Minspeak
Sometimes, to go forward, you have to go back. A fair number of AAC practitioners claim they know all about Minspeak and on the basis of their knowledge have determined that it is too
complicated. But how many people can actually give a text book definition of Minspeak? Aside from the fact that there isn’t an actual text book on the subject; not even a chapter in a text book – few people have ever come across a good, succinct, one-line definition that encapsulated what it means.
Minspeak is the systematic exploitation of secondary iconicity to provide large vocabularies within a small icon space.
You lost me at “systematic.” This looks like the sort of definition that an ivory-tower researcher might invent for a grant proposal, but it means nothing to me.
Sorry, but I did say we would have to go back in order to go forward. And that includes taking the basic definition and tearing it apart. But don’t worry, we’ll go slowly and by the end of this page you’ll find the definition as easy as saying 'TalkSense'.
Let’s start with the key word in the definition, and that’s 'iconicity'. This is a term bandied about frequently in AAC and refers to how pictures are used to represent words or ideas. Here’s a good working explanation:
Iconicity refers to how apparent the relationship between a symbol and its meaning (gloss) is to learners. Iconicity may be conceptualized as a continuum with transparency and opaqueness as the poles. Transparent symbols are readily guessable while the relationship between the symbol and the gloss of opaque words is not understandable even when both are provided the learner.
OK, so it’s written for an academic journal and might need some explaining to understand it fully. Maybe some examples will help. Let’s start with a picture that might best be a reasonable example of transparent i.e. obvious.
Sorry, but I did say we would have to go back in order to go forward. And that includes taking the basic definition and tearing it apart. But don’t worry, we’ll go slowly and by the end of this page you’ll find the definition as easy as saying 'TalkSense'.
Let’s start with the key word in the definition, and that’s 'iconicity'. This is a term bandied about frequently in AAC and refers to how pictures are used to represent words or ideas. Here’s a good working explanation:
Iconicity refers to how apparent the relationship between a symbol and its meaning (gloss) is to learners. Iconicity may be conceptualized as a continuum with transparency and opaqueness as the poles. Transparent symbols are readily guessable while the relationship between the symbol and the gloss of opaque words is not understandable even when both are provided the learner.
OK, so it’s written for an academic journal and might need some explaining to understand it fully. Maybe some examples will help. Let’s start with a picture that might best be a reasonable example of transparent i.e. obvious.
If we use this picture to represent the word apple, that would seem pretty logical and straightforward. Now let’s look at an example that might be deemed opaque i.e. you have to be TOLD what it is to have any hope of understanding what it means.
In the PCS picture set, this is used to represent the word easy. And that seems hard, yes? To understand that it means easy, you have to know that most people consider adding one and one to be easy. This is an example of something called metaphor. Briefly, a metaphor is a way of talking about one thing by comparing it to something else. We’ll talk more about metaphor later, but for now, it’s an example of a picture that is opaque. Of course, if we wanted to use the picture of the apple to represent something else, such as fruit, or round, or eat, then we are free to do so. It just shifts the transparency a little to the opaque side. Here’s a little graphic that might help.
apple fruit round hemisphere France
TRANSPARENT <------------------------------------------------------------------------------------------------------------------------------------------------> OPAQUE
So when you think about it – and I hope you are thinking about it – whether an icon is transparent or opaque has nothing to do with the image in itself but how well or not it represents what it is supposed to mean. That’s a tricky sentence so read it again just to make sure you grasp the importance of that.
So if iconicity is related to how well a picture represents what it stands for, and the more transparent an icon, the better it represents a meaning, then shouldn’t we be designing picture sets that are transparent?
You’d think so, wouldn’t you? After all, why use the picture of the apple for France when you could just go ahead and use a picture of the flag of France? And for the word hemisphere,how about using a picture of a semi-circle? Well, we could! But then not all words are as easy. How would you draw ‘this’, ‘should’, ‘the’, ‘in’, ‘across’, ‘it’, and ‘would’ (to name but a few) such that they were immediately transparent? We could draw a coin being placed in a piggy bank to represent the concept of in (for example) but wouldn’t ‘save’, ‘pig’, ‘money’, and even ‘pink’ be valid interpretations of this symbol? Why, even the apple picture could have multiple possible forms:
But more challenging is not just that one word such as apple can be represented by many different pictures, but one picture such as the apple can represent many other words. Or more accurately, can have more than one meaning assigned to it. And that’s where the notion of secondary iconicity comes in.
