Follow me on Twitter for more articles like this: @mj_bromley
Fast mapping is a process by which a new concept is learned – based only on a single exposure to it – alongside a familiar concept.
Allow me to illustrate… let’s assume a child has never seen or heard of an elephant before (our new concept) but that the child is well-acquainted with a goldfish (our familiar concept). You present the child with two pictures – one image of a goldfish and another image of an elephant. You then ask the child a question which refers to the elephant by name and includes details which preclude the goldfish from being the correct answer, such as ‘Does the elephant have big ears?’
By a process of elimination, the child is likely to conclude that the unfamiliar animal is an elephant because fish do not have ears whilst this unfamiliar creature does. The child therefore answers the question correctly and, in so doing, comes to learn a new concept which he/she now knows is called an elephant.
Retrieving information about and ruling out the goldfish allows the child to infer that the new word ‘elephant’ refers to the new concept in the picture.
It is important that both concepts are related – in this case, that they are both natural, living entities.
Fast mapping is thought to bypass the brain’s hippocampal system by rapidly incorporating new concepts into the cortical memory networks (hence the term ‘fast mapping’). As such, fast mapping is not only thought to help children learn new vocabulary very quickly (research including Bloom & Markson (1998) and Carey & Bartlett (1978) suggests that fast mapping can help two-year-olds learn new nouns quickly during a single exposure) but it may also help adults with hippocampal damage to learn new words for unknown items.
Let’s leave to one side the possible advantages that fast mapping may have for people with brain damage and focus instead on its potential use with young people in the classroom…
If fast mapping can lead to rapid vocabulary learning in infancy (and later childhood) then it may be worth experimenting with more widely in our schools…
First, let’s take a step back: the alternative to fast mapping (or FM) is a process called ‘explicit encoding’ (or EE). EE, to extend the example above, involves showing a child a picture of an elephant with the word ‘elephant’ written beneath it and asking the child to learn this new concept in isolation (i.e. without any reference to a familiar concept).
In experiments, researchers (e.g. Coutanche & Thompson-Schill, 2014) tested children using both the EE and FM methods. They tested them for:
Declarative memory – the ability to recall and recognise the new concept (in the example, above, children might be shown a picture of an elephant and later asked to name it).
Lexical integration – the ability to process new concepts and apply it to prior knowledge (in the example above, children might be asked to say whether an elephant was man-made or natural). Also wrapped up within lexical integration is the child’s processing speed.
Semantic integration – the ability to make associations between the new concept and similar words such as the names of other animals, versus made-up or unrelated words such as elephone or chair.
The researchers found that children who learned new names through EE had a better declarative memory immediately after the tests (in other words, those children who’d learn new concepts in isolation were quicker at recalling and recognising those new concepts within about ten seconds of first learning them) than those children who learned new names through FM. Indeed, EE children performed better in both recall and recognition compared to FM.
In contrast, FM children performed better when it came to lexical integration. In fact, children who learned new names via EE did not show any evidence of lexical integration soon after testing whereas FM children showed evidence of lexical integration ten minutes after testing and again the following day.
Again, when it came to semantic integration, FM children out-performed EE children. FM children demonstrated semantic priming on the first day and the second day whereas EE children did not.
In conclusion, research shows that learning new vocabulary by fast mapping (FM) – learning new words alongside familiar words as opposed to explicit encoding (EE) which is learning new words in isolation – promotes the rapid integration of those new words into memory networks. Although EE gives superior declarative memory, it does not lead to lexical integration or semantic integration. In other words, it does not lead to deep and meaningful learning. Words are learned but not integrated: learning new words through EE does not enable children to integrate those words into their existing memory networks. FM, meanwhile, enables this to happen and to occur rapidly.
A word of caution: because FM weakens a child’s declarative knowledge, the rapid integration that it does afford may be achieved at the cost of reduced on-demand memory access. In other words, if children learn new words via FM they may not be as good at recalling and recognising the words as they would had they learned it via EE.
