Kwantlen Polytechnic University

KORA: Kwantlen Open Resource Access
Honours Theses

Student Scholarship

Spring 2014

Proportional Manipulation of Produced Words
Tests the Distinctiveness and Strength Accounts of
the Production Effect
Dawn-Leah L. McDonald
Kwantlen Polytechnic University

Follow this and additional works at: http://kora.kpu.ca/honours_theses
Part of the Cognitive Psychology Commons
Recommended Citation
McDonald, Dawn-Leah L., "Proportional Manipulation of Produced Words Tests the Distinctiveness and Strength Accounts of the
Production Effect" (2014). Honours Theses. Paper 3.

This Honours Thesis is brought to you for free and open access by the Student Scholarship at KORA: Kwantlen Open Resource Access. It has been
accepted for inclusion in Honours Theses by an authorized administrator of KORA: Kwantlen Open Resource Access. For more information, please
contact kora@kpu.ca.

Running Head: PRODUCTION EFFECT

1

Proportional Manipulation of Produced Words Tests the Distinctiveness and Strength Accounts
of the Production Effect
Dawn-Leah L. McDonald
Kwantlen Polytechnic University
April 30, 2014

PRODUCTION EFFECT

2
Abstract

The production effect occurs when producing items enhances memory relative to reading items
silently. Although the most common mode of production used in research has been “reading
aloud”, singing, mouthing, whispering, writing, typing, and spelling out also produce a memory
advantage. Distinctiveness and strength of encoding have both been offered as explanations for
the memory enhancement. Production, relative to silent reading, makes items distinctive and
therefore, memorable. Production increases memory traces at encoding (strength). Two
experiments (N = 272) tested the distinctiveness and strength accounts by manipulating the
proportion of words produced at study. Experiment 1 involved the production method of reading
aloud in a classroom setting. Experiment 2 involved the production method of writing, tested in
a classroom setting, small groups, and individuals. Our results were consistent with both
accounts; however, neither account can explain our full data pattern.
Keywords: production effect, memory, distinctiveness, strength, proportions

PRODUCTION EFFECT

3

Proportional Manipulation of Produced Words Tests the Distinctiveness and Strength Accounts
of the Production Effect
The production effect, although studied for decades (Conway & Gathercole, 1987; Gathercole
& Conway, 1988; Hopkins & Edwards, 1972), was only defined and named in recent years
(MacLeod, Gopie, Hourihan, Neary, & Ozubko, 2010). The production effect appears as a
memory advantage for words that are produced, over words that are read silently. Words may be
produced in various ways such as reading aloud normally, reading aloud loudly, singing,
mouthing, whispering, writing, and typing (Conway & Gathercole, 1987; Forrin, MacLeod, &
Ozubko, 2012; Gathercole & Conway, 1988; Hopkins & Edwards, 1972; MacLeod et al., 2010;
Ozubko, Gopie, & MacLeod, 2012; Ozubko & MacLeod, 2010; Quinlan & Taylor, 2013).
There has been ongoing debate about what underlies the production effect. The two
explanations providing the strongest evidence are the distinctiveness account (Bodner & Taikh,
2012; Conway & Gathercole, 1987; Dodson & Schacter, 2001; Forrin et al., 2012; MacLeod et
al., 2010; Ozubko et al., 2012; Ozubko & MacLeod, 2010; Ozubko, Major, & MacLeod, in
press) and the strength account (Bodner & Taikh, 2012; Bodner, Taikh, & Fawcett, 2014;
Fawcett, 2013; MacLeod et al., 2010; Ozubko & MacLeod, 2010).
The distinctiveness account suggests that production provides an additional mode of encoding
to make the produced item stand out from the unproduced items (Bodner & Taikh, 2012, Castel,
Rhodes, & Friedman, 2013; Forrin et al., 2012; MacLeod et al., 2010; Ozubko et al., 2012;
Ozubko & MacLeod, 2010). The modes would be articulatory, in the case of mouthing and
reading aloud, and motor, in the case of writing or typing. The strength account posits that the
production of items creates stronger memory traces, thereby making the items easier to recollect