Ah yes, the Minspeak definition again: “…exploitation of secondary iconicity…” So how about explaining the secondary piece?
If you are now feeling comfortable with the idea of iconicity, there’s just one other thing you need to add to your memory, and that’s the difference between primary and secondary iconicity. Simply put, primary iconicity is the most common association that is made with regard to a particular image. Primary iconicity is really a statistical phenomenon. As an example, take a look at the picture below and think about what words come to mind when you look at it.
Ah yes, the Minspeak definition again: “…exploitation of secondary iconicity…” So how about explaining the secondary piece?
If you are now feeling comfortable with the idea of iconicity, there’s just one other thing you need to add to your memory, and that’s the difference between primary and secondary iconicity. Simply put, primary iconicity is the most common association that is made with regard to a particular image. Primary iconicity is really a statistical phenomenon. As an example, take a look at the picture below and think about what words come to mind when you look at it.
If you show this image to one hundred people and ask them to write a list of all the words that come to mind when they see it, around 45% put sun as their first choice. This is the primary iconic association for this picture. Other “first-words” include happy, hot, and smiling. These words would all be called secondary iconic associations. And what is really, really critical to remember is that ALL pictures will have a primary value and a bunch of secondary ones. In a 1994 study (Cross 1994), the SUN image above elicited no less than 22 different associations. A picture of a rainbow brought to mind 33 ideas to people who saw it. Even a simple letter gave rise to 12 associations other than just “A.” To summarize, here’s a quote from a 1997 article presented at the Closing The Gap conference in Minneapolis:
“Another way to look at this (iconicity) would be use a statistical/behavioral approach and talk about primaryversus secondary iconicity. The primary iconic value is the meaning most frequently elicited as a first response to the presentation of the icon. Thus, when shown a picture of an APPLE, the most frequently used word would likely be apple itself. Secondary iconic values are lower frequency elicited words, such 'eat' or 'bite' for the APPLE.”
And where I say that the most likely word would be apple, there’s another point to make here about the nature of the type of “likely” words; these are going to be nouns. That is they are going to be things: items you can see. If you want a sound-bite, try “The primary iconic label for any image is biased towards being a noun.”
So are you saying that for any picture, there will be more than one possible meaning associated with it?
Bingo! Why do you think people say a picture paints a thousand words? In fact, this is a good way of describing the first fundamental flaw
of using images for communication: that you can’t always be sure that the meaning one person derives from a picture is the same as that perceived by another. One man’s apple may be another man’s fruit. The second fundamental flaw of the model of language that is based on the idea that you can make a communication system from the simple process of representing every word with a picture is that it assumes – implicitly – that every word can be represented by a picture, and that every picture can be discrete from every other. As a mental exercise, think about how what pictures you would use for the following words: angry, mad, irate, irritated, livid, furious, enraged, annoyed. If you are using the one-picture for one-word idea, you have to make sure that each one is sufficiently different from the other, otherwise how would you know which is which. Of course, you could cheat and use the following solution:
“Another way to look at this (iconicity) would be use a statistical/behavioral approach and talk about primaryversus secondary iconicity. The primary iconic value is the meaning most frequently elicited as a first response to the presentation of the icon. Thus, when shown a picture of an APPLE, the most frequently used word would likely be apple itself. Secondary iconic values are lower frequency elicited words, such 'eat' or 'bite' for the APPLE.”
And where I say that the most likely word would be apple, there’s another point to make here about the nature of the type of “likely” words; these are going to be nouns. That is they are going to be things: items you can see. If you want a sound-bite, try “The primary iconic label for any image is biased towards being a noun.”
So are you saying that for any picture, there will be more than one possible meaning associated with it?
