiased
effect size for age differences in priming effects was virtually zero for
correct recall (Experiment 1), and was in the category of small effect sizes
(Cohen, 1992) for TOTs in Experiments 1 and 2. These age differences for TOT priming effects, however, were
in opposite directions with young adults showing smaller priming effects than
older adults in Experiment 1 and larger effects in Experiment 2.
The lack of reliable age
differences in priming effects on TOTs has clear implications within the TD
model. First, it suggests that the
increment in connection strength occurring after activation of phonological
nodes is unaffected by aging. Second, it is consistent with older adults suffering
not only more frequent transmission deficits than young adults, but also more severe transmission
deficits. Age differences in TOTs
remained in the related prime condition in Experiment 1. This would be expected
if phonological activation increased connection strength, and thus priming
transmission, the same amount across age: More transmission deficits would
remain in older than young adults because their deficits were more severe and
the increased strength still did not allow a critical level of priming
transmission. More frequent transmission deficits is consistent with the
greater frequency of TOTs with aging and more severe transmission deficits is
consistent with older adults reduced access to partial phonological information
during a TOT relative to young adults (A. Brown & Nix, 1996; Burke et al.,
1991; Heine et al., 1999; Maylor, 1990a).
Priming and ñSpontaneousî Resolution of TOT
The percent of TOTs that
are resolved spontaneously through "pop-ups" varies from 17% to 61%
in studies of naturally occurring TOTs (Burke et al., 1991; Cohen &
Faulkner, 1986; Heine et al., 1999; Reason & Lucas, 1984) and is somewhat
lower when TOTs are induced in the laboratory (Burke et al., 1991; Read &
Bruce, 1982; see Brown, 1991 for a review). For older adults, the majority of their naturally occurring
TOTs are resolved spontaneously and they report that this is their preferred
mode of resolution (Burke et al, 1991; Heine et al., 1999). Spontaneous
resolutions are an enigmatic experience and an intriguing problem in terms of
mechanism. The present results support the view that ñspontaneousî resolution
is triggered when the person inadvertently perceives or produces a word that
contains the missing phonology.
This activates inaccessible phonological components and may eliminate
transmission deficits. The person
may experience the TOT resolution as spontaneous if he or she is unaware of the
phonological similarity between the triggering word and the target word (e.g.,
Burke et al., 1991; Seifert, Meyer, Davidson, Patalano & Yaniv, 1994; Yaniv
& Meyer, 1987).
Under
this explanation, spontaneous resolution of TOTs is an instance of a
provocative phenomenon in problem solving, namely, the sudden appearance in
consciousness of a solution to a problem that had eluded conscious effort
(e.g., Polya, 1957). Our results
support the view that for some types of problems the "subconscious
work" (Polya, 1957, p.198) underlying resolution involves a mental
operation wherein critical components of the solution become more accessible
when relevant stimuli are encountered in the environment. The solution appears
spontaneous and unexpected because it was triggered by a repetition priming
process that occurs without awareness (cf. Posner, 1973; Yaniv & Meyer,
1987; see Seifert et al., 1994 for empirical and anecdotal evidence for
spontaneous solutions in problem solving).
The experimental
paradigm used in Experiment 2 differs in an important way from spontaneous
resolutions of TOTs that occur in everyday life. In Experiment 2, the question
was re-presented, whereas naturally occurring ñpop-upsî are reported as
occurring when attention is no longer directed towards resolving the TOT
experience (e.g., Burke et al, 1991).
How can target retrieval occur when there is no top-down conceptual
process driving it? An important
problem for future research is investigating whether phonological priming
effects on TOTs extend to spontaneous retrieval of the target word, in the
absence of the question.
Blaxton,
T. (1989). Investigating
dissociations among memory measures: Support for a transfer-appropriate
processing framework. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 15,
657-668.
Bock,
K., & Levelt, W. (1994). Language Production: Grammatical Encoding. In M. Gernsbacher (Ed.), Handbook of
Psycholinguistics (pp.
945-984). San Diego, CA: Academic
Press.
Bowles, N. L., Obler, L.