PRODUCTION EFFECT

4

later (Bodner & Taikh, 2012; Bodner et al., 2014; Fawcett, 2013; MacLeod et al., 2010; Ozubko
& MacLeod, 2010).
Research in this area has found that the memory performance increase due to the production
effect appears almost exclusively in mixed-list within-subject designs, where the participant
studies lists composed of items that are to be both produced and unproduced (Conway &
Gathercole, 1987; Forrin, et al., 2012; Hopkins & Edwards, 1972; MacLeod et al., 2010; Ozubko
et al., 2012; Ozubko & MacLeod, 2010). As an example of a mixed-list within-subject design,
participants read some items silently and some words aloud normally, experiencing both the
produced and unproduced conditions. This memory performance increase has not been shown in
a pure-list between-subjects design. Pure lists consist of only one condition. That is, participants
either produce all of the items on the study list or they read all of the items silently. It appears
that for items to seem unique or distinctive, it is necessary to have other items with which to
compare them. The between-subjects design does not provide participants with a manner of
comparison, thereby removing the opportunity to use distinctiveness.
Much of the research conducted thus far suggests that the most prominent feature of
distinctiveness is that it is relative (Bodner et al., 2014; Conway & Gathercole, 1987; Lin &
MacLeod, 2012). Participants must be able to determine at the time of retrieval that the
produced word was different from the word that was not produced. The assessment of whether
the word had been seen before could be determined by using a distinctiveness heuristic (Dodson
& Schacter, 2001; Forrin, MacLeod, & Ozubko, 2012; MacLeod, 2011; Ozubko & MacLeod,
2010; Ozubko et al., in press). Curiously, participants seem to have also used this heuristic in
reverse, determining that the item could not have been studied if it was not distinctive (Bodner &
Taikh, 2012; Dodson & Schacter, 2001).

PRODUCTION EFFECT

5

When groups of younger adults with an age range of 17 to 25 years and older adults with an
age range of 67 to 88 years were tested, both groups benefited from production in their memory
retrieval, but older adults showed a smaller advantage (Lin & MacLeod, 2012). These results
were interpreted as evidence for distinctiveness as an explanation for the production effect. Lin
and MacLeod (2012) stated that older adults experienced more difficulty in retrieving distinctive
information than younger adults.
There may be more than one explanation for the production effect, with more than one
process contributing to the memory benefits. A meta-analysis of production effect experiments
using a between-subjects design prompted Fawcett (2013) to suggest that strength at encoding
may only be a small aspect of the production effect, with other factors as contributors. The
between-subjects experiments showed a trend toward a production effect. If strength only plays
a small part in remembering, then it follows that there would be only a small effect when
strength is the only contributing factor to the production effect (Bodner & Taikh, 2012; Fawcett,
2013).
The experiments in this study tested the distinctiveness and strength accounts by manipulating
the proportion of produced words to unproduced words in the study phase. If the memory
benefit found in the production effect is attributable to distinctiveness, then the fewer the words
in the study list that are produced, the more those words would be remembered by participants
(e.g. if 80% of the words are produced and 20% of the words are silent, the silent words would
be more distinctive, and in turn, better remembered). If the memory benefit is due to strength,
then regardless of the proportion of words produced at study, participants would remember a
greater number of produced words. Production results in stronger memory traces at encoding,
allowing the produced words to be more easily remembered than non-produced words (Bodner

PRODUCTION EFFECT

6

& Taikh, 2012; Bodner et al., 2014; Fawcett, 2013; MacLeod et al., 2010; Ozubko & MacLeod,
2010). The benefits of production were also tested in a between-subjects design, where
participants either read all of the words silently (unproduced condition) or produced all of the
words.
When there are fewer words produced or unproduced, those words are more distinctive,
relative to the other words in the list. Since distinctiveness has been predominant in explanations
of the production effect in mixed-list, within-subjects designs, I predicted that when a
proportional manipulation of produced words was applied, there would be a memory advantage
for produced words in the 20% produced condition and unproduced words in the 80% produced
condition (Bodner & Taikh, 2012, Castel, Rhodes, & Friedman, 2013; Forrin et al., 2012;
MacLeod et al., 2010; Ozubko et al., 2012; Ozubko & MacLeod, 2010). I anticipated that in the
typical production effect paradigm where 50% of the words were produced, that produced words
would have a memory advantage over unproduced words. I also expected that there would be
either a small or no production effect in the between-subjects design, as evidenced in previous
experiments (Fawcett, 2013; Hopkins & Edwards, 1972). (See Table 1 for production conditions
and predictions).
Experiment 1
Method
Participants
Undergraduate students from Kwantlen Polytechnic University (KPU) Psychology classes
were recruited to participate in this experiment (N = 120). All five classes that participated were
first year Psychology classes. Participants received course credit as compensation for taking part
in the experiment.

