SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 1

Spaced Retrieval Improves Retention and Transfer of Foreign Vocabulary

Durna Alakbarova & B. Hunter Ball

Department of Psychology, University of Texas at Arlington

Author Note

Address correspondence to Hunter Ball, Department of Psychology, University of Texas

at Arlington, Arlington, TX 76019. Email: Hunter.Ball@uta.edu

Durna Alakbarova (https://orcid.org/0000-0002-9938-3851) & B. Hunter Ball, Department of

Psychology, University of Texas at Arlington, 501 Nedderman Drive, Arlington TX, 76109.

This data was collected as a Master’s Thesis at the University of Texas at Arlingoton. The data

for this study can be accessed at: https://osf.io/zst59/?view_only=4a7ff44f75ea4ce6b0c62ac8446ecbf2

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 2

Abstract

One strategy that has been shown to promote long-term retention is spaced retrieval practice. In this study,

we used standard and dropout procedures of spaced retrieval training to examine whether they produce

superior retention and transfer relative to repeated restudy in novel vocabulary learning. Across three

sessions spaced two days apart, participants learned 30 Swahili words with their English definitions via

spaced testing or by restudying. Definitions were the same (fixed) across days in Experiment 1 or different

each day (variable) in Experiment 2. A week later, all participants were tested using either the same

(repetition) or new (transfer) definitions from study. It was found that standard spaced retrieval practice

improved recall of repetition and transfer items compared to repeated restudy with both fixed and variable

learning. We also found that although the dropout procedure takes less time, standard spaced retrieval

results in more transfer. These findings have important implications for choosing learning techniques to

achieve the best outcomes.

Keywords: testing effect, spaced practice, learning, transfer

Public Significance: Previous research suggests that repeatedely retrieving information (e.g., quizzing
oneself) is a more effective study habit than simply rereading material. The results of the current study
showed spaced retrieval of Swahili words from English definition not only improves long-term retention
of learned information, but it also transfers to improved learning using novel definitions that were never
studied. However, the efficacy of retention versus transfer depends on the type of retrieval procedure
used. Although removing successfully recalled items from the study list across learning episodes (i.e.,
dropout procedure) is a time-efficient procedure that improve retention, multiple retrievals of the same
information (i.e., standard procedure) is required to transfer to novel definitions.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 3

Spaced Retrieval Improves Retention and Transfer of Foreign Vocabulary

One of the most common study techniques employed by students is restudying material (Karpicke

et al., 2009; Hartwig & Dunlosky, 2012). However, studies have shown that this technique is not the most

effective at promoting long-term retention of learned material (Rivers, 2021). In contrast, retrieval

practice has been shown to be a robust learning procedure (e.g., Butler et al., 2017; Roediger & Karpicke,

2006a; Agarwal et al., 2012). In a typical experiment on retrieval practice, participants first engage in a

learning phase, wherein they study the material (e.g., the English-Swahili word pair “baby – mtoto”).

Then, participants complete a relearning phase. In this phase, half of participants attempt to retrieve as

much of the material as they can (e.g., baby – ???), while the other half repeatedly restudy the information

(e.g., baby-mtoto). Either immediately or after a delay, all participants take a final test, the performance

on which is used as an indication of learning. Studies using variations of this procedure show that

engaging in retrieval by testing oneself during learning produces better recall than simply rereading the

material. The memorial benefit of retrieval over restudying has been shown in a variety of materials and

with different test types, and to be even more robust when used with corrective feedback (for a review see

Rowland, 2014). Additionally, studies have found that this testing effect persists after a week in

laboratory setting (Roediger & Karpicke, 2006b) and on semester and year-end exams in classrooms

(Roediger et al., 2011; McDaniel et al., 2011). Moreover, retrieval has also been shown to promote the

ability to apply knowledge to new tasks (i.e., transfer) better than restudy does (e.g., Butler, 2010;

Carpenter, 2012; Butler et al., 2017; Pan & Rickard, 2018a). The present study is designed to evaluate the

benefit of two spaced retrieval procedures over restudying on retention and transfer of knowledge in

novel vocabulary learning.

Several theories have been developed to explain the underlying mechanisms of the testing effect.

One such explanation is the elaborative retrieval hypothesis, which states that searching the semantic

network when attempting to retrieve a correct answers on a test activates information relevant to the

target, thereby creating a more elaborate memory trace for it (Carpenter, 2009; Carpenter & DeLosh,

2006). Some of this relevant information could then be reactivated by the cue during the final test and aid
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 4

in recalling the target. Restudying the material does not prompt a memory search, and therefore does not

lead to a similarly elaborate memory for the target. The elaborative retrieval hypothesis has been

supported by the findings that longer lag between retrieval attempts during study results in higher recall

than shorter lag (Rawson et al., 2015). Longer lag results in more forgetting than shorter lag and requires

a more extensive memory search for the target. Higher recall rates with longer lag therefore show that a

more extensive memory search leads to a more elaborated memory trace.

Another explanation of the testing effect is the retrieval effort hypothesis proposed by Pyc and

Rawson (2009). This theory states that conditions that make retrieval more difficult during the relearning

phase result in a better memory on the final test. Presumably, more difficult retrieval slows forgetting of

information and creates a more elaborate memory for the target via a more extensive memory search. Pyc

and Rawson (2009) provided support for this hypothesis with findings that increasing the number of

stimuli between retrieval of each target – which allowed for more forgetting and resulted in more difficult

retrieval – increased final recall. Additionally, they found that increasing the criterion level (the number

of times the target must be correctly recalled) on the relearning test led to diminishing returns on the final

test. Higher criterion levels lead to better memory – and therefore easier retrieval – with each recall

attempt. The finding that higher criterion (and by proxy, easier retrieval) leads to diminishing returns on

the final test therefore supports the retrieval effort hypothesis.

To explain the testing effect in cued-recall tests, where participants study paired associates and

then have to retrieve the target when presented with the cue at test, Pyc and Rawson (2010) developed the

mediator effectiveness hypothesis. According to his hypothesis, when learning cue-target pairs,

participants generate a keyword to tie the cue to the target in their memory (Carpenter, 2011). When the

cue is then presented during the test, it presumably activates the mediator, which can aid in retrieval of the

target information. Pyc and Rawson (2010) found that compared to restudying, retrieval practice

improved memory for the mediators and the strength of the association between the mediator and the

target. However, follow-up studies failed to find the relationship between mediator and target retrieval, or
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 5

the increased probability of mediator generation with testing (Cho et al., 2017; Lehman & Karpicke,

2016).

Spaced Retrieval and Long-Term Retention

While a single retrieval has been shown to benefit future recall over restudy, spaced retrieval

improves memory even further (Roediger & Karpicke, 2006b). Spaced retrieval refers to a learning

technique wherein participants retrieve information several times before the final criterion test, with some

time between each relearning session. Three mechanisms have been used to explain the benefit of spaced

retrieval for long-term memory. First, spaced retrieval includes spaced practice. Spacing learning over

several sessions has been shown to produce better memory than completing all practice in one sitting

(Cull, 2000; Pavlik & Anderson, 2005). When learning is done in different temporal contexts, each of

those contexts gets associated with the target and creates a new retrieval route to that information. An

alternative explanation of this mechanism is that unlike during massed retrieval practice, attempting to

retrieve information during spaced practice requires reconstruction of that information and pathways

leading to it from the previous study session. This results in a more meaningful organization of

information, which facilitates its accessibility (Furst, 2020). Additionally, spaced retrieval introduces

variability into the relearning sessions. This variable encoding links the target to more contextual

information (not necessarily temporal), creating more retrieval routes and thereby increasing the

probability of recall. Encoding variability can be introduced in many ways, such as practicing in different

places (e.g., bedroom vs classroom) or with different materials (e.g., learning a concept by reading about

it vs by looking at a diagram). The last mechanism underlying the benefit of spaced retrieval is explained

by the mediator shift hypothesis (Pyc & Rawson, 2012). According to this hypothesis, repeatedly

retrieving information leads participants to realize that some of their mediators are not effective for recall,

which causes them to shift to better mediators. Pyc and Rawson (2012) found that participants who

engaged in spaced retrieval at relearning shifted mediators more frequently and had higher recall on the

final test than those who restudied material repeatedly. Moreover, Dikmans et al. (2020) found that
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 6

repeated retrieval led participants to shift from mediated to direct retrieval, and this shift predicted better

recall on the final test.

One specific form of spaced retrieval that has shown to produce strong memorial benefit is a

dropout procedure (Bahrick, 1979). In a typical experiment using this paradigm, after initially studying

the material, participants engage in retrieval to criterion with dropout. During this retrieval procedure,

targets that are recalled to criterion (e.g., once) are dropped from the list of test items, while unrecalled

targets (or those that have not yet reached criterion) are moved to the end of the testing list for another

retrieval attempt. Participants then return after several days for another retrieval session with the same

procedure. After several of such relearning sessions, participants take the final test. The dropout

procedure has been shown to be more beneficial than restudy and to produce high levels of retention both

in laboratory and in naturalistic settings (Rawson et al., 2013; Rawson et al., 2018). The dropout

procedure benefits memory by engaging learners in elaborative retrieval, and through multiple relearning

episodes produces variability in encoding that results in multiple retrieval routes to the learned

information (Rawson et al., 2013).

The main difference between the standard spaced retrieval procedure and the dropout procedure

just described is that the former does not require retrieval for all items to be successful. In contrast, the

dropout procedure requires successful retrieval of each item once. Thus, while both learning techniques

reflect spaced retrieval, learning to a criterion of one may be more efficient in terms of time. Although we

compare the two conditions in this study, our primary interest lies in the difference in transfer between

either of these spaced retrieval conditions and restudying.