K., & Poon, L. W. (1989). Aging and word retrieval: Naturalistic, clinical
and laboratory data. In L. W. Poon, D. C. Rubin, & B. A. Wilson (Eds.), Everyday
cognition in adulthood and late life (pp. 244-264). Cambridge, UK:
Cambridge University Press.
Brennen, T., Baguley,
T., Bright, J., & Bruce, V. (1990).
Resolving semantically-induced tip-of-the-tongue states for proper nouns. Memory and Cognition, 18,
339-347.
Brown,
A.S. (1979). Priming effects in
semantic memory retrieval processes.
Journal of Experimental Psychology: Human Learning and Memory, 5,
65-77.
Brown,
A.S. (1991). The tip of the tongue
experience: A review and evaluation.
Psychological Bulletin, 10, 204-223.
Brown, A. S., Neblett,
D. R., Jones, T. C., & Mitchell, D. B. (1991). Transfer of processing in
repetition priming: Some inappropriate findings. Journal of Experimental
Psychology, 17, 514-525.
Brown, A.S., & Nix,
L.A. (1996). Age-related changes in the tip-of-the-tongue experience. American
Journal of Psychology, 109, 79-91.
Brown,
R., & McNeill, D. (1966). The
"tip of the tongue" phenomenon. Journal of Verbal Learning
Behavior, 5, 325-337.
Burke,
D., & Light, L. (1981). Memory
and aging: The role of retrieval processes. Psychological Bulletin, 90,
513-554.
Burke,
D.M., & MacKay, D.G. (1997). Memory, language and ageing. Philosophical
Transactions of the Royal Society: Biological Sciences, 352, 1845-1856.
Burke,
D.M., MacKay, D.G., Worthley, J.S., & Wade, E. (1991). On the tip of the tongue: What causes
word finding failures in young and older adults? Journal of Memory and Language, 30, 542-579.
Caramazza, A. (1997).
How many levels of processing are there in lexical access? Cognitive
Neuropsychology, 14, 177-208.
Cohen, J, (1992). A
power primer. Psychological Bulletin, 112 155-159.
Cohen,
G., & Faulkner, D. (1986).
Memory for proper names: Age differences in retrieval. British Journal of Developmental
Psychology, 4, 187-197.
Collins, A. F., & Ellis, A. W. (1992). Phonological
priming of lexical retrieval in speech production. British Journal of
Psychology, 83, 375-388.
Cooper,
P.V. (1990). Discourse production
and normal aging: Performance on oral picture description tasks. Journal of Gerontology:
Psychological Sciences, 45, P210-214.
Cutting, C. J., & Ferreira, V. S. (1999). Semantic
and phonological information flow in the production lexicon. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 25, 318-344.
Dell, G.S. (1986). A spreading-activation theory of
retrieval in sentence production. Psychological
Review, 93, 283-321.
Dell,
G.S., & O'Seaghdha, P.G. (1991).
Mediated and convergent lexical priming in language production: A
comment on Levelt et al. (1991). Psychological
Review, 98, 604-614.
Durso, F. T., & Johnson, M. K. (1979).
Facilitation in naming and categorizing repeated pictures and words. Journal
of Experimental Psychology: Human Learning and Memory, 5, 449-459.
Fleischman, D. A., & Gabrieli, J. D. E. (1998).
Repetition priming in normal aging and alzheimer's disease: A review of
findings and theories. Psychology and Aging, 13, 88-119.
Francis,
W.N., & Kucera, H. (1982). Frequency
analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin.
Freedman, J. L., & Landauer, T. K. (1966). Retrieval
of long-term memory: ñtip-of-the-tongueî phenomenon. Psychonomic
Science, 4, 309-310.
Gruneberg, M. M., &
Monks, J. (1974). ñFeeling of Knowingî and cued recall. Acta Psychologica,
38, 257-265.
Hanley, J. R., &
Cowell, E. S. (1988). The effects of different types of retrieval cues on the
recall of names of famous faces. Memory & Cognition, 16, 545-565.
Harley, T. A., &
Bown, H. E. (1998). What causes a tip-of-the-tongue state? Evidence for lexical
neighbourhood effects in speech production. British Journal of Psychology,
89, 151-174.
Hasher,
L., & Zacks, R.T. (1988).