Earlier, I said that I would leave to one side the possible advantages of FM for patients with brain damage but I can’t help but explore this concept further…
The fact that, through FM, you have to retrieve a familiar concept and engage in inference (relating it to the new concept, making decisions relating the both concepts) plays a crucial role in subsequent rapid lexical integration. This, alongside engaging the semantic memory (accessing memory representations of a familiar concept), means that we engage in increased neural activity when using FM than we do when using EE. Accessing neural representations of a familiar concept enables the new concept to take its place within the active neural network.
Research into the use of schema in learning already tells us that using existing relevant knowledge to help us acquire new knowledge (asking ourselves questions such as ‘How does this new concept relate to what I already know?’) can make learning extremely rapid and longer lasting.
In patients with hippocampal damage (for example, people with memory loss following a motorbike accident who can recall their childhood but are unable to acquire any new memories), instances of successful semantic learning are almost all confined to those patients who acquire new concepts that are related to existing prior knowledge. New information, it seems, can be anchored to existing knowledge even for those patients with severe hippocampal damage because the presence of a known item can activate a relevant schema which in turn facilitates the integration of the new item into semantic memory.
So what does all this mean for teachers?
In some ways, it reinforces what we already know about teaching knowledge before skill. After all, it is by applying prior knowledge – say about the goldfish – that children can learn new knowledge – say about an elephant – in a way that proves more meaningful and lasting.
It also reinforces what we already know about the importance of deliberate difficulties. For example – and sticking with our example – it is because the acquisition of the new concept of an elephant is made initially more difficult – it is harder to encode because it is introduced alongside another concept so we have to work harder at it, holding more information in our working memories – that it proves to be easier to retrieve later. We have seen that using EE – making something easier to encode initially – hinders children’s abilities to retrieve the information later. Learning with FM is harder to begin with (so declarative memory is weaker within ten minutes) but easier to retrieve later (so lexical integration and semantic integration is stronger after one and two days).
What I think is potentially new here is that, as well as making learning stronger over time, accessing prior knowledge and slowing down learning also makes learning deeper…
In other words, as well as remembering something for longer, children are better able to apply that new learning to different situations – the learning becomes deeper because it sits within a web of other ideas and concepts rather than in isolation.
This makes it easier to apply new knowledge to a range of different contexts which is surely what we need our students to do in order to be successful in tests and in life. Knowing that big grey thing with a long trunk is called an elephant is only useful up to a point. Being able to analyse that concept in relation to other concepts – knowing, for example, that an elephant is natural not man-made, that it is a mammal, and that there are two varieties, Africa and Asian, which can be discerned by the size of their ears (African elephants have large ears – shaped much like the continent of Africa itself – and the larger surface area of their ears helps to keep them cool in the African sun; Asian elephants tend to live in cool jungle areas so their ears are smaller. Asian and African elephants have very distinct head shapes, too – African elephants have fuller, more rounded heads, and the top of their head is a single dome; Asian elephants have a twin domed head with an indent in the middle. Only male Asian elephants grow tusks and even then not all males will have them; in African elephants, both sexes generally but not always have tusks) is much more useful.
A word of caution to conclude… using semantic cues such as “It lives in Africa” or phonological cues such as “It starts with /e” might impair learning for some children because it floods the working memory. Research by Graya & Brinkley (2011), for example, found that that phonotactic probability and previous lexical knowledge affected children’s ability to learn new words. Phonological or semantic encoding cues, therefore, might not be beneficial for fast mapping.
As we have seen, research has found that, generally, children retain a word they’ve learned through fast mapping for a significant period of time. Carey and Bartlett (1978) found that children were able to keep a new lexical entry in their working memories for several days, illustrating a process of gradual lexical alignment known as “extended mapping”. Markson and Bloom (1997) showed that children could remember a new word for up to a month.
However, more recent studies have shown that words learned through fast mapping tend to be forgotten over time. In a study conducted by Vlach and Sandhofer (2012), for example, memory supports (included in previous studies) were removed and this resulted in a low retention of words over time. Some researchers have also expressed concerns that experiments which test the effectiveness of fast mapping are produced in artificial settings. They believe that testing for fast mapping should focus more on the actual understanding of a word rather than simply its reproduction. Testing to see if a child can use a new word in a different situation constitutes true knowledge of the word.
There’s also evidence to suggest that fast mapping doesn’t enable children to learn words. Instead, they learn the probabilistic, predictive relationships between objects and sounds that develop over time.