Retrieval Practice and Transfer

Perhaps one of the most interesting findings from the retrieval practice literature is that testing not

only benefits memory for previously studied information, but also learning of new information (i.e.,

transfer; Pan & Rickard, 2018b). Transfer refers to successful application of acquired skills or knowledge

to new tasks that require similar skills or knowledge. For example, if participants learn that the Swahili

word “mtoto” refers to “a very young child”, they could use transfer to deduce that the Swahili word for
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 7

“a recently born human” would be “mtoto”. Facilitation of transfer is of great interest to researchers in the

field of learning as it is a skill needed in many areas of everyday life. For example, schoolchildren must

rely on concepts learned in previous years (e.g., multiplication and division) in order to master new

material (e.g., triangle area calculations) and adults must apply their skills from a previous position at

work to a new position to which they are promoted.

Butler (2010) examined whether the memorial benefit of retrieval can promote transfer. After

studying a series of passages, participants in his study engaged in four relearning blocks for the passages.

Some passages were presented for restudy, while others were tested using questions for the information

from the passages. For example, the former group reread a passage containing the concept, “A bird’s

wing has fairly rigid bone structure that is efficient at providing lift, whereas a bat has a much more

flexible wing structure that allows for greater maneuverability”, and the latter group had to answer a

question, “A bat has a very different wing structure from a bird. What is the wing structure of a bat like

relative to that of a bird?”. A week later, all participants took a final test which consisted of questions on

information from the passages (referred to as repetition items) and inferential questions for which the

answers must be deduced (referred to as transfer items). For repetition items, participants saw the same

questions as those asked during relearning. Transfer items involved applying information learned in the

passages to a new domain. An example of a transfer counterpart to the above example was, “The U.S.

Military is looking at bat wings for inspiration in developing a new type of aircraft. How would this new

type of aircraft differ from traditional aircrafts like fighter jets?”. The answer to this question would be

that aircrafts modeled after bird wings would provide better lift, whereas those modeled after bat wings

would provide better maneuverability. The results showed that retrieval practice not only enhanced

memory for repetition items relative to restudy, but also led to superior memory for transfer items. Butler

posited that testing promotes transfer because retrieving information not only improves its storage

strength, but also results in encoding of additional related information, thereby elaborating the memory

trace and potentially creating more retrieval routes. He argued that these mechanisms allow for better
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 8

understanding and ability to apply information to new contexts, thereby increasing the probability that

transfer will occur.

Studies have also found that testing facilitates several types of transfer, such as transfer across

temporal contexts (e.g., final test after a week; Coppens et al., 2011; Carpenter et al., 2008; Carpenter et

al., 2009), test formats (e.g., initial free-recall tests and final multiple-choice tests; Carpenter & DeLosh,

2006; Kang et al., 2007; McDaniel et al., 2007), and knowledge domains (e.g., previously studied but

untested items; Jacoby et al., 2010; Kang et al., 2011; Chan et al., 2006; for a review see Carpenter,

2012). Other studies have also showed transfer to new related cues. For example, Carpenter (2011)

showed that retrieval practice for the pair “mother-child” resulted in higher correct recall of the target

“child” when cued with the new related cue “birth” than restudying the pair. In the current study, we

examine transfer in a similar way: after studying and relearning Swahili words (e.g., mtoto) with one

English definition (e.g., a very young child), participants are asked to retrieve the Swahili word (mtoto) in

response to a new related cue (e.g., a recently born human) on the final test. Previous research has found

that transfer is less likely to occur with term-definition pairs than with term-term pairs, presumably

because definitions contain more words than single terms, which do not have prior strong associations to

the target or a chunked memory representation (Pan & Rickard, 2017). This might not be the case in the

current study, as the English definitions used are not new concepts to participants, and have

representations of commonplace English words (i.e., are already chinked into a memory representation,

such as “baby”).

Novel related cues have not only been investigated as targets of transfer, but also as potential

mechanisms of boosting transfer. For example, Butler (2010) examined whether relearning with different

related cues can boost transfer to new inferential questions. During each relearning phase, participants in

the testing condition were asked rephrased versions of each question. Surprisingly, Butler found that

encoding variability did not boost transfer. However, that might have been due to rephrased questions not

offering enough variability during learning. In fact, in a later study, Butler found that introducing

variability using application questions that require different answers based on the same concept helped
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 9

participants answer new application questions on the final test (Butler et al., 2017). Similarly, Goode et al.

(2008) found that practice solving variations of an anagram resulted in better transfer than solving the

same anagram several times. The current study similarly addresses the role of encoding variability on

transfer of new knowledge, but instead using novel term-definition pairs.

A recent meta-analysis provides a mechanistic account for how retrieval practice might benefit

transfer, referred to as the three-factor framework for transfer of test-enhanced learning (Pan & Ricard,

2018). According to the framework, for retrieval to facilitate transfer, three factors must be present:

response congruency, elaborated retrieval practice, and initial test performance. Response congruency

refers to the match between responses required on the relearning and final tests. For example, if the cue of

“a very young child” on the relearning test requires a response “mtoto” and the cue of “a human

offspring” on the final test requires the response “mtoto”, response congruency is present, and transfer is

more likely to occur. When response congruency is present, the response required on the final test is made

more accessible by the relearning test (via successful retrieval or feedback), which increases the

probability of successful transfer.

The second factor is referred to as elaborated retrieval practice and can take two distinct forms:

broad encoding methods and elaborative feedback. Broad encoding methods are present in initial tests

that are formatted in such a way as to elicit processing of additional information related to the target. For

example, an initial test may ask participants to report a detailed reason for their response or to attempt to

recall everything they can that is related to the target. This type of retrieval results in more elaborate

cognitive processing, which can be driven by reactivation of memories formed during initial study,

improved discrimination between answer choices (e.g., in a multiple-choice test), and/or better

organization of the information in memory. Elaborative feedback refers to providing participants with

detailed feedback on the initial tests (rather than simple corrective feedback). This can include providing

participants with the reason a particular answer is correct, explaining the underlying concept of the correct

answer, or allowing an opportunity to restudy all previously studied material. Providing elaborative
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 10

feedback produces improved postretrieval restudy of the material, which can result in a more elaborated

memory trace for the target.

The third factor – initial test performance – refers to the findings that higher performance on

initial test during relearning predicts the magnitude of transfer. Successful retrieval during relearning is

more likely to result in retrieval of aspects of the original study session other than the target itself (e.g.,

thoughts, inferences) than unsuccessful retrieval. These aspects, in turn, create a more elaborated memory

for the successfully retrieved targets and can serve as additional cues for successful retrieval on the

transfer test. Participants with higher initial recall then are more likely to experience transfer for more

items during the final test than participants with low initial recall.

This three-factor framework accounted for most of the findings in the literature on retrieval-

facilitated transfer. Furthermore, while each factor alone increased probability of transfer, including all

three factors had an additive effect on transfer. Thus, the framework can serve as a basis for evaluating

transfer in different learning techniques. For example, although there is growing literature on the effects

of spaced retrieval on transfer, no study to date has explored whether the dropout procedure promotes

transfer. Additionally, as Pan and Rickard (2018b) pointed out, there are still types of materials and

contexts that have not been investigated enough. One such context is novel vocabulary learning.

Therefore, we aimed at filling that gap by exploring the effects of spaced retrieval on transfer in learning

Swahili.

Current Study

The current study examined the benefit of spaced retrieval on memory for repetition and transfer

items in novel vocabulary learning. We utilized an adapted version of the procedure used by Rawson et

al. (2018). In the first session, participants first engaged in initial study, during which they learned

Swahili words with their English definitions. After studying all pairs, participants began the relearning

block. In this block, a third of the participants were shown the definitions and asked to retrieve each target

5 times (standard condition), a third had to successfully recall each target once (dropout condition), and a

third had to restudy the entire list 5 times (restudy condition). This number of list cycles was chosen
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 11

based on a pilot study showing that it took participants an average of 35 minutes, which translates to 5 list

repetitions, to successfully retrieve all words once. Participants then repeated their respective relearning

tasks in Sessions 2 and 3, each spaced 48 hours after the completion of the previous session. In Session 4,

which took place a week after Session 3, all participants took the same final test, where they had to recall

the appropriate Swahili words when presented with either the same (e.g., “a very young child”; repetition

items) or new (e.g., “a human offspring”; transfer items) English definitions. We chose Swahili – English-

definition paired associates as our materials because they mimic some instances of informal language

acquisition in everyday life. Informal language acquisition refers to contexts outside of formal instruction

in which people improve their knowledge of a language, such as reading or spending time with native

speakers. This mode of learning has been shown to provide benefits to language acquisition (Gass, 1999).

Butler (2010) found that participants who were repeatedly tested on information from a learned

passage experienced better recall for repetition and transfer items than participants who repeatedly

restudied the passages. The first goal of the current study was to investigate whether the same effect

would be found following testing of Swahili words using their English definitions relative to simply

restudying these word – definition pairs. The second aim of this study was to explore whether encoding

variability increased the effect of spaced retrieval on transfer. In Experiment 1 (fixed encoding)

participants studied the same definition (e.g., “a very young child”) associated with a single target (i.e.,

mtoto) across each successive session, whereas in Experiment 2 (variable encoding) participants learned a

different definition (e.g., “a very young child”, “a human offspring”, “the product of a human

reproductive process”) in each session. Prior research failing to show a benefit from variable encoding on

transfer could reflect the lack of variability introduced by materials (i.e., slightly rephrased questions;

Butler, 2010). In the current study, however, variable encoding (i.e., new definitions) orients attention to

features that might be diagnostic for final transfer (i.e., new definitions). Thus, match between encoding

and retrieval for transfer items might be stronger following variable compared to fixed encoding, resulting

in greater transfer with variable encoding.