Bingo! Why do you think people say a picture paints a thousand words? In fact, this is a good way of describing the first fundamental flaw
of using images for communication: that you can’t always be sure that the meaning one person derives from a picture is the same as that perceived by another. One man’s apple may be another man’s fruit. The second fundamental flaw of the model of language that is based on the idea that you can make a communication system from the simple process of representing every word with a picture is that it assumes – implicitly – that every word can be represented by a picture, and that every picture can be discrete from every other. As a mental exercise, think about how what pictures you would use for the following words: angry, mad, irate, irritated, livid, furious, enraged, annoyed. If you are using the one-picture for one-word idea, you have to make sure that each one is sufficiently different from the other, otherwise how would you know which is which. Of course, you could cheat and use the following solution:
angry mad irate livid furious enraged
But clearly, this assumes you can read: in which case, why on earth would you need a picture system? The way to fix this would seem to be to take each one of these pictures and change them subtly so they are different. Here are the results of a little tweaking with the help of my Xara X drawing program.
angry mad irate livid furious enraged
If you look through each one, you’ll see slight differences. Just to make things more interesting, now you have a discrete picture for angry, how would you go about representing angrier, angriest, angriness, angrily, anger, and angered? And remember, these must also be slightly
different from all the other words pictured above.
Now you’re just being plain silly. You’re taking things to the extreme just to make a fuss over nothing.
Not at all. This is precisely what you have to be doing if you want to have a truly pictorial communication system. And you can’t get away by saying things like “ah, but my client doesn’t need both irate and angry, so we don’t need to make two separate pictures,” because that still doesn’t address the fact that someone will need both. Secondary iconicity, like it or not, represents a real problem for the one-word one-picture solution for picture-based communication. As your vocabulary size increases, so does your bank of discrete pictures. And this is on top of another significant problem that needs to be addressed: how do you arrange all this images so that you can find them again? If you have a vocabulary of 30,000 words, you need 30,000 locations into which to slot your pictures. Just for fun, let’s do some number crunching. Assume you decide to put all these pictures in a book, and have 50 pictures on a page. How many pages will you need for the 30,000 word vocabulary? If you said 30, then you are wrong! But if you said 600, then you’d be spot on. If you want to play around with the numbers, here’s the formula to use:
Number of pages needed = total number of pictures / pictures per page
If you apply this to electronic communication aids, where you have screen pages with pictures on them, you can get an idea of the limits to what you can do. Consider the client with visual issues who can only see pages with 15 pictures; how many pages would there need to be to store a vocabulary of 500 words? Using our formula, that would give us 34 pages.
OK, enough with the numbers! Your point is that the larger a vocabulary becomes, the more difficult it is to (a) design discrete pictures and (b) arrange them in some logical order, yes?
Correct. If you only want a vocabulary of 100 words, then by all means use one word per picture, but if you aspire to larger and larger sets of words, this approach is going to make life harder and harder.
I don’t want to sound pessimistic, but this is beginning to sound like picture-based communication systems are doomed to failure. Secondary iconicity is a bad thing.
Secondary iconicity is, indeed, a confounding factor – but there is another viewpoint you can take. What if instead of seeing it as a problem, we try to use it in some structured way? What if we exploit this feature?
Wait, that sounds familiar. Doesn’t the definition if Minspeak say something about the “…exploitation of secondary iconicity…?” Is this what the exploitation bit means?
Absolutely. If we accept from the outset that when people see the they might possibly think of apple, pick, hemisphere, France etc., depending on their state of mind, how can we set up a system that uses this to an advantage? How do we arrange our pictures so that the person using the picture knows which meaning is being represented? The answer is to use sequences of images to code words. Instead of one picture = one word, we have N pictures = one word. A new formula.
Can you give an example?