PRODUCTION EFFECT

7

Materials
The study and test words were drawn from the pool of 120 words that were listed in the
appendix of MacDonald and MacLeod (1998) (see Appendix 1). This word list was used in
other experiments on the production effect with consistent results, reflecting its reliability
(Bodner & Taikh, 2012; Forrin et al., 2012; MacDonald & MacLeod, 1998; MacLeod et al.,
2010; Ozubko et al., 2012). The stimuli in the study phase consisted of 60 words randomly
selected from the item pool. The 60 words were randomly assigned to lists that varied in the
proportion of produced to unproduced words according to each of the five conditions as follows:
0 percent produced (0% produced words to 100% unproduced words), 20 percent produced
(20% produced words to 80% unproduced words), 50 percent produced (50% produced words to
50% unproduced words), 80 percent produced (80% produced words to 20% unproduced words),
and 100 percent produced (100% produced words to 0% unproduced words). Each list had the
words arranged in the same random fixed order. An icon accompanied each word to show the
participant how to read the word. The “speech bubble” icon ()) indicated that the participant
was to read the word aloud in a normal manner (produced condition). The “eyeglasses” icon
($) indicated that the participant was to read the word silently (unproduced condition). The
words were projected onto a classroom screen using Microsoft PowerPoint, presented in black 55
Calibri lowercase font against a white background. Each stimulus item during the study phase
stayed on the screen for 3000 ms and was offset from the next word by a 1000 ms blank screen.
Each word and its accompanying icon was presented in the middle of the screen.
Participants wrote the words for the free recall test on a blank piece of 8.5” by 11” paper. The
recognition test was composed of 120 words, with each word presented in black letters on a
piece of 8.5” by 11” piece of paper (Appendix 1). Sixty of the words were “old” and seen by the

PRODUCTION EFFECT

8

participants in the study phase, and 60 of the words were “new”, and had not been seen by the
participants during the experiment.
Procedure
Five groups of participants from first year Psychology classes completed the experiment in
classroom settings. Each class formed a separate group to study a different proportion of
produced to unproduced words. The groups were as follows: 0% produced words to 100%
unproduced words (n = 20); 20% produced words to 80% unproduced words (n = 25); 50%
produced words to 50% unproduced words (n = 19); 80% produced words to 20% unproduced
words (n = 27); 100% produced words to 0% unproduced words (n = 29).
Participants received verbal and written instructions to describe the procedure of the
experiment. The participants completed a practise phase of five slides that demonstrated the
study phase. The actual experiment commenced after it was determined that all the participants
understood the procedure. The study phase consisted of 60 words presented in a PowerPoint
presentation on a classroom screen. Each word in the 20% produced, 50% produced, and 80%
produced conditions was accompanied by an icon (either eyeglasses or speech bubble) that
indicated how the word was to be read (read silently or produced by reading aloud).
The study phase was followed by a free recall test and then a recognition test. For the recall
test, the participants were directed to write down all of the words that they remembered from the
study phase. The recognition test immediately followed the recall test. The recognition test
consisted of 120 words. Sixty words were “old” and had been seen by participants in the study
phase, and 60 were “new” and had not been seen by the participants in the study phase. The
participants were told that some of the words were “old” and had been presented during the study
phase and that some of the words were new. Participants were not told that the test contained all

PRODUCTION EFFECT

9

of the 60 study words. Participants were directed to circle all words that that they remembered
from the study phase.
Results
The proportional means of produced words remembered compared to unproduced words
remembered in each proportional condition were evaluated by conducting t-tests, calculating
average difference scores and 95% confidence intervals (CI) for each. The number of items
recalled or recognized by each participant was converted into proportions by dividing the
number recalled by the total number of items in that condition (e.g. In the 20% produced group,
the number of produced items recalled were divided by 12, the unproduced items were divided
by 48, and the errors were divided by 60.) The results from the recall test will be reported first,
followed by the results of the recognition test.
The data from the recall tests showed no differences in memory between produced and
unproduced words except in the 20% production condition (see Table 2). The 20% produced
group recalled more unproduced words than produced words, t(24) = -2.29, with a difference
score of M = -0.04, 95% CI [-0.08, 0.00]. The 50% produced group recalled the same proportion
of produced words and unproduced words, t(32) = 1.41, with a difference score of M = 0.02,
95% CI [-0.01, 0.05]. The 80% group recalled the same proportion of produced words as
unproduced words, t(26) = 1.39, with a difference score of M = 0.03, 95% CI [-0.01, 0.07]. The
100% produced group and the 0% produced group had unequal sample sizes, with the 100%
produced group having nine more participants. Only the data from 20 participants in the 100%
produced group were included in the analysis. These were the first 20 participants in the data set.
The 100% produced group recalled the same proportion of words as the 0% produced group,
t(38) = 2.33, with a difference score of M = 0.05, 95% CI [0.00, 0.11] (see Table 4).