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 12

The third goal of the study was to compare whether standard and dropout procedures elicit the

same amount of retention and transfer. As mentioned previously, the benefit of the dropout procedure on

transfer has not yet been evaluated in the literature, and the lack of massed retrieval at each session (like

that in the standard procedure) makes it a candidate for a more beneficial technique in terms of time-

efficiency.

Experiment 1

Experiment 1 tested whether spaced retrieval produced superior transfer to spaced restudying in

novel vocabulary learning. According to the three-factor framework of transfer of test-enhanced learning,

the probability of transfer increases with the number of factors included during relearning. In the current

study, all three of the conditions experience response congruency, as the answers required on the transfer

test are the same words that served as targets at relearning. All conditions also involve the same degree of

elaborative feedback, as participants in the retrieval conditions receive corrective feedback with the

correct response and the restudy condition rereads the pair, all without any additional details or

elaboration. However, spaced retrieval might involve a higher degree of broad encoding methods than

restudy does, as retrieval might result in greater cognitive processing than simply reading the material.

Additionally, participants in the spaced retrieval conditions are expected to have a higher degree of

learning after the relearning sessions (i.e., initial test performance) than restudy condition, due to the

classic testing effect. Thus, a finding of greater transfer in spaced retrieval conditions relative to the

restudy condition would indicate that transfer is driven by the difference in the cognitive processing that

occurs at relearning.

We are also interested in comparing the benefit of the two spaced retrieval procedures on transfer.

There were two reasons for including both retrieval conditions. The first was for practical reasons, as the

standard condition is more directly comparable to the restudy condition because the relearning phase

includes five list-learning cycles in each session. When considering the dropout method, the number of

exposures varies for each item, and so choosing an appropriate restudy comparison group is difficult. The

second reason was for application to real-world study contexts. Despite differences in exposure between
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 13

the two testing conditions, presumably the dropout method may be more efficient in terms of time. Thus,

if recall is identical between the two conditions, yet the dropout method is faster, this would suggest that

perhaps the dropout method is the ideal method for learning. No prior research has examined whether the

dropout procedure promotes transfer, and no studies directly compared the two procedures, so we did not

form predictions regarding the benefit of one over the other.

Method

Participants

An a priori power analysis using GPower 3.1 (α = .05, Power = .80) showed that a total of 150

participants (50 per condition) would be required to detect a medium effect size for the Condition x Item

Type interaction. However, due to the pandemic forcing the study online, we experienced low rates of

signups for participation and high rates of attrition. We decided to collect at least 30 participants per

condition. Participants (aged 17 – 47; M = 20.30, SE = 5.17) were undergraduate students from the

University of Texas at Arlington recruited through the SONA systems for class credit. After accounting

for exclusionary criteria (detailed below), the final sample consisted of 95 participants (restudy: 33;

standard: 29; dropout: 33).

Design and Materials

A 3 (condition: restudy, standard, dropout; between-subjects) x 2 (item type: repetition, transfer;

within-subjects) mixed design ANOVA was used to assess memory. Materials included 30 Swahili words

(e.g., mtoto) with their English definitions (e.g., a very young child). The English-Swahili word pairs for

the stimuli were obtained from Nelson and Dunlosky (1994) and Carpenter et al. (2008). For each word,

we created five converging definitions using online dictionaries. Four of the five definitions were used as

the final materials. Four definitions were used in anticipation of using variable encoding during

Experiment 2, which requires a total of four definitions for each Swahili word.

A pilot study (N = 197) was conducted with 60 words to test how well the created definitions

evoked the correct word. We excluded 12 Swahili words that did not have at least four definitions
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 14

averaging at least 75% accuracy. Then, we selected 30 words for which the definitions, on average,

elicited the correct response most frequently. For those words, the four of five definitions with the highest

average recall were selected for the final stimulus materials. Thus, for the final 30 selected Swahili words,

each had four converging definitions. For example, for the Swahili word “mtoto”, the four definitions

would be (1) a very young child, (2) the product of a human reproductive process, (3) a recently born

human, and (4) human offspring.

To counterbalance the procedure, we created two versions of the final test. In one version, half of

the definitions were previously studied by participants (repetition items) and half were new definitions

(transfer items). In the second version, repetition items from the first version appeared as transfer items

and transfer items from the first version appeared as repetition items.

All data collection was completed online using PsychoPy3 software. QuestionPro survey software

obtained participant consent and randomly assigned participants to each condition. The PsychoPy3

experiment was hosted on Pavlovia.com. The study was preregistered on Open Science Framework

website.

Procedure

Initial Learning (Session 1). During Session 1, all participants engaged in two phases. Phase 1

consisted of learning and Phase 2 consisted of relearning. Before beginning Session 1, participants filled

out a demographic questionnaire. After the questionnaire, participants were shown task instructions for

both phases. Participants were told that they would be studying Swahili words with their English

definitions.

Participants in the restudy condition were told that after the initial learning, they would be shown

each Swahili word with its English definition for 10 seconds for restudy. They were instructed to press

the spacebar once they feel they have sufficiently studied the pair, after which they would be presented

with the pair for 4 more seconds (to equate presentation time between conditions). Instructions included

examples of what the participants would see and what they were to do in in this session.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 15

Participants in the retrieval conditions were told that after initial study, they would be shown the

definitions, one by one, and given up to 10 seconds to retrieve the corresponding Swahili words. If they

retrieve the word before that time is up, they were instructed to press the ENTER key. Once that key was

pressed, or the time has run out, participants were presented with the definition, the correct answer, and

colored text to indicate whether their answer was correct (green) or incorrect (red). Before moving on to

the task itself, participants engaged in a practice trial corresponding to their condition.

Phase 1 (Learning). During Phase 1, all participants studied the 30 words and their definitions,

one at a time. Each word-definition pair was presented for 10 seconds for study (DeWinstanley & Bjork,

2004). Then participants engaged in Phase 2.

Phase 2 (Relearning). During Phase 2, participants in the restudy condition were shown the same

word-definition pairs from Phase 1, in random order. They were told to study each pair for up to 10

seconds and press the space bar once they felt that they had learned the word. After 10 seconds had

passed or participants had pressed the space bar, they were presented with the pair again for 4 seconds for

further study. This was done to equate the length of presentation of each stimulus between the conditions

(Pan & Rickard, 2017). Participants restudied the entire list for 5 cycles or 35 minutes, whichever came

first, so as to equate the length of relearning to the other conditions.

Participants in the standard condition engaged in repeated testing for Phase 2. Participants were

presented with the definitions, one at a time and in a random order (Rawson et al., 2018), and given up to

10 seconds to type out the corresponding Swahili words. After each retrieval attempt, regardless of

success, participants were presented with the correct word and its definition for 4 seconds for restudy.

They were tested on the entire list 5 times, so as to equate the length of relearning to the other conditions.

Participants in the dropout condition engaged in retrieval. During retrieval, they were presented

with the definitions, one at a time and in a random order, and given up to 10 seconds to type out the

corresponding Swahili words (Rawson et al., 2018). On both correct and incorrect trials, participants were

presented with the correct answer for 4 seconds for restudy (Pan & Rickard, 2017). If participants

responded incorrectly or omitted a response, the items in those trials were then placed at the end of the
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 16

stack for another recall trial. Participants had up to 35 minutes to recall all of the words. This was done to

equate the total amount of exposure time between the conditions. Pilot studies showed that 35 minutes,

which roughly translates to 5 cycles, was on average enough time for participants to recall all words

correctly.

Relearning (Sessions 2 and 3). Sessions 2 and 3 were relearning sessions. Session 2 was

administered 2 days after initial learning and Session 3 was administered 2 days after Session 2. In each

of these sessions, participants in the retrieval conditions were told that they would be tested on their

memory for the Swahili words from the previous session, and the participants in the restudy condition

were told that they would be restudying the pairs from the previous session. The rest of the instructions

were the same as the ones in Phase 2 of the Initial Learning session. Just as in Session 1, participants

engaged in a practice trial corresponding to their condition before completing the task itself. During these

sessions, participants engaged in the same relearning procedure as the one they completed in Phase 2 of

the initial study session using the same definitions they studied. That is, participants in the dropout

condition engaged in the same cued-recall test with feedback for up to 35 minutes as they did at initial

study, participants in the standard condition were tested on the entire list 5 times, and participants in the

restudy condition restudied all materials for 35 minutes/5 cycles.

Final Test (Session 4). A week after Session 3 (Rawson et al., 2018), all participants took the

same final test, in which half of the items were tested using previously seen definition cues (repetition;

e.g., a very young child) and half were tested using new definition cues (transfer; e.g., the product of a

reproductive process). Participants were told that they would be shown English definitions and that they

had unlimited time to retrieve the corresponding Swahili words. The instructions stated that some of the

definitions would be from previous sessions, and some would be new English definitions corresponding

to previously studied Swahili words. Participants in the retrieval conditions were told that the test would

be in a similar format to tests from relearning, but that each word would only be tested once, and no

feedback would be provided.

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 17

Post-Session Questionnaire. At the end of each session, participants filled out a questionnaire

asking them if they a) had used any outside aid (such as an online dictionary), b) written any words or

definitions down, and c) to provide any other information that might have affected their performance.

Attention Checks. For all experiments, each session included an attention check interspersed

with other questions that asked for demographic information. The demographic questions were added so

that participants did not catch on to attention checks and only paid attention at the end of the experiment.

The answers to these questions were obvious, and the attention checks were created in order to exclude

data of participants who were not paying attention during the tasks.