No problem. Let me introduce a symbol called DAN. Here he is:
Here’s a guy who is a little mixed up: he’s half black and half white, he’s half male and half female, he’s half hairy and half bald, and he’s half happy and half sad. Dan kind of describes the way a person can be or might feel at any one time. Indeed, D A N stands for Describes A Noun; which is what an adjective does! So when DAN gets dressed in a jacket, how do you think he looks or feels? In other words, how could we describe him? If you said ‘smart’ or even‘scruffy’ they would be good answers? If he picks up a cup that contains a drink how might we describe his state? Did you say ‘thirsty’? And suppose he laid down on a bed, why would he do that? Because he’s …. If ‘tired’ came to mind that, too, would be a brilliant answer. Let’s translate that into using pictures.
Here he is: he must feel smart in his jacket
Here he is: he must feel smart in his jacket
Here is Dan. He's got two drinks! He must feel thirsty.
Here is Dan. He’s in bed. He must feel tired.
So you can see how we could represent words in the following manner:
smart = DAN + JACKET
thirsty = DAN + CUPS
tired = DAN + BED
The innovative thing we have done here is to exploit the secondary iconicity values of the three pictures by using a second picture, the purpose of which is effectively to say “use the word association for the icon that is related to how you might describe how DAN feels in or with it.”
To make the example clearer, think about the CUPS picture and think of the associations that might come from it. Here is a typical set:
smart = DAN + JACKET
thirsty = DAN + CUPS
tired = DAN + BED
The innovative thing we have done here is to exploit the secondary iconicity values of the three pictures by using a second picture, the purpose of which is effectively to say “use the word association for the icon that is related to how you might describe how DAN feels in or with it.”
To make the example clearer, think about the CUPS picture and think of the associations that might come from it. Here is a typical set:
To decide which one of all these possible associations is the one needed, a first icon is used to give the vital clue to the category. As we have already seen, Dan points to the adjective form – thirsty. If we used a different first (CATEGORY) symbol, for exampple one that related specifically to 'colour' the word represented could be 'blue' (COLOUR + CUPS = blue).
To demonstrate how we can then choose other associations, let me now introduce another picture called BRIDGE. In the Bridge picture a Land Rover is on a bridge crossing across with a caravan in tow. Let’s think of the things that a Land Rover can do to a bridge. It can go over it, under it, around it, through it, on it, across it, about it, along it … What are those words? They are not adjectives so they are not DAN words: if you looked them up in a dictionary it might say prep after the word and before the definition. That stands for preposition. A preposition is a word that describes a spatial relationship between two objects (The Land Rover and the Bridge for example). Thus BRIDGE words are prepositions. What prepositions come to mind when we see some of the pictures we have already encountered? As a bit of fun, here is a list of prepositions, try to match them with the pictures (which have been deliberately mixed up in a different order) and see if we agree afterwards:
over across through in of under
in = BRIDGE + PIG (the hand is putting some money IN the piggy bank)
under = BRIDGE + UMBRELLA (The underpants are UNDER the umbrella)
across = BRIDGE + BRIDGE (The Landrover is going ACROSS the bridge)
of = BRIDGE + CUPS (a cup OF tea)
through = BRIDGE + NUT&BOLT (A bolt goes THROUGH a nut)
over = BRIDGE + SHEEP (The sheep is jumping over the arrow shape)
You may not necessarily agree with those choices but that is OK: we all don’t necessarily see things in the same way but, can you understand the rationales for my choice? Do you think you could learn them and commit them to memory? I bet you could! Using our picture sequences, we could then generate those words with an SGD (Speech Generating Device).You are now discovering the world of Minspeak! What we’ve done is to use the combination of pictures in a systematic way to help represent a specific item of vocabulary. It is systematic because it uses a consistent rule and a consistent structure. It is so systematic that it can even help someone guess the meaning of a sequence even if they have never seen it before. For example: What word do you think this sequence represents?