PRODUCTION EFFECT

10

The recognition tests showed differences in the memory for produced words and unproduced
words across all proportional manipulations (see Table 3). As with the recall test, the 20%
produced group recognized fewer produced words than unproduced words t(24) = -3.74, with a
difference score of M = -0.11, 95% CI [-0.17, -0.05]. The 50% produced group recognized more
produced words than unproduced words, t(32) = 7.13, with a difference score of M = 0.16, 95%
CI [0.12, 0.21]. The 80% group recognized more produced words and unproduced words, t(26)
= 5.74, with a difference score of M = 0.14, 95% CI [0.09, 0.19]. As with the recall results, the
0% produced group and the 100% produced group had unequal sample sizes, with the 100%
produced group having nine more participants. Only the data from 20 participants in the 100%
produced group were included in the analysis. The 100% produced group recognized
significantly more words than the 0% produced group, t(38) = 3.83, with a difference score of M
= 0.195, 95% CI [0.07, 0.31] (see Table 4).
Discussion
Experiment 1 tested the production effect paradigm in a classroom setting. All published
production effect experiments have tested participants individually rather than in groups, except
in MacLeod (2011), where each participant was paired with one other individual. In MacLeod’s
experiment, similar to the typical production effect paradigm, there were items to be read silently
and items to be produced by reading aloud. The items were produced by either one individual or
the other, or by both participants. MacLeod (2011) found that participants remembered the items
that they had produced themselves best. However, there was still a memory advantage for items
produced at the same time as the other participant, and for items produced solely by the other
participant. The benefit was less in the latter two cases, but the production effect was still robust
(MacLeod, 2011). MacLeod’s experiment demonstrated that it was possible to generate a

PRODUCTION EFFECT

11

production effect by testing more than one individual at a time. Experiment 1 of the two
experiments described in this paper, investigated whether testing the production effect in a
classroom setting was a viable option. The mean proportion of words recognized in the 50%
production condition (the typical proportion produced in production effect studies) in the
recognition test was comparable to results found in other production effect studies, although the
proportion recalled was lower than what might normally be found (Forrin, Jonker, & MacLeod,
2013; MacLeod et al., 2010).
There were some questions about the findings of Experiment 1, particularly in the 20%
produced condition, where participants recalled more silent words than produced words. These
findings support neither the distinctiveness account nor the strength account. Distinctiveness
would predict that more produced words would be remembered, as fewer produced words would
stand out compared to the unproduced words. The strength account would have predicted that
producing the words would strengthen the memory traces at encoding, allowing more produced
words to be remembered. There may be another unknown variable contributing to the effect that
was found, or it may be a Type I error. Since all of the participants tested were in first year
Psychology classes, it is unlikely that this particular class of students was different from the other
classes that were tested. However, to ensure that there were no appreciable differences, it would
be necessary to replicate this condition.
The results of this experiment indicated that there were no significant differences in recall
between the silently read words and the produced (read aloud) words, except when 20% of the
words were produced and 80% were silent (see Table 2). Based on these results, distinctiveness
would not appear to explain the production effect. If distinctiveness accounted for the memory
advantage, then having fewer produced words would make them more, not less, memorable.

PRODUCTION EFFECT

12

They would be distinctive relative to the unproduced words, and would stand out in memory
(Bodner & Taikh, 2012; Conway & Gathercole, 1987; Dodson & Schacter, 2001; Forrin et al.,
2013; Forrin et al., 2012; MacLeod, 2011; MacLeod et al., 2010; Ozubko et al., 2012; Ozubko &
MacLeod, 2010; Ozubko et al., in press). Strength does not provide an explanation either.
According to the strength account, producing words provides an additional memory trace at
encoding (Bodner & Taikh, 2012; Bodner et al., 2014; Fawcett, 2013; MacLeod et al., 2010;
Ozubko & MacLeod, 2010). Therefore, produced words would be more easily recalled than
unproduced words, regardless of the proportion of produced words presented at study. If
strength at encoding allowed for enhanced memory, then the 20% produced group would show a
production effect, but this is not the case.
Analysis of the recognition data for Experiment 1 displayed significant differences between
produced and unproduced words in all the groups (see Table 3). Production of words provided
enhanced memory, except in the group where 20% of the words were produced. For the reasons
outlined above, neither distinctiveness nor strength appears to explain the results. The
recognition test data were confounded by the recall test that was administered immediately prior
to the recognition test. Therefore, it is important to evaluate the recognition results in light of
this confound.
When compared to reading silently, production by reading aloud provides an additional
auditory modality to provide distinctiveness (Conway & Gathercole, 1987; Forrin et al., 2012;
Hopkins & Edwards, 1972; Ozubko et al., in press). Writing also offers the additional mode of
orthographic processing to add distinctiveness (Conway & Gathercole, 1990; Forrin et al., 2012).
MacLeod (2011) noted that when both participants produced words by reading aloud at the same
time, the decreased benefit over solo reading could be due to a disruption in the distinctiveness at