Data Scoring

For all experiments, recall was hand-scored by the experimenter. As the target words that

participants had to type in were in Swahili and might have been challenging to spell, a separate strict and

lenient scoring systems were utilized. For the strict scoring system, a word was only counted as correct if

it exactly matched the correct Swahili word. For the lenient scoring system, a word was counted as

correct if the word a) was only misspelled by one letter (e.g., mtota); b) was missing one letter (e.g.,

mtot); c) had two consecutive letters that were switched (e.g., motto); d) had one extra letter (e.g.,

matoto); or e) was inputted correctly, but because of a technical issue was scored as incorrect. All

analyses were conducted using each scoring system separately. However, as they produced a very similar

pattern of results, only the lenient scoring results are presented below.

Exclusionary Criteria

Exclusionary criteria were preregistered. The exclusionary criteria were as follows, with the

number of subjects fitting each criteria noted in parentheses: a) native language was not English or had

previous knowledge of Swahili, as determined by responses in the Demographics questionnaire at the

beginning of the study (n = 10); b) failed 75% (3 out of 4) attention checks (n = 0); and c) reported

writing words or definitions down, using outside aid, or otherwise cheating (n = 7). Additionally, three

participants were excluded due to system error with the experiment. This criterion was not preregistered

or foreseen, but the experimental error compromised the data obtained from these participants.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 18

We originally preregistered that we would exclude participants in the restudy condition that did

not press space at least once during a single relearning session (as an index of “accuracy”) or for

participants whose average study time was less than 1 second. However, many participants never pressed

the spacebar, while others had brief durations. In hindsight, we realized that is possible that 10 seconds

was not sufficient time for some participants to feel that they have learned the words, while for others 1

second may have been enough (since after terminating study the information was presented again for 4

seconds). We therefore analyzed the data with and without these participants. As there was no difference

in the results of the two analyses, we report analyses including all of these participants.

Attrition

Attrition rates were high for this experiment (52% in total). We are not sure of the reason for such

high attrition rates, but it is possible that students struggled to participate in four online experimental

sessions that were spread out over two weeks. In the restudy condition, 75 participants started the study,

and 39 participants completed all sessions. In the dropout condition, 76 participants started the study, and

41 participants completed all sessions. In the standard condition, 67 participants started the study, and 34

participants completed all sessions. Importantly, a chi-square test revealed that there were no differences

in attrition rates between the 3 conditions, [X2(4, N = 221) = 2.26, p = .69].

Results

Data from 95 participants are reported below. We analyzed final recall using strict and lenient

scoring schemes. All results from both analyses were significant at an alpha level of .05. Therefore,

analyses shown below have been conducted using the lenient scoring scheme.

Final Test Performance

Figure 1 displays the means and standard errors for retention and transfer in the dropout,

standard, and restudy conditions. A 3 (condition: restudy, standard, dropout; between-subjects) X 2 (item

type: repetition, transfer; within-subjects) ANOVA revealed that participants recalled more repetition than

transfer words on the final test [Item Type: F(1, 92) = 30.74, p < .001, ηp2 = .25]. There was also a
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 19

significant main effect of Condition [F(2, 92) = 9.11, p < .001, ηp2 = .17]. These main effects were

qualified by a significant interaction [Condition x Item Type: F(2, 92) = 7.02 p = .001, ηp2 = .13].

To probe the interaction, we conducted separate one-way ANOVAs for repetition and transfer

items. For repetition items, there was a significant main effect of condition [Condition: F(2, 92) = 11.32,

p < .001, ηp2 = .20]. Participants in the restudy condition (M = .42, SE = .05) recalled significantly fewer

words than participants in the dropout condition (M = .67, SE = .05) and those in the standard condition

(M = .75, SE = .05). There was no difference in recall between the dropout and standard conditions. For

transfer items, there was a main effect of condition [Condition: F(2, 92) = 6.33, p = .003, ηp2 = .12].

Participants in the restudy condition (M = .39, SE = .05) recalled significantly fewer words than

participants in the standard condition (M = .65, SE = .05), but not significantly fewer than those in the

dropout condition (M = .47, SE = .05). Additionally, participants in the dropout condition recalled

significantly fewer transfer items than participants in the standard condition.

Figure 1
Retention and Transfer in Experiment 1

1.00 Repetition
Proportion correct

.80 Transfer

.60

.40

.20

.00
Restudy Standard Dropout

Note. Spaced retrieval produced better recall for repetition items than
restudy, but only the standard condition produced better transfer than the
restudy condition. Bars represent standard error of the mean.

Rate of Learning
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 20

Figure 2 displays the means and standard errors for recall accuracy during each study session in

the dropout and standard conditions. We conducted a preregistered analysis of rate of learning for the two

conditions to examine whether different relearning techniques yielded differences in how quickly

participants acquired material during learning. To do that, we calculated the proportion of words (out of

30) participants in each condition recalled in the first list cycle they were tested on in each session1. A 2

(condition: standard, dropout; between-subjects) x 3 (session: 1, 2, 3; within-subjects) mixed-effects

ANOVA revealed that number of items recalled on the first test in each session increased across sessions

[Session: F(2, 120) = 108.57, p < .001, ηp2 = .64]. Recall did not differ between the two conditions

[Condition: F(1, 60) = 3.42, p = .07, ηp2 = .05] and there was no interaction between the two variables,

[Session x Condition: F(2, 120) = 2.97, p = .055, ηp2 = .05].

Figure 2
Relearning Accuracy Across Days in Experiment 1

1.00 Standard
Dropout
.80
Proportion correct

.60

.40

.20

.00
Session 1 Session 2 Session 3

Note. Overall, there was no difference in learning rates between the dropout and standard
conditions. Bars represent standard error of the mean.

We also conducted exploratory (but pre-registered) correlational analysis between accuracy in the

first recall cycle of each relearning session and the final recall of repetition and transfer items, collapsed

1
Note that this comparison cannot be done for subsequent cycles (e.g., cycle 5) because recalled items are not
included in later cycles in the dropout condition.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 21

across the two retrieval conditions. For final recall of repetition items, there was no correlation with

Session 1 performance, but there were sizable positive correlations with Session 2 and 3 performance,

[Session 1: r(60) = .12, p = .36; Session 2: r(60) = .50, p < .001; Session 3: r(60) = .75, p < .001. The

same pattern appeared for correlations between final recall for transfer items and the three sessions,

[Session 1: r(60) = .10, p = .44; Session 2: r(60) = .48, p < .001; Session 3: r(60) = .75, p < .001].

Additionally, there was a strong positive correlation between final recall for repetition items and recall for

transfer items [r(60) = .75, p < .001]. Given the small sample size, these results should be interpreted with

caution.

Total Relearning Time

Differences in the duration of relearning can be an important consideration in research and

learning environments, where time is a limited resource. We therefore conducted exploratory analyses to

examine the time spent relearning in each session in the standard and dropout retrieval conditions.

Relearning time reflects the relearning duration in each session and does not include time spent on

instructions, practice, or post-session questionnaires.

Figure 3 displays the means and standard errors for time spent on relearning in each session in the

standard and dropout conditions. A 2 (condition: standard, dropout; between-subjects) x 3 (session: 1, 2,

3; within-subjects) mixed-effects ANOVA on relearning time revealed that participants in the standard

condition spent more time relearning overall [Condition: F(1, 60) = 139.74, p < .001, ηp2 = .70]. There

was also a significant main effect of session, [Session: F(2, 120) = 120.29, p < .001, ηp2 = .67], with

relearning duration decreasing each session. These main effects were qualified by a significant

interaction, [Condition x Session: F(2, 120) = 42.95, p < .001, ηp2 = .42]. This interaction reflects that the

dropout procedure was faster than the standard procedure in each session, and this difference increased

across sessions.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 22

Figure 3
Relearning time Experiment 1

35 Session 1
30 Session 2
25 Session 3
Time (min)

20
15
10
5

Standard Dropout

Note. Participants in the dropout condition spent significantly

less time training than participants in the standard condition.
Bars represent Standard error of the mean.

Discussion

The purpose of Experiment 1 was to examine whether spaced retrieval produced superior

retention and transfer relative to restudying in novel vocabulary learning. A second aim was to compare

the two spaced retrieval conditions in terms of retention and relearning time, as this comparison has not

been conducted previously (but see Karpicke & Roediger, 2007 for comparison of different single-session

criterion levels with dropout). We found that both retrieval procedures improved recall of repetition items

over restudy. In contrast, only the standard procedure improved memory for transfer items over restudy.

To examine why standard, but not dropout procedure of relearning promotes transfer, we refer to the

three-factor framework for transfer of test-enhanced learning. Any differences in the levels of factors

between the two testing procedures could point to the reason for these results.

The first factor, response congruency, was equated between the two retrieval conditions, as both

conditions were required to produce the same responses on the transfer test as they did during testing. The

factor of elaborative feedback was also equated between the two conditions: both received non-
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 23

elaborative feedback at relearning. The results of our analysis of learning rates showed that there were no

differences in overall recall between the two conditions.

This leaves broad encoding methods as a possible imbalanced factor driving the difference in

transfer between the two conditions. Broad encoding methods are present in relearning procedures that

orient participants to engage in deeper cognitive processes of the material relatively to simply restudying.

One possibility is that the standard, but not the dropout, procedure induced deeper cognitive processing in

our experiment by repeatedly retrieving information in each session. Our dropout condition only had to

retrieve each word correctly once, limiting the number of retrieval attempts for some words. Experiment 2

was designed to examine whether increased levels of broad encoding through encoding variability might

reduce transfer differences between the two retrieval conditions and improve transfer for the dropout

condition relative to the restudy condition.

Finally, our analysis of the relearning time in the two retrieval conditions in each session revealed

that participants in the standard condition spent more time on learning overall. Thus, the improved

memory for repetition and transfer items in the standard conditions comes at a cost in terms of time taken

to complete the relearning.

