The following post is what I intended to get across at the recent researchEd conference and didn’t have time to finish! The post covers some of the important issues raised by John Sweller, Paul Kirschner, John Hattie, Daniel Willingham, David Geary and others in a number of academic pieces published on human cognitive architecture and how we learn. I’ve tried to relate to the teaching of phonics some of what I think is most relevant for teaching practitioners to think about.
It is a long post, for which I apologise.The post is also a tacit plea to government to train teachers properly – something it (whether this or the last government) has signally failed to do. Learning to read and write underpins everything a child does and will do in the future, and it is vital that it is taught to a very high level of proficiency as soon as children enter school.
According to John Sweller, Emeritus professor of education at the University of New South Wales in Adelaide, instruction will only be effective to the extent that it takes into account the characteristics of human cognition (‘Human Cognitive Architecture’).
‘Ideal learning environments in accord with human cognitive architecture are not always in accord with realistic learning environments that mimic the real world’ (Sweller, ‘Human Cognitive Architecture’, p.370) or we take for granted as naturalistic human behaviour. If they did, then there would be no need for schools and universities or indeed for specific instructional procedures. Simply being in the world with an eye and an ear to what is going on around us would be sufficient.
Sweller and others contend that we have evolved to ‘assimilate and process information’ in order to direct human action. In so doing we have evolved what they refer to as a ‘natural information processing system’: human long-term memory. This repository is no longer viewed as a jumble of unrelated facts we store and retrieve from time to time. It is now seen as absolutely central to the structure of human cognitive architecture. In other words, it is central to how we learn.
So, what is it that differentiates experts and novices in solving particular classes of problems, by which I mean domain-specific (mathematics, chess, writing systems, etc.) problems? The answer appears to be simply the amount of knowledge or information held in long-term memory. To be skilful in any domain, it is necessary to have at one’s disposal a huge stock of knowledge that is instantly available, a point also made by E.D. Hirsch on many occasions.
As Sweller et al contend, we have evolved cognitive systems to impart and receive knowledge from others around us. From our first moments we begin to imitate what others do. We are primed for naturalistic human behaviour, such as learning our own language or languages, recognising faces, assimilating our own specific cultural practices, and spotting predators and prey. These we learn pretty well without explicit instruction though that isn’t to say we can’t improve these abilities with further guided instruction and practice.
However, what we, as educators, are concerned with is what Sweller and his associates refer to as secondary learning, or ‘biologically secondary knowledge’. In this respect the essential task for educators lies in devising efficient instructional procedures for transmitting knowledge to the long-term memories of learners.
All efficient procedures for the transmission of knowledge are dependent on what Sweller calls the ‘borrowing principle’ or probably what we think of more generally as learning from presented information or direct instruction. The alternative is leaving the learner to work things out for themselves, or, as Sweller calls it, ‘random generation followed by effectiveness testing’, which is really another way of saying trial and error approach. Trial and error approaches are highly inefficient for a number of reasons, not least of which are that they are extremely time-consuming and, perhaps more importantly, they run the risk of pupils misunderstanding important concepts or being misled by surface impressions (classifying whales as fish because they live in the sea, for example). As Michelene Chi points out in “Two Approaches to the Study of Experts’ Characteristics” (in Ericsson et al, p.23), experts are more able to ‘perceive the “deep structure” of a problem or situation’ because they are more knowledgeable. So, for a number of reasons, the research evidence seems to indicate that, at least in the early stages of learning a new domain, direct instruction is much more effective.
Now apply this to the situation of teaching reading and writing to young children in school. We have one of the most complex writing systems, if not the most complex, and we know that the long tail of underachievement in learning to read and write is enormous: as much as 50% if we rely on figures given by the OECD (Canada) in 1997. So, given the prodigious task of learning to read and write to a very high level of mastery – one which is vital to an individual if they are to maximise whatever potential for learning they have – why would we expect a young child to work out for him/herself how our complex writing system works? As Sweller confirms, ‘The absence of explicit instruction that works perfectly in the case of biologically primary knowledge is likely to fail abysmally when dealing with secondary knowledge’ (Sweller, J., ‘Human Cognitive Architecture and Constructivism’ in Constructivist Instruction: Success or Failure, (2009), p.130.
In my long experience of training teaching practitioners, I have found that very few have a clear and unambiguous idea of how the alphabetic writing system in English relates to the sounds of the language.
According to one of the world’s experts on writing systems Peter Daniels, ‘writing is defined as a system of more or less permanent marks used to represent an utterance in such a way that it can be recovered more or less exactly without the intervention of the utterer.’ (Daniels, P.T, Bright, W., (1996), ‘The Study of Writing Systems’ in The World’s Writing Systems, London, OUP, p.3) A writing system must represent the sounds of the language and for that you need a graphic symbol inventory. The decisive step in the development of writing was phoneticisation of graphic representation. This came with the realisation that if you worked out how many sounds there are in a language and you invented a series of squiggles to represent those sounds, you would have an entirely accurate means of recording anything. Writing, then, is a cognitive tool, a huge systemic mnemonic, if you like.