PRODUCTION EFFECT

13

encoding. It is possible that the disruption described by MacLeod was amplified in the
classroom setting used in Experiment 1. Experiment 2 was designed as a replication of
Experiment 1 to address the problem of disruption, by changing the mode of production from
reading aloud to writing.
Experiment 2
Method
Participants
Undergraduate students from Kwantlen Polytechnic University (KPU) Psychology classes and
the Psychology Research Pool were recruited to participate in this experiment (N = 152). The
participants that comprised the 0% (n = 22), 50% (n = 34), and 100% (n = 28) produced
conditions were tested in a classroom setting, small groups, and individually. All but three of the
participants in the 100% produced condition were tested in a classroom setting. Participants in
the 20% (n = 34) and 80% (n = 34) produced conditions were tested in small groups or
individually. All of the classes used were first year Psychology classes, except those in the 0%
production condition, whose class was a second year Psychology class. To ensure that there
were no appreciable differences between participant classes, additional participants from a
second year Psychology class were tested in the 100% production condition. Participants
received course credit as compensation for taking part in the experiment.
Materials
The apparatus was identical to that used in Experiment 1 aside from three exceptions. The
stimuli were the same as Experiment 1 except that the assignment of “produced” or
“unproduced” words was counterbalanced to eliminate the potential confound of order effects
(Frensch, 1994). Writing was the mode of production rather than the reading aloud mode of

PRODUCTION EFFECT

14

production used in Experiment 1. Each participant was provided with pieces of paper measuring
4.75” by 5.25” on which to write each of the produced words. Each stimulus item during the
study phase stayed on the screen for 5000 ms and was offset from the next word by a 2000 ms
blank screen to allow participants sufficient time in which to write each word.
Procedure
The procedure was identical to that of Experiment 1, except that participants wrote the
produced words rather than read the produced words aloud. A pencil icon (!) indicated that the
participants were to write the word. Participants wrote each word to be produced on the paper
provided, immediately turning the paper over in front of them (see Conway and Gathercole,
1990; Forrin et al., 2012). Each paper used in the study phase was checked by the researcher to
ensure that each participant had sufficient time to write each word completely.
Results
The proportional means of produced words remembered compared to unproduced words
remembered in each proportional condition were evaluated by conducting t-tests, calculating
average difference scores and 95% confidence intervals (CI) for each. As in Experiment 1, the
number of items recalled or recognized by each participant was converted into proportions by
dividing the number recalled by the total number of items in that condition. The results from the
recall test will be reported first, followed by the results of the recognition test.
The data from the recall tests showed no differences in memory between produced and
unproduced words except in the 20% production condition (see Table 2). The 20% produced
group recalled more produced words than unproduced words, t(32) = 4.77, with a difference
score of M = 0.13, 95% CI [0.08, 0.19]. The 50% produced group recalled the same amount of
produced words and unproduced words, t(32) = 2.35, with a difference score of M = 0.07, 95%

PRODUCTION EFFECT

15

CI [0.01, 0.12]. The 80% group recalled the same amount of produced words as unproduced
words, t(32) = -1.61, with a difference score of M = -0.35, 95% CI [-0.09, 0.02]. The 100%
produced group and the 0% produced group had unequal sample sizes, with the 100% produced
group having nine more participants. Only the data from the first 22 participants in the data set
from the 100% produced group were included in the analysis. The 100% produced group
recalled the same amount of words as the 0% produced group, t(41) = -2.65, with a difference
score of M = -0.09, 95% CI [-0.16, -0.02] (see Table 4).
The recognition tests showed significant differences in the memory for produced words and
unproduced words across all proportional manipulations in the within-subjects groups (see
Tables 3 and 4). The 20% produced group recognized more produced words than unproduced
words t(33) = 4.81 with a difference score of M = 0.13, 95% CI [-0.16, -0.02]. The 50%
produced group recognized more produced words than unproduced words t(33) = 3.32, with a
difference score of M = 0.12, 95% CI [0.05, 0.19]. The 80% produced group recognized more
produced words than unproduced words t(33) = 3.32, with a difference score of M = 0.09, 95%
CI [0.04, 0.15]. The 100% produced group was essentially the same as the 0% produced group
in recognition of produced versus unproduced words t(38) = 0.95, with a difference score of M =
-0.05, 95% CI [-0.15, 0.06].
Discussion
A production effect was observed in all the within-subject groups, with significant differences
between the produced words and unproduced words recalled. The only exception was the 80%
produced group in the recall test, where there was essentially no difference between the
proportional mean of the produced words recalled and the unproduced words recalled.