Experiment 2

The purpose of Experiment 2 was to investigate whether encoding variability would promote

transfer in novel vocabulary learning. To do that, we used the same procedure as in Experiment 1, with

the exception that participants were presented with different English definitions of the same Swahili

words in each relearning session. On the final test, half of the words were tested with definitions from the

last relearning session (repetition items) and half were tested with new, never-before-seen definitions

(transfer items).

One possible way to promote cognitive processing is encoding variability. Introducing variability

at encoding has been shown to improve learning by creating more retrieval routes (Estes, 1955; Martin,

1968; Kukull et al., 2002). It has also been shown to promote transfer to novel tasks (e.g., Wahlheim et

al., 2012; Barcroft & Sommers, 2005). Some studies have found the benefit of variable encoding on
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 24

transfer following retrieval practice (Goode et al., 2008; Smith & Handy, 2014, 2016; Butler et al., 2017).

However, this effect is only found when learning stimuli introduce a sufficient degree of variability

(Butler, 2010). In the current experiment, we explored whether practicing with different definitions of the

same Swahili word would introduce enough variability to promote transfer and whether encoding

variability would induce greater cognitive processing in the dropout condition.

We expected that varied definitions will benefit recall of transfer items more than repetition

items. Transfer is dependent on elaborative encoding of learned material (i.e., broad encoding methods),

which encoding variability supports. Retrieval of learned information without application to a new task,

while benefitting from elaborative encoding, does not necessarily depend on it to the same degree. If so,

we should find that variable learning benefits transfer items more than repetition items.

We also hypothesized that varied definitions would benefit transfer more in the spaced retrieval

conditions relative to the restudy condition. The reason for this prediction is that we expected the greater

cognitive processing introduced by varied definitions to have an additive effect with the already present

greater cognitive processing in the spaced retrieval (versus restudy) conditions. According to the

elaborative retrieval hypothesis, testing results in a more extensive memory search than restudying, which

creates a more elaborative memory trace (Carpenter & DeLosh, 2006). Encoding variability using varied

definitions improves memory by creating more retrieval routes. Therefore, testing and variability improve

learning by two different processes which might have an independent boosting effect on items.

Alternatively, it might be that testing and elaborative encoding both enhance cognitive processing through

the same mechanisms of greater elaboration and strengthening of memory traces. If that is the case, the

restudy condition, which engages in less cognitive processing than the testing conditions do, might see

more benefit from encoding variability.

The second aim of this experiment was to examine whether encoding variability would have a

different effect on the two spaced retrieval conditions. Results of Experiment 1 showed that the dropout

method did not show improved memory for transfer memory. Assuming that the testing condition is

already engaging in broad encoding following multiple retrievals, we would expect that encoding
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 25

variability would boost transfer more in the dropout than the restudy condition. We note, however, that

this is a post-hoc hypothesis.

Method

Participants

An a priori power analysis using GPower 3.1 (α = .05, Power = .80) showed that a total of 150

participants (50 per condition) would be required to detect a medium effect size. However, due to the

pandemic forcing the study online, we have experienced low rates of signups for participation and high

rates of attrition. We again decided to collect at least 30 participants per condition. Participants (aged 17 –

45; M = 19.48, SE = 3.93) were undergraduate students from the University of Texas at Arlington

recruited through the SONA systems for class credit. After accounting for exclusionary criteria (detailed

below), the final sample consisted of 95 participants (restudy: 36; standard: 30; dropout: 29).

Design

A 3 (condition: restudy, standard, dropout; between-subjects) X 2 (item type: repetition, transfer;

within-subjects) mixed design ANOVA was used to analyze the proportion of words correctly recalled on

the criterion test.

Materials and Procedure

The materials and procedure were the same as in Experiment 1, with one exception. During each

session, participants were presented with a different English definition for the same Swahili word. The

instructions for each session were similar to those in Experiment1, but in Sessions 2 and 3 participants

were told that all of the definitions would be new English definitions corresponding to the Swahili words

from the previous session.

To counterbalance the procedure, we created two versions of the materials. In one version, half of

the words on the final test were tested with the definitions from Session 3, and half were tested with new,

never before seen definitions. In the second version, the former words were tested with new definitions

and the latter words were tested with definitions from Session 3.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 26

Exclusionary Criteria.

Exclusionary criteria were preregistered. The exclusionary criteria were as follows, with the

number of subjects fitting each criterion noted in parentheses: a) native language was not English or had

previous knowledge of Swahili, as determined by responses in the Demographics questionnaire at the

beginning of the study (n = 5); and b) reported writing words or definitions down, using outside aid, or

otherwise cheating (n = 2).

Like in Experiment 1, we did not exclude participants that scored 0% accuracy on any single

relearning session and participants whose study duration was less than 1 second in the restudy condition.

Analyses with and without those participants showed no difference in the results.

Attrition

Attrition rates were high for this experiment (50% in total). In the restudy condition, 75

participants started the study, and 37 participants completed all sessions. In the standard condition, 60

participants started the study, and 33 participants completed all sessions. In the dropout condition, 71

participants started the study, and 33 participants completed all sessions. A chi-squared test of

independence showed there were no differences in the attrition rates between conditions [X2(4, N = 207) =

3.17, p = .53].

Results

Data from 95 participants are reported below. We analyzed final recall using strict and lenient

scoring schemes, which produced the same results. We therefore report analyses using the lenient scoring

scheme.

Final Test Performance

Figure 5 displays means and standard errors for retention and transfer in the dropout, standard,

and restudy conditions. A 3 (condition: restudy, standard, dropout; between-subjects) X 2 (item type:

repetition, transfer; within-subjects) ANOVA revealed no effect of test, [F < 1] and no Item Type x

Condition interaction [F(2, 92) = 1.94, p = .15, ηp2 = .04]. However, there was a significant effect of

Condition [F(2, 92) = 4.17, p = .02, ηp2 = .08]. The condition effect reflects that participants in the restudy
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 27

condition (M = .49, SE = .05) recalled fewer words than participants in the standard condition (M = .70,

SE = .05), but not significantly so than those in the dropout condition (M = .62, SE = .05). There was no

difference in recall between the dropout and standard conditions.

Although there was no test x condition interaction, we conducted preregistered analyses for recall

differences between conditions separately for repetition and transfer items. As can be seen in Figure 4, for

repetition items, participants in the restudy condition recalled significantly fewer words than participants

in the standard condition, [F(1, 64) = 10.14, p = .002, ηp2 = .14], and dropout condition, [F(1, 63) = 4.35,

p = .04, ηp2 = .07]. Recall in the standard condition did not differ significantly from the dropout condition

[F < 1]. For transfer items, participants in the restudy condition recalled significantly fewer words than

participants in the standard condition, [F(1, 64) = 5.81, p = .02, ηp2 = .08], but not the dropout condition,

[F(1, 63) = 1.52, p = .22, ηp2 = .02]. Recall in the standard condition did not differ significantly from the

dropout condition, [F(1, 57) = 1.01, p = .32, ηp2 = .02].

Figure 4
Retention and Transfer in Experiment 2

1.00 Repetition
Proportion correct

.80 Transfer

.60

.40

.20

.00
Restudy Standard Dropout

Note. Standard condition produced better recall for repetition, but not
transfer items than the restudy condition. Dropout condition did not
produce better recall of repetition or transfer items relative to restudy.
Bars represent standard error of the mean.

Rate of Learning
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 28

Figure 5 displays means and standard errors for recall accuracy in each session in the dropout and

standard conditions. A 2 (condition: standard, dropout; between-subjects) x 3 (session: 1, 2, 3; within-

subjects) mixed-effects ANOVA revealed that recall increased with each new session, [Session: F(2, 114)

= 50.20, p < .001, ηp2 = .47]. There was no effect of condition, [Condition: F(1, 57) = 1.70, p = .20, ηp2 =

.03] and no significant interaction between the two, [Session x Condition: F < 1].

Figure 5
Learning Accuracy Across Days in Experiment 2

1.00 Standard
Dropout
.80
Proportion correct

.60

.40

.20

.00
Session 1 Session 2 Session 3

Note. Overall, there were no differences in the rate of learning between the dropout and
standard conditions. Bars represent standard error of the mean.

We also conducted correlational analysis between accuracy in the first recall cycle of each

relearning session and final recall of repetition and transfer items, collapsed across retrieval conditions.

As with Experiment 1, this analysis showed no correlation of final recall for repetition items with initial

performance in session 1, but there was a strong positive correlation in sessions 2 and 3, [Session 1: r(56)

= .06, p = .67; Session 2: r(56) = .68, p < .001; Session 3: r(56) = .82, p < .001]. The same pattern

appeared for correlations between final recall for transfer items and the three sessions, [Session 1: r(56) =

-.07, p = .63; Session 2: r(56) = .72, p < .001; Session 3: r(56) = .89, p < .001]. Additionally, there was a

strong positive correlation between final recall for repetition and transfer items, [r(56) = .87, p < .001].

Given the small sample size, these results should be interpreted with caution.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 29

Total Relearning Time.

Figure 6 displays means and standard errors for time spent on relearning in each session in the

dropout and standard conditions A 2 (condition: standard, dropout; between-subjects) x 3 (session: 1, 2, 3;

within-subjects) mixed-effects ANOVA on relearning time revealed that participants in the standard

condition spent more time relearning overall [Condition: F(1, 56) = 41.03, p < .001, ηp2 = .42]. There was

also a significant main effect of session, [Session: F(2, 112) = 50.15, p < .001, ηp2 = .47]. These main

effects were qualified by a significant interaction, [Condition x Session: F(2, 112) = 11.60, p < .001, ηp2 =

.17]. This interaction reflects that the dropout procedure was faster in each session, but this difference

increased across sessions.

