The Production Effect: Delineation of a Phenomenon
Colin M. MacLeod, Nigel Gopie, Kathleen L. Hourihan, Karen R. Neary, and Jason D. Ozubko
University of Waterloo
In 8 recognition experiments, we investigated the production effect—the fact that producing a word aloud
during study, relative to simply reading a word silently, improves explicit memory. Experiments 1, 2, and
3 showed the effect to be restricted to within-subject, mixed-list designs in which some individual words
are spoken aloud at study. Because the effect was not evident when the same repeated manual or vocal
overt response was made to some words (Experiment 4), producing a subset of studied words appears to
provide additional unique and discriminative information for those words—they become distinctive. This
interpretation is supported by observing a production effect in Experiment 5, in which some words were
mouthed (i.e., articulated without speaking); in Experiment 6, in which the materials were pronounceable
nonwords; and even in Experiment 7, in which the already robust generation effect was incremented by
production. Experiment 8 incorporated a semantic judgment and showed that the production effect was
not due to “lazy reading” of the words studied silently. The distinctiveness that accrues to the records of
produced items at the time of study is useful at the time of test for discriminating these produced items
from other items. The production effect represents a simple but quite powerful mechanism for improving
memory for selected information.
Keywords: memory, encoding, production, generation, distinctiveness
In 1978 Slamecka and Graf reported a thorough set of experi-
ments demonstrating that producing a word from a cue (e.g.,
producing fast from the cue rapid-f) leads to considerably better
memory for that word than does simply reading the word. The
phenomenon that they called the “generation effect” has subse-
quently become one of the most widely used manipulations in
memory research, leading to their article becoming a citation
classic (see Slamecka, 1992), now having been cited more than
500 times and having spawned hundreds of directly related articles
(for a meta-analysis, see Bertsch, Pesta, Wiscott, & McDaniel,
2007).
The generation effect is a member of a very select club: manip-
ulations that have consistent, reliable effects on retention of a
stimulus that was presented only once. Other such manipulations
include imagery (Paivio, 1971) and elaboration (Craik & Lockhart,
1972). All three of these involve some recoding of the stimulus,
and all produce quite substantial memory benefits, often an im-
provement of 10% or even more relative to a “standard” baseline
of simply reading the word. Enriching the encoding of the stimulus
definitely enhances memory for it, in line with what we know
generally about mnemonic techniques (see Higbee, 1988).
Improving Memory by Saying a Word Aloud
In the same period, another factor was first reported to benefit
memory but, unlike the others, it failed to attract subsequent
research attention. In the study in question, Hopkins and Edwards
(1972) tested a key assumption of frequency theory (Ekstrand,
Wallace, & Underwood, 1966)—that recognition should be better
for pronounced than for unpronounced words because pronuncia-
tion would increment the item’s frequency (see Hopkins, Boylan,
& Lincoln, 1972, for evidence that pronunciation does increase
judged frequency). To test this prediction, Hopkins and Edwards
used two recognition tests: two-alternative forced choice (Exper-
iment 1) and yes/no (Experiment 2). In both experiments, three
groups of subjects studied 100-word lists. There were two pure-list
groups— one read all 100 words aloud, and one read all 100 words
silently—and one mixed-list group, which read 50 of the words
aloud and 50 silently.
In comparing the two pure-list conditions, Hopkins and Edwards
(1972) found no between-subjects benefit to reading words aloud.
But in the mixed-list condition, words read aloud were better
recognized than those read silently. This pattern held for both types
of recognition test, with the within-subject benefit of reading aloud
being approximately 10%. Hopkins and Edwards suggested that
the effect was at encoding.
Colin M. MacLeod, Nigel Gopie, Kathleen L. Hourihan, Karen R.
Neary, and Jason D. Ozubko, University of Waterloo, Waterloo, Canada.
Nigel Gopie is now at the Rotman Research Institute, Toronto, Canada.
Kathleen L. Hourihan is now at the University of Illinois at Urbana-
Champaign.
This research was supported by Natural Sciences and Engineering
Research Council of Canada Discovery Grant A7459. This article is
dedicated to the memory of Norman J. Slamecka, longtime colleague of
Colin M. MacLeod at the University of Toronto. In tribute, the title of this
article is a (slight) variation on that of the original article on the generation
effect (Slamecka & Graf, 1978); we thank Peter Graf for his permission.
We thank Keehan Bailey, Emily Bryntwick, Samantha Chaves, Domenica
De Pasquale, Nicole Deckert, Grace Hsiao, Jingjing Ji, Marita Partanen,
and Molly Pottruff for assistance with collecting the data and Amy Criss,
Neil Mulligan, and Hubert Zimmer for their advice concerning the original
version of this article.
Correspondence concerning this article should be addressed to Colin M.
MacLeod, Department of Psychology, University of Waterloo, 200 Uni-
versity Avenue West, Waterloo, Ontario N2L 3G1, Canada. E-mail:
cmacleod@uwaterloo.ca
Journal of Experimental Psychology: © 2010 American Psychological Association
Learning, Memory, and Cognition
2010, Vol. 36, No. 3, 671– 685
0278-7393/10/$12.00 DOI: 10.1037/a0018785
671