PRODUCTION EFFECT

16

The between-subjects group that produced all of the words by writing recalled proportionally
fewer produced words than the group that read all of the words silently (unproduced words).
This result, where more unproduced words were recalled than produced words in a recall test, to
our knowledge, has not been reported previously. It may be a feature that arises when testing
written production with recall tests. In previous studies, written production was tested with
recognition tests only (Conway and Gathercole, 1990; Forrin et al., 2012).
There was a possible confound in the between-subjects groups. Twenty-five of the 28
participants in the 100% produced condition were from a first year Psychology course while 14
of the 22 participants in the 0% produced condition were from a second year Psychology course.
To address this possible confound, another second year Psychology class (n = 19) was tested in
the 100% produced condition. Their data was essentially the same as the original group. Thus,
the finding that more unproduced words were recalled than produced words in our betweensubjects comparison is not due to year 1 versus year 2 students comprising the 100% and 0%
produced conditions, respectively.
General Discussion
This experiment set out to test the distinctiveness account and the strength account as
explanations for the production effect, by manipulating the proportion of produced words
relative to unproduced words. The data from the recall tests in Experiment 1 provided no
evidence to support the distinctiveness account in any of the production conditions. In fact,
Experiment 1 did not provide evidence for either distinctiveness or strength in any of the
production conditions. Only in the between-subjects design, where a group that studied all
produced words was compared to a group that studied all unproduced words, was there support
for either of the accounts. In this comparison, participants recalled more produced words than

PRODUCTION EFFECT

17

unproduced words, supporting the strength account, which posits that production will result in a
memory advantage.
Experiment 1 used reading aloud as the mode of production, the same paradigm as a majority
of other production effect experiments (for example, MacLeod et al., 2010; Ozubko & MacLeod,
2010). Other production effect experiments used individual testing, whereas a classroom setting
was used in Experiment 1. It is possible that self-consciousness or embarrassment moderated the
responses of some of the participants. In addition, it was difficult to monitor whether
participants were actually reading aloud.
Experiment 2 provided evidence to support distinctiveness in the 20% produced and 50%
produced conditions, with more produced words recalled than unproduced words.
Distinctiveness could explain the production effect in the 20% produced and 50% production
conditions, since production has made the words in those conditions stand out, relative to the
unproduced words (Bodner et al., 2014; Conway & Gathercole, 1987; Lin & MacLeod, 2012). If
distinctiveness provided the explanation for this effect, then the 80% produced condition should
have yielded more unproduced words recalled and recognized than produced words, since the
smaller proportion of unproduced words made them more distinctive. However, this was not the
case. Because more produced words were recalled than unproduced words, evidence for the
strength account surfaces. The strength account posits that production increases the memory
traces at encoding, aiding memory (Bodner & Taikh, 2012; Bodner et al., 2014; Fawcett, 2013;
MacLeod et al., 2010; Ozubko & MacLeod, 2010). Strength could also be offered as an
explanation for the results found in the 20% produced and 50% produced conditions in
Experiment 2, since strength posits that production increases the memory traces at encoding to
provide a memory advantage.

PRODUCTION EFFECT

18

Transfer-appropriate processing might explain the production effect when writing is the mode
of production. This theory suggests that there is a memory advantage when the study and test
modalities match (Graf & Ryan, 1990). In Experiment 2, the match of production by writing and
the subsequent written recall test might explain the production effect when production conditions
were proportionally manipulated. However, transfer-appropriate processing does not explain
why more unproduced words were recalled by the 0% produced condition than produced words
recalled by the 100% produced condition, in Experiment 2. According to this theory, the match
of study and test modality should result in more produced words recalled in the 100% produced
condition.
False alarms in Experiment 2 were consistently higher than in Experiment 1 across all
production conditions. Production by writing may lead participants to feel more confident in
their recognition judgments than those participants who produce by reading aloud. A follow-up
study might include confidence judgments by participants to test this assumption.
The between-subjects design in Experiment 2 yielded results that have not yet been seen in
the published literature. More unproduced than produced words were recalled when comparing
the groups of 100% produced versus 0% produced. It is possible that the 0% produced group
used the time given to them to view the words, to employ study strategies that provided a
memory advantage over the group that produced all the words (G. E. Bodner, personal
communication, March 26, 2014). Some participants may have grouped words that had a
common theme (e.g. vacation, resort, ticket, travel). The 100% produced group participants
were busy writing each word and turning over each paper, so did not have the same amount of
time to visually encode the words, putting this group at a disadvantage. This effect may not have
been seen in the proportionally produced conditions because the produced words would interrupt

PRODUCTION EFFECT

19

the study strategies. The mixture of produced and unproduced words would make it difficult to
consciously adopt a strategy that could persist throughout the entire study phase.
Replications of these experiments will be necessary. It may be found that both the
distinctiveness and strength accounts contribute to the production effect, with an additional
contribution from another, yet unknown source. Understanding the underlying mechanisms of
the production effect could result in more effective strategies for studying (Ozubko, Hourihan, &
MacLeod, 2012) and memory improvement in older adults (Lin & MacLeod, 2012).