Figure 6
Relearning time Experiment 2

35 Session 1
30 Session 2
Time (min)

25 Session 3
20
15
10
5

Standard Dropout

Note. Participants in the dropout condition spent significantly

less time training than participants in the standard condition.
Bars represent Standard error of the mean.

Discussion

Experiment 2 was designed to examine whether learning novel vocabulary using varied

definitions promotes greater transfer relative to recall of repetition items and to a greater degree in spaced

retrieval relative to repeated restudy conditions. We were also interested in whether this encoding

variability had a different effect on the two spaced retrieval conditions. The results are largely consistent
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 30

with Experiment 1, such that overall, the standard condition produced the best recall, the restudy

condition had the worst recall, and the dropout condition fell somewhere in between. When examining

item types separately as a function of condition, the results showed that the standard and dropout

procedures produced superior recall of repetition items compared to the restudy condition. In contrast,

only the standard condition showed superior transfer over the restudy condition (although the dropout

condition was numerically greater than the restudy condition and did not differ from the standard

condition). We hypothesized that studying with different definitions would promote greater cognitive

processing and accentuate the differences between the retrieval and restudy conditions for both repetition

and transfer items. However, it appears that introducing variability had relatively little influence on

performance, other than producing a numerical increase in transfer for the dropout condition. However, as

encoding variability was manipulated across different experiments, it is not possible to directly explore

this idea. In the following section we therefore conducted preregistered exploratory cross-experimental

analyses.

Cross-Experimental Analyses

To better understand how variable encoding affects retention and transfer, we conducted an

analysis including data from both experiments. Direct comparison of these two encoding manipulations

can provide a better insight into the mechanisms underlying retention and transfer in the three conditions.

Therefore, we conducted separate 2 (item type: repetition, transfer; within-subjects) x 2 (encoding: fixed,

variable) ANOVAs for each condition. Figure 7 displays means and standard errors for retention and

transfer in the restudy (A), standard (B), and dropout conditions (C).

Analysis for the restudy condition revealed no main effect of Item Type, [F < 1], or Encoding,

[F(1, 67) = 1.51, p = .22, ηp2 = .02], and there was no interaction between the two, [Item Type x Encoding:

F(1, 67) = 2.16, p = .15, ηp2 = .03]. Thus, encoding variability did not influence memory.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 31

Figure 7 (A)
Restudy Condition

1.00 Repetition
.80 Transfer
Proportion correct

.60

.40

.20

.00
Fixed Encoding Variable Encoding

Note. Encoding variability had no effect on retention and

transfer in the restudy condition. Bars represent standard
error of the mean.

Analysis for the standard condition revealed that participants recalled more repetition items [Item

Type: F(1, 57) = 7.71, p = .01, ηp2 = .12]. Recall did not differ across encoding conditions [Encoding: F <

1] and there was no interaction between the two [Item Type x Encoding: F(1, 57) = 3.25, p = .08, ηp2 =

.05]. Thus, encoding variability did not influence memory.

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 32

Figure 7 (B)
Standard Condition

1.00 Repetition
Transfer
.80
Proportion correct

.60

.40

.20

.00
Fixed Encoding Variable Encoding

Note. Encoding variability had no effect on retention and

transfer in the standard condition. Bars represent standard
error of the mean.

Analysis for the dropout condition revealed that participants recalled more repetition than transfer

items, [Item Type: F(1, 60) = 27.45, p < .001, ηp2 = .31]. Although there was no effect of encoding,

[Encoding: F < 1], there was a significant interaction between the two factors, [Item Type x Encoding:

F(1, 60) = 14.76, p < .001, ηp2 = .20]. This reflects that variable encoding did not affect recall of repetition

items [F < 1], but numerically (but not significantly) improved recall for transfer items [F(1, 60) = 3.27, p

= .075, ηp2 = .05]. Thus, encoding variability does numerically increase transfer.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 33

Figure 7 (C)
Dropout Condition

1.00 Repetition
Transfer
Proportion correct

.80

.60

.40

.20

.00
Fixed Encoding Variable Encoding

Note. Encoding variability improved transfer but not

retention in the dropout condition. Bars represent standard
error of the mean.

The results of our condition-level analyses showed that variable encoding numerically promotes

transfer in dropout, but not standard or restudy conditions. This finding is in line with our prediction that

variable encoding promotes greater cognitive processing (i.e., broad encoding methods) in the dropout

condition. It is possible that the manipulation did not have the same effect in the standard condition

because retrieving items repeatedly in a single session might have already forced participants to engage in

greater cognitive processing. For the restudy condition, it seems that encoding variability failed to induce

greater cognitive processing, as evident by the lack of the effect of encoding.

General Discussion

In this study we explored whether spaced retrieval promotes learning of Swahili words from their

English definitions compared to repeated restudy and whether this transfers to improved memory for

vocabulary using never-before seen definitions. Learning novel vocabulary is a universal human

experience: from toddlers learning their mother tongue, to schoolchildren acquiring more complex

vocabulary, to anyone trying to master a new language. Much of language acquisition occurs in informal

settings, such as from conversing with those who already have a mastery of the language or from reading

books. To simulate such learning, we used Swahili words with the English definitions as our stimuli.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 34

Although there is previous research on the benefits of retrieval practice for retention and transfer, no

studies have examined these processes with such materials. We found that the standard spaced retrieval

procedure resulted in higher recall of words cued with the same (repetition) and new (transfer) definitions

than studying by rereading the material regardless of whether the same (fixed) or different (variable)

definitions were studied across relearning sessions. Additionally, we found that although more efficient in

terms of relearning duration, the dropout procedure of spaced retrieval produces less transfer than the

standard procedure. Below we discuss the theoretical and applied ramifications of these findings.

Memorial Benefit of Spaced Retrieval on Repetition Items

The first aim of this study was to investigate the effect of spaced retrieval and repeated restudy on

retention of repeated definitions and terms during new language acquisition. In line with our predictions,

both Experiment 1 and 2 showed that both spaced retrieval procedures promoted retention over

restudying. The benefit of retrieval over restudy was not surprising. According to the elaborative retrieval

hypothesis, when attempting to retrieve information, participants engage in a memory search that

reactivates information that is relevant to the target information, creating a more elaborative memory

trace for the target. At test, some of this relevant information can be used to cue the target information,

making retrieval more likely. Restudying information, on the other hand, does not result in a memory

search, and thus does not produce an equally elaborated memory trace for target information that can be

helpful at test (Carpenter & DeLosh, 2006). Additionally, the desirable difficulty hypothesis poses that

conditions that make initial learning more difficult result in better retention of information (Bjork, 1994,

1999). Retrieval practice is more difficult than restudying, because it forces participants to engage in

memory search and recall the target, thereby improving retention. Finally, the mediator effectiveness

hypothesis states that in cued-recall paradigms, such as the one used in this study, participants generate

mediators to link the cue to the target at study. During retrieval, the mediator can be activated by the

presented cue and aid in retrieval of the target. Retrieval leads to creation of more effective mediators

than restudy, thereby promoting recall (Carpenter, 2011).

Memorial Benefit of Spaced Retrieval on Transfer Items

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 35

The second aim of the study was to investigate whether spaced retrieval would benefit recall of

Swahili terms using definitions that were semantically related to previously studied definitions, but never

actually learned. Given that prior research has shown that transfer can occur for different types of

materials (for a meta-analysis see Adesope et al., 2017), we hypothesized that spaced retrieval would be

particularly beneficial for transfer using term-definition pairs. Our results supported this prediction, with

the qualification that it depends on the type of spaced retrieval used. Regardless of whether the same

definitions were used during each relearning session (Experiment 1) or different definitions were used

(Experiment 2), we found that the standard condition promoted transfer over restudy. The finding of

superior transfer in the standard condition compared to restudy is in line with previous research on the

transfer benefits of retrieval (Pan & Rickard, 2018b), as well as previous findings on improved transfer

following spaced retrieval compared to repeated restudy (Butler et al., 2017; Butler, 2010). In contrast, we

were surprised to find the lack of transfer benefit in the dropout condition. It is possible that learning to a

criterion of one is not sufficient for transfer to occur.

According to the three-factor framework of transfer of test-enhanced learning, three factors affect

transfer: response congruency, elaborated retrieval practice, and initial learning. Response congruency

refers to the similarity between responses required at relearning and responses required on the final test

(e.g., “mtoto” as a response to “a very young child” at relearning and to “a human offspring” at final test).

Elaborated retrieval practice contains broad encoding methods, which refers to study designs that orient

participants to information from study session other than the target (e.g., “mtoto” refers to “baby”), and

elaborative feedback, which refers to feedback that explains the underlying concept of the correct

response (e.g., both “a very young child” and “a human offspring” refer to a baby). Finally, initial

learning refers to the amount of information retained at the end of relearning.

In the current study, the restudy, standard, and dropout condition all had the same degree of

response congruency, as they were all required to retrieve the same targets on the final test as the ones

they encountered at relearning. The three conditions were also equated on the degree of elaborative

feedback. While the retrieval groups received corrective feedback (i.e., “correct!” or “incorrect!”) with
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 36

the opportunity to restudy the correct answer, and the restudy group was given the same length of time to

study the correct answer, no condition was given any further information about why a particular response

is correct. In contrast, retrieval conditions presumably had a higher degree of the factor of initial learning

than the restudy condition. This assumption is based on the classic testing effect that testing results in

better memory than restudy does. The analysis of relearning accuracy on the first cycle of each session

showed that initial learning (i.e., recall accuracy in Session 3) did not differ between the standard and

dropout conditions. Broad encoding methods (i.e., greater cognitive processing) were also assumed to be

greater in the retrieval conditions compared to the restudy condition, as retrieving information requires

more cognitive processing (e.g., memory search) than simply restudying it. The results of Experiment 1

showed that the standard, but not dropout, procedure promoted transfer over restudy led us to explore

whether there was an imbalance in the degree of broad encoding methods between the two retrieval

conditions, as it was the only factor that might have been different between them. Broad encoding

methods are supported by study designs that orient participants to engage in more elaborative processing

(Pan & Rickard, 2018b). Therefore, we argued that lack of the transfer benefit in the dropout condition

was due to the procedure not inducing as much elaborative processing as the standard procedure does.