in = BRIDGE + PIG (the hand is putting some money IN the piggy bank)
under = BRIDGE + UMBRELLA (The underpants are UNDER the umbrella)
across = BRIDGE + BRIDGE (The Landrover is going ACROSS the bridge)
of = BRIDGE + CUPS (a cup OF tea)
through = BRIDGE + NUT&BOLT (A bolt goes THROUGH a nut)
over = BRIDGE + SHEEP (The sheep is jumping over the arrow shape)
You may not necessarily agree with those choices but that is OK: we all don’t necessarily see things in the same way but, can you understand the rationales for my choice? Do you think you could learn them and commit them to memory? I bet you could! Using our picture sequences, we could then generate those words with an SGD (Speech Generating Device).You are now discovering the world of Minspeak! What we’ve done is to use the combination of pictures in a systematic way to help represent a specific item of vocabulary. It is systematic because it uses a consistent rule and a consistent structure. It is so systematic that it can even help someone guess the meaning of a sequence even if they have never seen it before. For example: What word do you think this sequence represents?
Remeber DAN represents an ADJECTIVE. So what adjective does a fire represent? Flammable? Yes but think of a much more common word than that... 'hot'! DAN + FIRE = hot. If you thought hot, then you have just benefited from a systematic exploitation of secondary iconicity. If you thought of some other secondary aspect of the fire like ‘fire’ or ‘heat’ or ‘burn’ then that’s OK but remember DAN means the next symbol is an
adjective. To generate those other words we need to use a different symbol in place of DAN as the category symbol.
Amazing! I think I now grasp the meaning of “…the systematic exploitation of secondary iconicity…;” and I don’t even have a headache. But it seems to me that we’ve made life harder by using two pictures instead of one for each word. Why is this better?
Ah, that’s where the last piece of the definition comes in; “…to provide large vocabularies within a small icon space.” Let me explain. Here’s the list of all the words we’ve now represented using our symbol sequences:
In, under, along, of, across, through, over, inside-out, smart, thirsty, tired, and hot
Now, count how many words and how many different pictures? Twelve words; eleven pictures.
Notice something odd? We have fewer pictures in our system than words! Do you want to see the system become even more powerful?
Here’s another picture, it’s a tiger, actually, I affectionally call the tiger ‘Tigger’ and the whale 'Ernest'.
Tigger sits next to DAN on the Minspeak overlay. Do you think Tigger is bigger than DAN, do you think he is fiercer? Tigg-ER adds ER to the end of DAN words such that smart becomes smarter, thirsty becomes thirstier, and hot becomes hotter.
Next to the tiger on the overlay there is a whale ('Ernest the whale'). Who do you think is the biggest: Dan, Tigger, or Ernest? Guess what Ernest does when combined with another picture? He adds the ‘est’ ending to Dan words. Smart becomes smartest and thirsty becomes
thirstiest, and hot becomes hottest. Thus:
Tigger sits next to DAN on the Minspeak overlay. Do you think Tigger is bigger than DAN, do you think he is fiercer? Tigg-ER adds ER to the end of DAN words such that smart becomes smarter, thirsty becomes thirstier, and hot becomes hotter.
Next to the tiger on the overlay there is a whale ('Ernest the whale'). Who do you think is the biggest: Dan, Tigger, or Ernest? Guess what Ernest does when combined with another picture? He adds the ‘est’ ending to Dan words. Smart becomes smartest and thirsty becomes
thirstiest, and hot becomes hottest. Thus:
DAN + JACKET = smart
TIGGER + JACKET = smarter
ERNEST + JACKET = smartest
smart = DAN + JACKET smarter = TIGGER + JACKET smartest = ERNEST + JACKET
thirsty = DAN + CUPS thirstier = TIGGER + CUPS thirstiest = ERNEST + CUPS
hot = DAN + FIRE hotter = TIGGER + FIRE hottest = ERNEST + FIRE
Not only do DAN, TIGGER, and ERNEST represent the superordinate concepts for adjectives, comparatives and superlatives respectively but they are also sequentially arranged on the Minspeak overlay:
A quick recount now shows we have a vocabulary size of eighteen and a symbol set of thirteen. If we then add take just one more example, the DAN and BRIDGE combination, we can now add three more words: wide, wider and widest
DAN + BRIDGE = wide
TIGGER + BRIDGE = wider
ERNEST + BRIDGE = widest
We’re up to 21 words and still only using 13 pictures. Indeed, as there are 128 possible spaces for symbols (icons) on the overlay and we are combining icons together (like DAN and JACKET) to generate a word (like smart) then, if you do the maths, as each of the 128 sysmbols can combine with every other of the 128 symbols: 128 x 128 = 16,384 possibilities (vocabulary items). The actual possibility is far greater than this as the icon sequence length can be longer than two. If you go to a three icon sequence, for example, we now have 128x128x128 possibilities = 2,097,152!