PRODUCTION EFFECT

20
References

Bodner, G. E. & Taikh, A. (2012). Reassessing the basis of the production effect in
memory. Journal of Experimental Psychology: Learning, Memory, and Cognition. Advance
online publication. doi: 10.1037/a0028466
Bodner, G. E. Taikh, A., & Fawcett, J. M. (2014). Assessing the costs and benefits of production
in recognition. Psychonomic Bulletin & Review, 21(1), 149-154. doi: 10.3758/s13423-0130485-1
Castel, A. D., Rhodes, M. G., & Friedman, M. C. (2013). Predicting memory benefits in the
production effect: the use and misuse of self-generated distinctive cues when making
judgments of learning. Memory and Cognition, 41, 28-35. doi: 10.3758/s13421-012-0249-6
Conway, M. A., & Gathercole, S. E. (1987). Modality and long-term memory. Journal of
Memory and Language, 26(3), 341-361. doi: 10.1016/0749-596X(87)90118-5
Dodson, C. S., & Schacter, D. L. (2001). “If I had said it I would have remembered it”:
Reducing false memories with a distinctiveness heuristic. Psychonomic Bulletin & Review,
8(1), 155-161. doi: 10.3758/BF03196152
Fawcett, J. M. (2013). The production effect benefits performance in between-subject designs:
A meta-analysis. Acta Psychologica, 142, 1-5. doi: 10.1016/j.actpsy.2012.10.001
Forrin, N. D., Jonker, T. R., & MacLeod, C. M. (2013). Production improves memory
equivalently following elaborative vs non-elaborative processing. Memory, doi:
10.1080/09658211.2013.798417
Forrin, N. D., MacLeod, C. M., & Ozubko, J. D. (2012). Widening the boundaries of the
production effect. Memory & Cognition, 40, 1046-1055. doi: 10.3758/s13421-012-0210-8

PRODUCTION EFFECT

21

Frensch, P.A. (1994). Composition during serial learning: a serial position effect. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 20(2), 423-443.
Gathercole, S. E., & Conway, M. A. (1988). Exploring long-term modality effects:
Vocalization leads to best retention. Memory & Cognition, 16(2), 110-119. doi:
10.3758/BF03213478
Graf, P. & Ryan, L. (1990). Transfer-appropriate processing for implicit and explicit memory.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(6), 978-992.
doi: 10.1037/0278-7393.16.6.978
Hopkins, R. H. & Edwards, R. E. (1972). Pronunciation effects in recognition memory.
Journal of Verbal Learning and Verbal Behavior, 11, 534-537. doi: 10.1016/s00225371(72)80036-7
Lin, O. H., & MacLeod, C. M. (2012). Aging and the production effect: A test of the
distinctiveness account. Canadian Journal of Experimental Psychology/Revue Canadienne
de Psychologie Expérimentale, 66(3), 212-216. doi: 10.1037/a0028309
MacDonald, P. A. & MacLeod, C. M. (1998). The influence of attention at encoding on direct
and indirect remembering. Acta Psychologica, 98, 291-310. doi: 10.1016/s00016918(97)00047-4
MacLeod, C. M. (2011). I said, you said: The production effect gets personal. Psychonomic
Bulletin & Review, 18(6), 1197-1202. doi: 10.3758/s13423-011-0168-8
MacLeod, C. M., Gopie, N., Hourihan, K. L., Neary, K. R., & Ozubko, J. D. (2010). The
production effect: Delineation of a phenomenon. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 36(3), 671-685. doi: 10.1037/a0018785

PRODUCTION EFFECT

22

Ozubko, J. D., Gopie, N., & MacLeod, C. M. (2012). Production benefits both recollection and
familiarity. Memory & Cognition, 40, 326-338. doi: 10.3758/s13421-011-0165-120
Ozubko, J. D., Hourihan, K. L., & MacLeod, C. M. (2012). Production benefits learning: The
production effect endures and improves memory for text. Memory, 20(7), 717-727. doi:
10.1080/09658211.2012.699070
Ozubko, J. D. & MacLeod, C. M. (2010). The production effect in memory: Evidence that
distinctiveness underlies the benefit. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 36(6), 1543-1547. doi: 10.1037/a0020604
Ozubko, J. D., Major, J., & MacLeod, C. M. (in press). Remembered study mode: Support for
the distinctiveness account of the production effect. Memory.
Quinlan, C. K. & Taylor, T. L. (2013). Enhancing the production effect in memory. Memory,
21(8), 904-915. doi: 10.1080/09658211.2013.766754