Indeed, Experiment 2 showed that when greater cognitive processing is promoted via encoding

variability, the differences in retention and transfer between the two conditions were attenuated. In other

words, it seems that multiple retrievals (at each session) in the standard condition produce substantial

“broad encoding” to promote transfer, whereas a single successful retrieval does not.

Although the dropout method did not result in greater transfer than the restudy condition, it is

worth noting that cross-experimental comparison revealed that variable encoding increased transfer for

the dropout methods (albeit only marginally), but not the standard or restudy conditions. This suggests the

variable encoding alone does not produce transfer if participants are simply restudying materials.

Likewise, if participants have engaged in multiple retrievals of the same item, as in the standard

condition, variable encoding may have little influence on memory for transfer items. However, variable

encoding combined with minimal retrieval (i.e., a criterion of one) may at least have some benefit for
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 37

performance. To address this issue, in future work we plan to compare fixed versus variable encoding

within-subjects and to manipulate the criterion level (e.g., 1 vs. 3 vs. 5) required in the dropout condition.

In any manner, the failure to find benefits of variable encoding in standard and restudy conditions, and

only a marginal benefit in the dropout condition, could be because the varied definitions used in our

experiment did not introduce enough encoding variability to produce an effect. Butler (2010) found that

rephrasing questions participants had to answer at relearning did not boost transfer over relearning with

the same questions. A later study using similar design overcame this problem by creating questions that

required participants to apply a concept embedded in previously learned information to a new example

(Butler et al., 2017). Perhaps the definitions we created did not differ enough from each other to simulate

applying knowledge to a new example.

Comparing the Two Repeated Retrieval Techniques

We were also interested in examining differences between two spaced retrieval techniques:

standard spaced retrieval and dropout spaced retrieval. The robust memorial benefit of spaced retrieval

has prompted researchers to recommend its use in school settings (Roediger & Pyc, 2012; Dunlosky &

Rawson, 2015; Roediger & Karpicke, 2006a). Time usage of any learning method is an important

consideration for students, and therefore finding a learning technique that is most time efficient while still

producing long-term retention and transfer is important for naturalistic applications of learning research.

The results from the current study demonstrated the dropout procedure was more time-efficient and

resulted in comparable memory for repetition items to the standard condition. However, this was not the

case for the transfer items. Thus, recommendations for which procedure is most optimal depends on the

goal of the learner (i.e., retention versus transfer). In addition to application, the results from the current

study have important implications for the theoretical mechanisms underlying standard versus dropout

methods.

Both the standard and dropout techniques provide a robust memorial benefit on long-term

learning over restudying (Rowland, 2014; Rawson et al., 2013; Rawson et al., 2018). However, these two

techniques have never been compared to each other with regard to their benefits on transfer. While both
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 38

techniques involve several sessions of retrieval, within in each session, the standard procedure requires

spaced retrieval of all items regardless of retrieval success, and the dropout procedure only requires that

items be retrieved until the criterion is met. This leads to an imbalance of exposure and retrieval attempts

for some targets between the conditions. For example, if the word “mtoto” is successfully retrieved on the

first cycle of the standard condition, it will have to be retrieved several more times. In contrast, “mtoto”

only has to be successfully retrieved in the dropout condition until it is recalled to criterion (e.g., once),

after which it is dropped from further testing and restudy. In essence, the difference between the two

conditions is one of intersession repeated retrieval. As a result, the dropout method might be less time

consuming, but the question remained whether it is as beneficial as the standard method.

We found that when learning to a criterion of one, the dropout procedure did take less time and

resulted in the same degree of initial relearning (i.e., on cycle 1 of each relearning day). However, unlike

the standard condition, performance on the final recall for transfer items in the dropout condition was not

better than repeated restudy when studying with the same definitions or different definitions (although

performance was at least numerically greater following variable encoding). One possibility for the

differences in performance across retrieval conditions is memory strength. Every time information is

successfully retrieved, the memory trace for that information is strengthened and decays at a slower rate

(Pavlik & Anderson, 2005; Bjork & Bjork, 1992). Information in the dropout condition could only be

successfully retrieved once at each relearning session, and therefore could only be strengthened once. In

contrast, because participants in the standard condition had to attempt retrieval of each item 5 times in

each relearning session, there is a possibility that some items were successfully retrieved more than once.

If this is the case, more targets had stronger memory traces in the standard condition than in the dropout

condition at the end of each session. Moreover, these memory traces would decay at a slower rate than the

memory traces for targets in the dropout condition. A week after the last relearning session, then, the

standard condition would have more memory for the studied material than the dropout condition, which

would result in better retention and a higher probability of transfer occurring.

SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 39

An alternative explanation of our findings is proposed by the desirable difficulties hypothesis

(Bjork, 1994, 1999). The hypothesis states that learning that is slower and more effortful results in better

long-term retention than learning that is fast and easy. Karpicke and Roediger (2007) had participants in

one condition engage in several restudy sessions, after each of which they took a free-recall test on all of

the material (STST condition). For participants in another condition, the words that were recalled

correctly were dropped from future restudy and testing blocks (STS nTn condition). A week after the initial

session, all participants returned for a final free-recall test on all of the words studied initially. Their

results showed that recalling each word only once resulted in worse retention after a week than retrieving

all words several times. Similarly, Rawson et al. (2020) found that the dropout procedure provided only a

marginal benefit on retention over single-session learning. They proposed that learning to a criterion of

one stops learning before functional mastery is achieved. This explanation is supported by the literature

on overlearning, which suggests that continuing learning after material has been learned enough to be

recalled enhances long-term retention (Postman, 1962; for a meta-analysis, see Driskell et al., 1992).

These two explanations are not mutually exclusive. It is possible that the mechanism driving the

learning benefit of overlearning is that it introduces desirable difficulties by forcing learners to engage in

more elaborative processing. This would explain superior transfer in the standard condition relative to the

dropout condition we found in Experiment 1. To explore whether this is the case, future research should

manipulate learning criterion in the dropout condition to examine learning and transfer.

Conclusion

The results of the current study showed that spaced retrieval improve retention and transfer over

restudy in novel vocabulary learning. Moreover, the comparison of the standard and dropout procedure of

spaced retrieval revealed that while standard spaced retrieval takes more time, it results in more transfer.

Our findings suggest that if the goal of learning is retention, the dropout method is a time efficient

procedure that yields long-term retention, even without variability in learning materials. To achieve the

goal of transfer, the standard procedure is more beneficial. However, when used with variable materials,

the dropout method provides a faster alternative to the standard method. These findings have important
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 40

implications for future research on mechanisms underlying spaced retrieval as well as recommendations

made to instructors.
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 41

References

Adesope, O. O., Trevisan, D. A., & Sundararajan, N. (2017). Rethinking the use of tests: A meta-
analysis of practice testing. Review of Educational Research, 87(3), 659-701.

Agarwal, P. K., Bain, P. M., & Chamberlain, R. W. (2012). The Value of Applied Research:
Retrieval Practice Improves Classroom Learning and Recommendations from a Teacher,
a Principal, and a Scientist. Educational Psychology Review, 12.

Bahrick. (1979). Maintenance of knowledge: Questions about memory we forgot to ask. Journal
of Experimental Psychology: General, 108(3).

Barcroft, J., & Sommers, M. S. (2005). Effects of acoustic variability on second language
vocabulary learning. Studies in second language acquisition, 27(3), 387-414.

Bjork, R. A. (1994). Memory and metamemory considerations in the. Metacognition: Knowing

about knowing, 185(7.2).

Bjork, R. A. (1999). Assessing our own competence: Heuristics and illusions. Attention and
Performance.

Bjork, R. A., & Bjork, E. L. (1992). A new theory of disuse and an old theory of stimulus
fluctuation. From learning processes to cognitive processes: Essays in honor of William
K. Estes, 2, 35-67.

Butler, A. C. (2010). Repeated testing produces superior transfer of learning relative to repeated
studying. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(5),
1118-1133. https://doi.org/10.1037/a0019902

Butler, A. C., Black-Maier, A. C., Raley, N. D., & Marsh, E. J. (2017). Retrieving and applying
knowledge to different examples promotes transfer of learning. J Exp Psychol Appl,
23(4), 433-446. https://doi.org/10.1037/xap0000142

Carpenter, S. K. (2009). Cue strength as a moderator of the testing effect: the benefits of
elaborative retrieval. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 35(6), 1563.

Carpenter, S. K. (2011). Semantic information activated during retrieval contributes to later

retention: Support for the mediator effectiveness hypothesis of the testing effect. J Exp
Psychol Learn Mem Cogn, 37(6), 1547-1552. https://doi.org/10.1037/a0024140
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 42

Carpenter, S. K. (2012). Testing enhances the transfer of learning. Current Directions in

Psychological Science, 21(5), 279-283.

Carpenter, S. K., & DeLosh, E. L. (2006). Impoverished cue support enhances subsequent
retention: Support for the elaborative retrieval explanation of the testing effect. Memory
& cognition, 34(2), 268-276.

Carpenter, S. K., Pashler, H., & Cepeda, N. J. (2009). Using tests to enhance 8th grade students'
retention of US history facts. Applied Cognitive Psychology: The Official Journal of the
Society for Applied Research in Memory and Cognition, 23(6), 760-771.