As Daniels is also at pains to point out, writing is different from spoken language ‘in a very fundamental way. Language is a natural product of the human mind … while writing is a deliberate product of the human intellect: no infant illiterate absorbs its script along with its language; writing must be studied.’ (Ibid., p.2)
To make such a system work someone has to set about working out what the sounds of a language are and inventing symbols to represent them. Since its invention in Mesopotamia, it has passed from the Phoenicians to the Greeks, to the Etruscans and on to the Romans. The Romans brought it to the British Isles, since which time it has undergone many changes. Nevertheless, regardless of the changes to the language over time, it remains the case still that writing, no matter its complexity, represents the sounds of the language.
When we present young children (or anyone for that matter) with the written material that makes up our writing system, this novel information for which they have little, incorrect or no prior knowledge, their working memories are very limited. What we must do therefore is to present them with a phonics approach that builds a schema for sounds and individual spellings and teaches them from simple to ever more complex, turning that schema into an ever more sophisticated tool.
Schemas enable us to organise and store knowledge. They categorise the elements of information you want to teach according to the way in which you want them to be used. They also reduce working memory load because, as K Anders Ericsson (Ericsson et al: The Cambridge Handbook of Expertise and Expert Performance) and his colleagues have long argued, even a highly complex schema can be dealt with as if it is a unitary element in working memory.
Because schemas can incorporate facts, functions, procedures and entities specific to a domain, the building of schemas is a long process and there are no shortcuts. I was once told by a disdainful head teacher that phonics was a quick fix! If only that were the case! All the literature, tells us the same story: development of individual performance in any domain is ‘relatively slow and graduated even when a large amount of time is invested’ (Hattie, 2014, p.95). Growth, as we always tell people on our courses, is uneven and is punctuated by sudden improvements, long plateaus, and/or periods of regression. That’s what learning looks like. Messy!
To continue to make improvement in any domain, you need to devote time to deliberate practice, which is ‘mindful, sequential and highly structured’ (Hattie, 2014, p.96).
Of course, you also need expert tuition, persistence and goal setting. I don’t know whether you listened to Professor John Hattie talking on Radio 4 (http://www.bbc.co.uk/programmes/b04dmxwl) a few weeks ago but, aside from telling us what doesn’t work, he highlighted a few things that do. Chief among these was expertise in teaching… and to get that, you need to train the teachers.
In the beginning at least, confirms Hattie (Hattie, 2014, p.96), the focus should be on short-term, immediate goals, where the teacher concentrates attention on critical aspects of practice, helps to refine the procedural skills of blending, segmenting and phoneme manipulation, provides time and space for repetition, and offers corrective feedback where necessary.
One of the most critical aspects of successful teaching is practice. The more anyone practises an activity, the more able they are to switch from a conscious to an automatic function. Automaticity frees working memory capacity for other activities because, it is argued by the proponents of Cognitive Load Theory, a schema that has been made automatic acts as a central executive that directs activity without the need for processing in working memory. This is what enables people to perform an activity without even thinking about it: the activity dips below the level of conscious attention. In one way, this is very useful because it enables the performer, in Hattie’s words, to ‘peg a skill at a given level’.
Furthermore, as Fletovich, Prietula and Ericsson point out, ‘[a]utomaticity is important to expertise. It appears to have at least two main functions. The first has to do with the relationship between fundamental and higher-order cognitive skills, and the second has to do with the interaction between automaticity of processes and usability of available knowledge. With regard to the first, in complex skills with many different cognitive components, it appears that some of the more basic ones (e.g., fundamental decoding, encoding of input) must be automated if higher-level skills such as reasoning, comprehension, inference, monitoring, and integration are ever to be proficient.’ Fletovich, P.J., Prietula, M.J. & and K. Anders Ericsson, in ‘Studies of Expertise from Psychological Perspectives’, in K. Anders Ericsson et al, Eds, (2006), The Cambridge Handbook of Expertise and Expert Performance, London, CUP.
In terms of reading then, this is what enables us to decode text and understand it; in terms of writing, it is what enables us to think about what we want to write while actually writing it. The two processes can operate in parallel simultaneously.
The paradoxical thing about this is that simply spending time honing a skill you have already acquired doesn’t automatically enhance your actual level of performance in that skill. People tend to learn a complex skill to the extent that they are satisfied with its functionality and then they don’t usually bother to burnish it. They only want, as in, say, the case of golf, to be good enough to keep up with their friends, or, as with driving, to negotiate the particular driving conditions they are routinely faced with.
So, if we want learners in a domain to continue to improve, we need to present them with further challenges. This, in the case of learning a musical instrument may mean finding a more highly skilled teacher. In the case of teaching children to read and write, it means training teachers to a high level of proficiency.