PRODUCTION EFFECT

23

Table 1.
Production Proportion Conditions and Predictions for Distinctiveness and Strength Accounts
Production Condition

Experimental
Design

Prediction if
Distinctiveness
Account

Prediction if
Strength Account

20% produced

Within-subjects

Produced words >
unproduced words

Produced words >
unproduced words

50% produced

Within-subjects

Produced words >
unproduced words

Produced words >
unproduced words

80% produced

Within-subjects

Produced words >
unproduced words

Produced words >
unproduced words

100% produced versus
0% produced

Between-subjects

Produced words =
unproduced words

Produced words >
unproduced words

Table 2.
Experiments 1 and 2: Mean Proportions of Words Remembered in Recall Test
Proportion

Produced

Unproduced

Errors

Exp 1. 20% produced

0.07
(0.02)

0.11
(0.01)

0.01
(0.00)

Exp 1. 50% produced

0.14
(0.01)

0.12
(0.01)

0.02
(0.00)

Exp 1. 80% produced

0.13
(0.01)

0.10
(0.02)

0.02
(0.00)

Exp 2 20% produced

0.26
(0.02)

0.13
(0.01)

0.02
(0.00)

Exp 2. 50% produced

0.16
(0.02)

0.09
(0.02)

0.02
(0.00)

Exp 2. 80% produced

0.14
0.17
0.03
(0.02)
(0.03)
(0.00)
Note. Standard errors are shown in parentheses below each corresponding mean.

PRODUCTION EFFECT

24

Table 3.
Experiments 1 and 2: Mean Proportions of Words Remembered in Recognition Test
Proportion

Produced

Unproduced

False Alarms

Exp 1. 20% produced

0.23
(0.04)

0.34
(0.05)

0.03
(0.01)

Exp 1. 50% produced

0.52
(0.03)

0.35
(0.03)

0.03
(0.01)

Exp 1. 80% produced

0.40
(0.04)

0.26
(0.03)

0.04
(0.01)

Exp 2 20% produced

0.54
(0.03)

0.41
(0.02)

0.07
(0.01)

Exp 2. 50% produced

0.48
(0.04)

0.36
(0.04)

0.12
(0.02)

Exp 2. 80% produced

0.52
0.43
0.07
(0.04)
(0.03)
(0.01)
Note. Standard errors are shown in parentheses below each corresponding mean.
Table 4
Experiment 1 and 2: Mean Proportions of Words Remembered in Between-Subjects Conditions
Experiment/Test

Produced

Errors/False
Alarms

Unproduced

Errors/False
Alarms

Exp 1/Recall

0.17
(0.02)

0.04
(0.01)

0.12
(0.01)

0.02
(0.01)

Exp 1/Recognition

0.53
(0.03)

0.06
(0.01)

0.33
(0.04)

0.04
(0.01)

Exp 2/Recall

0.19
(0.02)

0.02
(0.00)

0.28
(0.03)

0.03
(0.01)

Exp 2/Recognition

0.58
0.05
0.63
(0.04)
(0.01)
(0.03)
Note. Standard errors are shown in parentheses below each corresponding mean

0.10
(0.02)

PRODUCTION EFFECT

25

Appendix 1: Words to be read silently or produced
forest

pocket

branch

traffic

machine

leather

lesson

invention station

education history

village

theatre

wagon

minute

factory

direction

century

amount

record

debate

furniture

wheel

address

judge

ticket

account

powder

uniform

teacher

answer

package

quarrel

victory

captain

trousers

shoulder

afternoon election

ocean

resort

laugh

market

capital

industry

entrance

school

dinner

vacation

clothes

partner

merchant

foundation stream

garden

kettle

winter

glass

beauty

queen

avenue

evening

language

painting

gravity

friend

engine

basket

treasure

office

plate

campaign

pebble

speech

battery

thread

distance

summer

knock

valley

invitation guardian

attitude

wheat

whisper

reward

handle

daughter

building

steam

neighbor

travel

attention

peace

harbor

author

kingdom

river

uncle

meadow

nephew

message

arrow

holiday

fashion

turnip

department island

journey

ladder

sailor

shadow

porch

castle

kitchen

orchard

border

quarter

justice

envelope