Oh I see! As the words increase, the pictures don’t increase at the same rate. This means that we can actually have a language system with lots of words yet relatively few images.
Exactly!. If we use the word icon instead of “picture,” we are now looking at “… large vocabularies within a small icon space.”
Oh my goodness, I now understand “…the systematic exploitation of secondary iconicity to provide large vocabularies within a small icon space.” Well would you believe it!
See? By going backwards, we’ve actually come forwards. The definition of Minspeak is now no longer mysterious. Sure, at first glance it seemed complex, but when you look at it step by step, you can see that it is an innovative solution to the problem of using pictures
to communicate. Everyone who uses a one-picture = one-word approach is already on the road to using Minspeak. They are already using
secondary iconicity on many of their pictures without knowing it. Anyone who uses a witch for Halloween is using secondary iconicity; anyone using eyes to represent look is using secondary iconicity; anyone using a mouth for talk is using secondary iconicity. All that Minspeak does is to take this to the next level by providing a systematic method of using pictures.
References
Cross, R.T. (1994). Iconicity and Associative Meaning: What are we looking for?
In Proceedings of the 6th Biennial Conference of the International Society for Augmentative and Alternative Communication, 126-127. Hoensbroek: IRV.
Luftig, R.L. and Bersani, H.A. (1985). An Investigation of Two Variables Influencing Blissymbol Learnability with Nonhandicapped Adults. Augmentative and Alternative Communication, 1, 1, 32-37.
DAN + BRIDGE = wide
TIGGER + BRIDGE = wider
ERNEST + BRIDGE = widest
We’re up to 21 words and still only using 13 pictures. Indeed, as there are 128 possible spaces for symbols (icons) on the overlay and we are combining icons together (like DAN and JACKET) to generate a word (like smart) then, if you do the maths, as each of the 128 sysmbols can combine with every other of the 128 symbols: 128 x 128 = 16,384 possibilities (vocabulary items). The actual possibility is far greater than this as the icon sequence length can be longer than two. If you go to a three icon sequence, for example, we now have 128x128x128 possibilities = 2,097,152!
Oh I see! As the words increase, the pictures don’t increase at the same rate. This means that we can actually have a language system with lots of words yet relatively few images.
Exactly!. If we use the word icon instead of “picture,” we are now looking at “… large vocabularies within a small icon space.”
Oh my goodness, I now understand “…the systematic exploitation of secondary iconicity to provide large vocabularies within a small icon space.” Well would you believe it!
See? By going backwards, we’ve actually come forwards. The definition of Minspeak is now no longer mysterious. Sure, at first glance it seemed complex, but when you look at it step by step, you can see that it is an innovative solution to the problem of using pictures
to communicate. Everyone who uses a one-picture = one-word approach is already on the road to using Minspeak. They are already using
secondary iconicity on many of their pictures without knowing it. Anyone who uses a witch for Halloween is using secondary iconicity; anyone using eyes to represent look is using secondary iconicity; anyone using a mouth for talk is using secondary iconicity. All that Minspeak does is to take this to the next level by providing a systematic method of using pictures.
References
Cross, R.T. (1994). Iconicity and Associative Meaning: What are we looking for?
In Proceedings of the 6th Biennial Conference of the International Society for Augmentative and Alternative Communication, 126-127. Hoensbroek: IRV.
Luftig, R.L. and Bersani, H.A. (1985). An Investigation of Two Variables Influencing Blissymbol Learnability with Nonhandicapped Adults. Augmentative and Alternative Communication, 1, 1, 32-37.
Back to Home or Top of Page
Click on the Left Arrow below to return to the Types Page or hit the Up Arrow to return to the top of this page.