Novel information (squiggles on a page combining to represent sounds in words) coming through the senses is unorganised and random and often imposes on working memory too heavy a cognitive load. Sweller maintains that ‘anything beyond the simplest cognitive activities appear to overwhelm working memory’. So, our problem as teachers of reading and spelling is one of how to present the information necessary for young children (or even older learners who are illiterate or semi-literate) in such a way as not to overload working memory and to enable successful transfer of information to long-term memory, where information is organised and structured. Throwing all the complexities of the English alphabet code at them in the form of whole words that they are asked to remember as wholes is an impossible task.
To help young children learn, we need, in the first instance, to restrict the amount of new information we present them with. In essence, this is one of the basic tenets of cognitive load theory. As Hattie says, ‘Novices need to concentrate as deeply as they can on specific ideas without encumbrance from other sources’ (Hattie, 2014, p.150). In other words, we need to take things one step at a time and do our best not to distract learners with unnecessary information, and distractions, such as the ‘all singing, all dancing’ features that often go alongside phonics activities: what Sweller and van Merriënboer call ‘extraneous cognitive load’, or load that is unnecessary for learning.
Hattie describes working memory as ‘the workbench of the conscious mind’ and as a ‘bottleneck to our ability to learn’. It can only deal with a very limited amount of information at any one time. While this would appear to be an impediment to learning, in fact it safeguards long-term memory from being deluged with too much unstructured information pouring in at any one time
So, first, the question remains of how we can facilitate the loading knowledge from working memory into the long-term memory system. At the same time, we need to ensure that the material being loaded into the system is coherently structured so that new information can be processed and integrated in a way that is commensurate with what has already been learnt. For example, after teaching all the one-to-one sound spelling correspondences, it is very easy to teach the double consonants, ff, ll, ss and zz (because they look very similar to the single letter spellings), while making explicit that we can spell sounds with more than one letter. The third problem is one of retrieval: how to enable ease of access to the store?
Because teachers of good quality phonics programmes know how the alphabet system works in relation the sounds of the language, they should be the arbiters of what we deem important enough to transfer into long-term memory. How can we expedite this? The answer is simple: we create a schema in which the material we want young children to learn from the start is from simple to more complex. For this reason, we need to introduce a limited amount of information at any one time and to recycle information that has already been learned, by which I mean committed to long-term memory.
Thus we can begin by presenting simple information in cumulative steps in the form of a limited number of ‘templates’ or lessons or instructional procedures that are unchanging. If the way in which lessons are presented is unchanging and provided that they are an effective means of conveying what it is we want to teach, once established and practised, cognitive load is reduced to input of the new material.
If done properly, the learner can now devote all their attention to what Sweller et al call ‘intrinsic cognitive load’ or only what the learner needs to learn to achieve the desired outcome. At the same time, ‘extrinsic cognitive load’ or all the stuff that the intrinsic cognitive load comes wrapped up in (instructional design, the practice that goes into creating automaticity) is reduced to a minimum.
As I’ve already made clear, working memory is limited. There’s general agreement that it can handle only about seven bits of information at any one time. We also know that it can actually only operate on between two and four new elements at a time and that, unless rehearsed constantly even those elements are likely to be lost within as short a period of time as twenty seconds.
These capacity limits only apply to the input of new knowledge because we also know that working memory has no known limits when operating on information stored in long-term memory. What this means is that whatever is stored in long-term memory has the capacity to alter dramatically what is happening in working memory.
So, how we develop expertise in reading and writing depends on how well we design instructional materials and build a comprehensive schema, which is additive, open to complexity and automation. When posed in this way, teaching reading and spelling is no different qualitatively from teaching mathematics or other knowledge domains.
Daniels, P.T. & Bright, W., (1996), ‘The Study of Writing Systems’ in The World’s Writing Systems, London, OUP.
Ericsson, K.A., Charness, N., Feltovich, P.J. & Hoffman, R.R., (2006), The Cambridge Handbook of Expertise and Expert Performance, London, CUP.
Hattie, J & Yates, G., (2014), Visible learning and the Science of How We Learn, London, Routledge.
Geary, D.C, ‘Educating the Evolved Mind: Conceptual Foundations for an Evolutionary Psychology’, https://www.google.co.uk/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=David+Geary%2C+%E2%80%98Educating+the+Evolved+Mind%3A+Conceptual+Foundations+for+an+Evolutionary+Psychology%E2%80%99
Van Meriënboer, J. & Sweller, J., ‘Cognitive Load Theory and Complex Learning: Recent Developments and Future Directions’, in Educational Psychology Review, Vol 17, No 2, June 2005 (p.150)
Sweller, J., ‘Human Cognitive Architecture’, http://thesedominiquebellec.fr/Vrac%20articles/Human%20Cognitive%20Architecture.pdf
Sweller, J., (2009), ‘Human Cognitive Architecture and Constructivism’ in Constructivist Instruction: Success or Failure, London, Routledge.