Carpenter, S. K., Pashler, H., Wixted, J. T., & Vul, E. (2008). The effects of tests on learning and
forgetting. Memory & cognition, 36(2), 438-448.

Chan, J. C., McDermott, K. B., & Roediger III, H. L. (2006). Retrieval-induced facilitation:
initially nontested material can benefit from prior testing of related material. Journal of
Experimental Psychology: General, 135(4), 553.

Cho, K. W., Neely, J. H., Brennan, M. K., Vitrano, D., & Crocco, S. (2017). Does testing
increase spontaneous mediation in learning semantically related paired associates? J Exp
Psychol Learn Mem Cogn, 43(11), 1768-1778. https://doi.org/10.1037/xlm0000414

Coppens, L. C., Verkoeijen, P. P., & Rikers, R. M. (2011). Learning Adinkra symbols: The
effect of testing. Journal of Cognitive Psychology, 23(3), 351-357.

Cull, W. L. (2000). Untangling the benefits of multiple study opportunities and repeated testing
for cued recall. Applied Cognitive Psychology: The Official Journal of the Society for
Applied Research in Memory and Cognition, 14(3), 215-235.

Dikmans, M. E., van den Broek, G. S. E., & Klatter-Folmer, J. (2020). Effects of repeated
retrieval on keyword mediator use: shifting to direct retrieval predicts better learning
outcomes. Memory, 28(7), 908-917. https://doi.org/10.1080/09658211.2020.1797094

Driskell, J. E., Willis, R. P., & Copper, C. (1992). Effect of overlearning on retention. Journal of
Applied Psychology, 77(5), 615.

Dunlosky, J., & Rawson, K. A. (2015). Practice tests, spaced practice, and successive relearning:
Tips for classroom use and for guiding students’ learning. Scholarship of Teaching and
Learning in Psychology, 1(1), 72-78. https://doi.org/10.1037/stl0000024
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 43

Estes, W. K. (1955). Statistical theory of distributional phenomena in learning. Psychological

review, 62(5), 369.

Furst, E. (2020). Successive Relearning aka Spaced Retrieval Practice. Teaching with learning in
mind.

Gass, S. (1999). Discussion: Incidental vocabulary learning. Studies in second language

acquisition, 21(2), 319-333.

Goode, M. K., Geraci, L., & Roediger, H. L. (2008). Superiority of variable to repeated practice
in transfer on anagram solution. Psychonomic bulletin & review, 15(3), 662-666.

Hartwig, M. K., & Dunlosky, J. (2012). Study strategies of college students: Are self-testing and
scheduling related to achievement? Psychonomic bulletin & review, 19(1), 126-134.
https://doi.org/10.3758/s13423-011-0181-y

Jacoby, L. L., Wahlheim, C. N., & Coane, J. H. (2010). Test-enhanced learning of natural
concepts: effects on recognition memory, classification, and metacognition. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 36(6), 1441.

Kang, S. H., McDaniel, M. A., & Pashler, H. (2011). Effects of testing on learning of functions.
Psychonomic bulletin & review, 18(5), 998-1005.

Kang, S. H., McDermott, K. B., & Roediger III, H. L. (2007). Test format and corrective
feedback modify the effect of testing on long-term retention. European Journal of
Cognitive Psychology, 19(4-5), 528-558.

Karpicke, J. D., Butler, A. C., & Iii, H. L. R. (2009). Metacognitive strategies in student
learning: Do students practise retrieval when they study on their own? Memory, 17(4),
471-479.

Karpicke, J. D., & Roediger, H. L. (2007). Repeated retrieval during learning is the key to long-
term retention. Journal of Memory and Language, 57(2), 151-162.

Kukull, W. A., Higdon, R., Bowen, J. D., McCormick, W. C., Teri, L., Schellenberg, G. D., van
Belle, G., Jolley, L., & Larson, E. B. (2002). Dementia and Alzheimer disease incidence:
a prospective cohort study. Arch Neurol, 59(11), 1737-1746.
https://doi.org/10.1001/archneur.59.11.1737
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 44

Lehman, M., & Karpicke, J. D. (2016). Elaborative retrieval: Do semantic mediators improve
memory? J Exp Psychol Learn Mem Cogn, 42(10), 1573-1591.
https://doi.org/10.1037/xlm0000267

Martin, E. (1968). Stimulus meaningfulness and paired-associate transfer: an encoding

variability hypothesis. Psychological review, 75(5), 421.

McDaniel, M. A., Agarwal, P. K., Huelser, B. J., McDermott, K. B., & Roediger, H. L. (2011).
Test-enhanced learning in a middle school science classroom: The effects of quiz
frequency and placement. Journal of Educational Psychology, 103(2), 399-414.
https://doi.org/10.1037/a0021782

McDaniel, M. A., Anderson, J. L., Derbish, M. H., & Morrisette, N. (2007). Testing the testing
effect in the classroom. European Journal of Cognitive Psychology, 19(4-5), 494-513.

Pan, S. C., & Rickard, T. C. (2017). Does Retrieval Practice Enhance Learning and Transfer
Relative to Restudy for Term-Definition Facts? Journal of Experimental Psychology:
Applied, 23(3).

Pan, S. C., & Rickard, T. C. (2018a). Does retrieval practice enhance learning and transfer
relative to restudy for term-definition facts? Journal of Experimental Psychology:
Applied, 23.

Pan, S. C., & Rickard, T. C. (2018b). Transfer of test-enhanced learning: Meta-analytic review
and synthesis. Psychological Bulletin, 144(7), 710-756.
https://doi.org/10.1037/bul0000151

Pavlik , P. I., & Anderson, J. R. (2005). Practice and forgetting effects on vocabulary memory:
An activation‐based model of the spacing effect. Cognitive Science, 29(4), 559-586.

Postman, L. (1962). Retention as a function of degree of overlearning. Science, 135(3504), 666-

667.

Pyc, M. A., & Rawson, K. A. (2009). Testing the retrieval effort hypothesis: Does greater
difficulty correctly recalling information lead to higher levels of memory? Journal of
Memory and Language, 60(4), 437-447. https://doi.org/10.1016/j.jml.2009.01.004

Pyc, M. A., & Rawson, K. A. (2010). Why testing improves memory: mediator effectiveness
hypothesis. Science, 330(6002), 335. https://doi.org/10.1126/science.1191465
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 45

Pyc, M. A., & Rawson, K. A. (2012). Why is test-restudy practice beneficial for memory? An
evaluation of the mediator shift hypothesis. J Exp Psychol Learn Mem Cogn, 38(3), 737-
746. https://doi.org/10.1037/a0026166

Rawson, K. A., Dunlosky, J., & Janes, J. L. (2020). All Good Things Must Come to an End: a
Potential Boundary Condition on the Potency of Successive Relearning. Educational
Psychology Review, 32(3), 851-871. https://doi.org/10.1007/s10648-020-09528-y

Rawson, K. A., Dunlosky, J., & Sciartelli, S. M. (2013). The Power of Successive Relearning:
Improving Performance on Course Exams and Long-Term Retention. Educational
Psychology Review, 25(4), 523-548. https://doi.org/10.1007/s10648-013-9240-4

Rawson, K. A., Vaughn, K. E., & Carpenter, S. K. (2015). Does the benefit of testing depend on
lag, and if so, why? Evaluating the elaborative retrieval hypothesis. Mem Cognit, 43(4),
619-633. https://doi.org/10.3758/s13421-014-0477-z

Rawson, K. A., Vaughn, K. E., Walsh, M., & Dunlosky, J. (2018). Investigating and explaining
the effects of successive relearning on long-term retention. Journal of Experimental
Psychology: Applied, 24(1), 57-71. https://doi.org/10.1037/xap0000146

Rivers, M. L. (2021). Metacognition about practice testing: a review of learners’ beliefs,

monitoring, and control of test-enhanced learning. Educational Psychology Review,
33(3), 823-862.

Roediger, H. L., Agarwal, P. K., McDaniel, M. A., & McDermott, K. B. (2011). Test-enhanced
learning in the classroom: long-term improvements from quizzing. J Exp Psychol Appl,
17(4), 382-395. https://doi.org/10.1037/a0026252

Roediger, H. L., III, & Karpicke, J. D. (2006a). The power of testing memory: Basic research
and implications for educational practice. Perspectives on Psychological Science, 1(3),
181-210. https://doi.org/10.1111/j.1745-6916.2006.00012.x

Roediger, H. L., III, & Karpicke, J. D. (2006b). Test-enhanced learning: Taking memory tests
improves long-term retention. Psychological Science, 17(3), 249-255.
https://doi.org/10.1111/j.1467-9280.2006.01693.x

Roediger, H. L., & Pyc, M. A. (2012). Inexpensive techniques to improve education: Applying
cognitive psychology to enhance educational practice. Journal of Applied Research in
Memory and Cognition, 1(4), 242-248. https://doi.org/10.1016/j.jarmac.2012.09.002
SPACED RETRIEVAL IMPROVES RETENTION AND TRANSFER 46

Rowland, C. A. (2014). The effect of testing versus restudy on retention: a meta-analytic review
of the testing effect. Psychol Bull, 140(6), 1432-1463. https://doi.org/10.1037/a0037559

Smith, S. M., & Handy, J. D. (2014). Effects of varied and constant environmental contexts on
acquisition and retention. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 40(6), 1582.

Smith, S. M., & Handy, J. D. (2016). The crutch of context-dependency: Effects of contextual
support and constancy on acquisition and retention. Memory, 24(8), 1134-1141.

Wahlheim, C. N., Finn, B., & Jacoby, L. L. (2012). Metacognitive judgments of repetition and
variability effects in natural concept learning: Evidence for variability neglect. Memory &
cognition, 40(5), 703-716.