Working memory, comprehension, and aging: A review and a new view. In G.H. Bower (Ed.), The psychology of learning and
motivation Vol.22 (pp. 193-225).
San Diego, CA: Academic Press.
Hedges,
L.R. (1984). Advances in statistical methods for meta-analysis. In J.W. Yeaton
& P.M. Wortman (Eds.), Issues in data synthesis (pp.25-42). San
Francisco: Jossey-Bass.
Heine, M. K., Ober, B.
A., & Shenaut, G. K. (1999). Naturally occurring and experimentally induced
tip-of-the-tongue experiences in three adult age groups. Psychology and
Aging, 14, 445-457.
Heller,
R.B., & Dobbs, A.R. (1993).
Age differences in word finding in discourse and nondiscourse
situations. Psychology and
Aging, 8, 443-450.
Jones,
G.V. (1989). Back to Woodworth:
Role of interlopers in the tip of the tongue phenomenon. Memory and Cognition, 17,
69-76.
Jones,
G.V., & Langford, S. (1987).
Phonological blocking in the tip of the tongue state. Cognition, 26, 115-122.
Kelley,
C.M., & Lindsay, D.S. (1993).
Remembering mistaken for knowing: Ease of retrieval as a basis for
confidence in answers to general knowledge questions. Journal of Memory and Language, 32, 1-24.
Kemper,
S. (1992). Adults' sentence fragments: Who, what, when, where, and why. Communication Research, 19, 444-458.
Kohn,
S.E., Wingfield, A., Menn, L., Goodglass, H., Berko Gleason, J., & Hyde, M.
(1987). Lexical retrieval: The
tip-of-the-tongue phenomenon. Applied
Psycholinguistics, 8, 245-266.
Koriat, A., &
Lieblich, I. (1974). What does a person in a ñTOTî state know that a person in
a ñdon't knowî state doesn't know.
Memory and Cognition, 2, 647-655.
Kozlowski,
L.T. (1977). Effects of distorted
auditory and of rhyming cues on retrieval of tip-of-the-tongue words by poets
and nonpoets. Memory and
Cognition, 5, 477-481.
Laver,
G.D., & Burke, D.M. (1993). Why do semantic priming effects increase in old
age? A meta-analysis. Psychology and Aging, 8, 34-43.
LaVoie,
D., & Light, L.L. (1994).
Adult age differences in repetition priming: A meta-analysis. Psychology and Aging, 9,
539-553.
Levelt,
W.J.M. (1989). Speaking: From
intention to articulation.
Cambridge, MA: MIT Press.
Levelt,
W.J.M., Schriefers, H., Vorberg, D., Meyer, A.S., Pechmann, T., & Havinga,
J. (1991). The time course of
lexical access in speech production: A study of picture naming. Psychological Review, 98,
122-142.
Light, L. (1992). The
organization of memory in old age. In F.I.M. Craik & T.A. Salthouse
(Eds.), The handbook of aging
and cognition (pp. 111-166). Hillsdale, NJ: Lawrence Erlbaum Associates.
Lovelace,
E.A., & Twohig, P. T. (1990). Healthy older adults' perceptions of their
memory functioning and use of mnemonics. Bulletin of the Psychonomic Society,
28, 115-118.
MacKay,
D.G. (1987). The organization
of perception and action: A theory for language and other cognitive skills.
New York: Springer-Verlag.
MacKay,
D.G., & Abrams, L. (1996).
Language, memory and aging: Distributed deficits and the structure of
new versus old connections. In
J.E. Birren & W.K. Schaie (Eds.),
Handbook of the psychology of aging: Fourth Edition (pp.
251-265). San Diego: Academic
Press.
MacKay,
D.G., & Burke, D.M. (1990).
Cognition and aging: A theory of new learning and the use of old
connections. In T. Hess (Ed.), Aging
and cognition: Knowledge organization and utilization (pp.213-263). Amsterdam: North Holland.
Martin,
N., Weisberg, R.W., & Saffran, E.M. (1989). Variables influencing the occurrence of naming errors:
Implications for a model of lexical retrieval. Journal of Memory and Language, 28, 462-485.
Maylor,
E.A. (1990a). Age, blocking and
the tip of the tongue state. British
Journal of Psychology, 81, 123-134.
&nbs