SR-EAL, the forgetting curve and the Spacing Effect! No it's not a neuromyth

Not so long ago 'Learning Styles' occupied a secure place in any list of EAL 'buzz words.' Fortunately, times have changed and many of us now recognize that despite the hype the notion of learning styles has proved unhelpful and misleading. Certain academics might deny all this, but teachers seem more reluctant to acknowledge learning styles in their teaching and lesson plans. This is all to the good. Skepticism makes sense if we agree that credible research should inform classroom practice. My blog on learning styles did rather get me thinking, however: What other 'neuro' myths are floating around to entrap the unwary? How do you feel about about Brain Gym, for example, Brain Based learning, Performance Based Learning, or Maketon and even Numicon? How well do these stand up to scrutiny now the dust has settled? Someone has to ask. We can all benefit from stepping back to reflect on our beliefs if we aim to improve as teachers. Whether we should take that 'awkward' step of speaking out about some dubious educational practice isn't quite so easy to answer. Who wants to tell Mrs. Forthright that the pet project for Primary (and apparently trialed ... Ahem) is pseudo-scientific claptrap? No one takes kindly to being told they're doing it all wrong! I'm no exception.

While mulling over how to get points across without becoming a staffroom pariah it dawned on me that I was going about it all the wrong way. If you want to challenge the orthodoxy, or question some classroom practice an all-out frontal assault might make sense at times. But perhaps better still, one might just turn the whole myths topic on its head. Rather than argue that some practice doesn't work, how about encouraging peers to adopt those overlooked educational practices that indeed have real merit? No bruised egos if we take that path. It didn't take long before I had dug up a true gem of real pedagogical value. Today psychologists call it the spacing effect. And it really is worth looking into.

The spacing effect.

The term 'spacing effect' dates back to the pioneering experiments of Hermann Ebbinghaus in the 1890s. Ebbinghaus's somewhat singular and all-absorbing passion lay in uncovering the mysteries of memory and learning and, in particular, how building secure long term memories might depend upon the total time and number of learning sessions over which learning efforts occurred. Before Ebbinghaus examined the topic no one really knew much more than a few simple and uncontroversial truths: For example, that the more time expended on studying, the more we learn and that learning clearly called for some type of periodic reinforcement to minimize forgetting. What no one understood at the time, and what seized Ebbinghaus's attention, was how for any total study time learning gains might differ if you interspersed the studying period with non-study intervals: Would student 'X' learn as much, more, or less from, say, one uninterrupted hour-long study session as s/he would from the same study time but interrupted with 'short' on-study breaks? This is what Ebbinghaus set out to determine. Of course, this project raised other issues, too. If you indeed learnt more, what about retention of that knowledge over time?

Starting way back in the mid-1880s Ebbinghaus embarked on a dedicated program of self-experimentation. He began by compiling a list of 2,300 'three letter' CVC syllables (Bok, Qud). No consonant could appear twice in the same syllable -ruling such constructions such as GoG or QaQ. Nor could the syllables correspond to real words (In fact, Ebbinghaus used the term 'nonsense syllables' to describe his 'trigram' creations). From his main list, Ebbinghaus then set about compiling various sub-lists that he had decided he would commit himself to learning. Over the next few years Ebbinghaus periodically selected a list and then learnt it until he could recite it correctly two times, with each syllable in its correct order. He recorded his total learning time and would then repeat the experiment at a later occasion -i.e. either 20 minutes after the initial learning, 1 hour, 1 day, 2 days, 6 days, or 31 days after the previous study session. By controlling the time intervals in this way Ebbinghaus could determine how learning depended upon total study time and the duration and number of no-study intervals for the list in question. When he plotted his findings Ebbinghaus realized he had created a revealing representation of memory losses over time. The curve that he uncovered -and the one that you will find in most psychology books- not surprisingly became known as the forgetting curve. More importantly, perhaps, Ebbinghaus clearly demonstrated that for any total study time not only did more learning arise from relatively short study sessions with intervals between them, but also that by optimizing the number and duration of intervals the time savings (for x amount of study time) could be substantial. The phenomenon became known as the spacing effect.

The Forgetting Curve (Memory: A Contribution to Experimental Psychology. Hermann Ebbinghaus (1885)

Despite its obvious relevance to those working in education, few seemed willing to incorporate Ebbinghaus's findings into lessons or instructional materials, however. Some 35 years ago, and nearly 100 years after Ebbinghaus had published his findings, Dempster (1988) wrote of the The spacing effect as a 'case study in the failure to apply the result of psychological research.' That was then, of course, and a lot can change over several decades you might think. Yet even today instances of teachers applying the spacing effect in classroom settings remains uncommon. Antero Garcia (2022) sees the spacing effect as still very much underutilized by educators in general, as do Carl Hedrick and Robin Macpherson (2017).

The evidence.

But what about credibility? Does the evidence really point to spaced learning efficacy in classroom settings, and does it indeed lead to durable gains? If the answers are 'yes' then can we generalize across all grade levels and student ages? According to Seabrook et al (2005), in a study involving 119 participants ranging from 5 years old to adults, the spacing effect indeed ensured more learning irrespective of the study participants age. Good news for teachers, so it seemed. The study did note a methodological problem, however. As in most research the comparisons involved spaced and massed learning, but as Seabrook et al point out, "Teachers already distribute material: Learning to read, for example, is distributed over many years and even much smaller topics will be covered over a week or more. " The authors went on to describe a new form of learning that they labelled 'clustered' and which better captured the typical classroom setting. They defined clustered learning as teaching that lies intermediate between massed presentations and many short presentations (i.e. learning opportunities); the term was, that is, a useful shorthand for ‘less distributed than the other schedule in the experiment.’ Because clustered learning typifies the typical classroom experience for most children, it offered a more pedagogically useful base against which to compare distributed learning outcomes than the massed learning opportunities that previous research had typically focused upon for comparison purposes.

So, which type of learning condition -distributed or clustered- proved most effective? For the experiment the authors recruited adults (16 students from Warwick University), along with twenty children from Year 2 with an average age of 6 years and ten months. Using the same digit recall task employed in their earlier experiment (above). The study found that clustered presentations (the school 'norm,' remember) produced recall performance no better than that following massed presentations. In fact, "Distributed presentations resulted in significantly better recall than seen in either of the other two conditions" with this holding true for both children and adults." Moreover, no significant difference appeared in the "pattern of results between the two age groups" (p.116). The authors draw attention to the potential practical implication, describing the then 'literacy hour' program common in schools as an example of just the clustered teaching that compared so unfavorably to distributed teaching practices.

This still left open whether the findings carried across to the rough and tumble of real classrooms as opposed to the laboratory like conditions under which the authors had conducted the experiments thus far. To compare distributed with clustered learning in authentic school settings Seabrook et al. (2005) recruited 34 year 1 children from two different schools. One set of children received phonics instruction (they learnt grapheme sound associations) divided up into three two minute sessions per day, while the another group received a single daily 6 minute instruction of the same sounds/phonemes. The experiment continued for two weeks after which all children then underwent tests of single letter phoneme correspondences, multiple letter sound phoneme correspondences and ability to read phonically regular words (e.g. sit, bend, stand).

So, what did the authors learn from all this? First, the scores for the various skills examined tended to correlate (Pearson's r) highly with p-values ranging from 0.84 to 0.73. As for the phonics scores those children exposed to distributed teaching showed 'significantly more improvement than those who had received clustered teaching sessions (8.3 vs. 1.3 points improvement; t(22) ¼ 3.05, p < 0.01).' It seemed clear that laboratory type testing (experimnt1), could indeed furnish useful insights into what 'works' in an authentic classroom setting. What's more, the gains Seabrook et al observed proved far from trivial. 'Over 2 weeks, children whose teaching consisted of three 2-min sessions per day showed more than six times the improvement of those who were taught for one 6-min session per day.' It's hardly surprising, then, that the authors pushed for more spaced learning in place of the (then) Literacy Hour. After all, imagine if we could translate results like that over the course of an academic year. We would hit the jackpot!

But what about older children -those in secondary, for example? And what about skills other than phonics? Just how ubiquitous is the spacing effect in education more generally? In the field of vocabulary acquisition, Goossens et al (2012) investigated uptake of new word meanings among grade three children. The experiment involved 33 students from two classes, each of whom received instruction in the meanings of 30 words that the researchers had divided into two separate lists of 15. Each child learnt the words from one list under massed conditions and the words form the other under distributed conditions. To counterbalance the design, those words that a group studied under massed conditions were those learnt by the other under distributed conditions, and vice versa. The total time children spent learning their lists was identical as was the teaching method. The latter involved teaching the same 15 words (1 list) over three learning sessions (distributed learning) and for the mass learning 5 words in one session, a different five from the same list in the next session and a final five words in the last session. The experiment took place over the course of four consecutive days.

After completing their sessions children undertook assessments at one week and five weeks post study. The results on both occasions were essentially the same -children displayed higher word retention scores for words studied under the spaced condition as we see from table 1.

All this only confirmed what Ebbinghaus had predicted more than a century earlier. But how impressive really are the results? It kind of depends on where you're coming from and your expectations. From my own (somewhat dismal) attempts at language learning, five weeks seems ample enough time to forget any new words unless I have had regular encounters with them in the intervening period. I would, however, suggest that gains of the order Goossens (2012) et al. report are far from trivial. More research into memory durability certainly seems warranted. That said, reports of associations between spaced learning and long term gains do continue to make appearances in the academic literature. A recent(ish) example is Lotfolahi et al (2017) who had 2-5 graders (average age of 9.17 years) learn word-pairs having children give each other corrective feedback on tests of words they had studied. At both one week, and five weeks, after the last study session children retained statistically significantly more words from their spaced learning experiences. The authors conclude that spaced learning offers a 'pedagogically powerful' means of developing vocabulary.

Despite a bias that has seen far more experiments in laboratory settings rather than classrooms (section 1.2), several studies report school based findings. Reynolds and Glaser (1964) claimed time savings from a secondary level biology course incorporating distributed learning tasks. Rea and Modigliani (1987) noted impressive results among primary school children from a spaced spelling and multiplication program while Bloom and Shuell (1981) recorded a substantial 35% increase in word gains from distributed learning among high school students studying French as an L2. And it's not just from immediate post tests that gains reveal themselves. Knowledge retention, too, appears more durable given spaced learning opportunities. Sobel, Capeda, and Kapler (2011), reported superior vocabulary retention among 39 Grade 5 pupils from spaced instruction (this involved presenting slides, oral practice along with paper and pencil assessments) from tests conducted one week after the last tuition session. The authors concluded that optimal intervals between study periods facilitates vocabulary uptake “in applied settings” with “middle-school-aged children” (p.763).

Perhaps what really makes the spacing effect so interesting is its pervasiveness. You don't observe it only in traditional school based settings but in almost any situation where the need to learn arises. According to Carpenter et al 2012, the spacing effect has proved one of the oldest and most reliable findings in research on human learning. It applies to implicit as well as explicit learning, among all age groups and across a broad range of learning contexts including, among others, video game mastery (Donoval & Radosevich, 1999), mirror tracing, paired associate learning and paragraph recall (Janiszewski, Noel, & Sawyer, 2003).

Final comments

The essence of SR-EAL lies in adopting our teaching to particular student needs which, in turn, obliges us to apply well established research findings to our teaching practices. Something as effective as spaced learning seems far too good to miss out on. Both laboratory experiments and classroom based research show non-trivial gains from spaced learning programs; we surely need to bear this in mind. But what would this look like in your average EAL lesson? Most obviously, it suggests far more review than most of us are accustomed to; it suggests constantly returning to spelling, grammar points and vocabulary rather than ticking of check lists of 'done' work. It also implies that we should alert students to the value of revision and educate them, as well as colleagues, of just what spaced learning has to offer. As Emeny, Hartwig and Rohrer (2021) remind us,  teachers are still not fully integrating spaced learning opportunities into their teaching methods.