Imagine for a moment, that you have been thrown back into the Ellisonesque world of the 1980’s, with a delightful perm and even better trousers. One fragile Monday morning, you are sitting innocently enough at your cubicle, when your boss comes to you with the summary of a report you have never read, on a topic you know nothing about. “I’ve read the précis and I’d love to take a peek at the report," he intones, leaning in. "Apparently, they reference some fairly intriguing numbers on page 76.” You stare blankly at him, wondering where this is going. “Yess—so I’d love if you could generate the report for me.” He smirks at you expectantly. You blink, twice, then begin to stutter a reply. But your boss is already out the door. “On my desk by five, Susie!” he whistles (as bosses are wont to do) and scampers off to terrorize another underling.
You would be forgiven if, at that moment, you decided it was time to knock a swig or two off the old bourbon bottle and line up some Rick Astley on the tapedeck.
Because the task is, in a word, impossible.
Certainly you might, on a slightly less surreal morning, tidily summarize a long research article into a brief abstract. But the reverse process simply doesn’t work. When an abstract is created, information is discarded for that purpose. Which means that it can’t be recovered. Which means that you, sweet Susie, have a problem.
The insight that this gives rise to – that abstraction can only work in one direction – becomes critically important when we study language, because words are abstractions of what they mean. Think about the difference between say, the word ‘dog’ and a real life dog. The word is a sound symbol : a brief sequence of phonemes. The dog is a complex perceptual entity : visually, it has a quite a number of different discernible features (its size, its coat, its snout); it has a distinctive woof and a distinctive gait; it smells in a particular way when it comes in out of the rain. This reality is not—and cannot be—fully captured by the word. You simply cannot code all of the complexity of dog into 'dog.'
This means that while the world can get abstracted into words, we can’t get all of the ‘world’ back out of words.
That may seem blindingly obvious.
But it might not appear that way if you were a trained philosopher.
In analytic philosophy – and to some extent in linguistics and cognitive psychology – there are those who believe that words get their meaning by reference to a set (or class) of things in the world. Take the following example, from the Stanford Encyclopedia of Philosophy:
How do I manage to talk about George W. Bush and thereby say meaningful and true things about him? In a word: Reference. More picturesquely, we are able to use language to talk about the world because words, at least certain types of words, somehow ‘hook on to’ things in the world — things like George W. Bush.
If you aren’t intimately familiar with this idea, not to worry. In the history of ideas, 'reference' isn't going to have much of a shelf life. Luckily, the basics are fairly easy to grasp.
Just think of the way we use the word ‘refer’ in everyday language. In the course of a conversation, for example, a friend might confide to you that Todd was ‘referring’ to his mother-in-law in salacious terms, and you might decide you’d rather he didn’t ‘refer’ to your mother that way.
‘Refer’ is quite a handy word conversationally, because it allows us to specify “Oh, but I meant this not that.” The problem arises when it is understood literally by theorists of various persuasions, who want to say that words get their meaning by ‘referring’ to things in the world. A theory of reference is inconsonate with the idea that words are abstractions. Words cannot ‘latch onto’ the world and feed us back meaning any more than an abstract can feed us back a much longer paper; the metaphor simply doesn’t work. To build a theory around the idea that they can and do, is to be dealings in fictions.
As you may remember from an earlier post, our understanding of the mind and of language is governed by the metaphors we use to guide our study. There are better and worse metaphors—subject to empirical scrutiny— and some that make no sense at all. It would be hard to say how language is ‘like a toaster,’ for example, or how using a ‘toaster model’ of language would add anything useful to our methods of study. Of course, posing a 'referential' model of language is far more tempting than elaborating a toaster model; so long as we don't push on the analogy too hard, it seems that language just might work that way.
The problem is that when we do push on the metaphor, and think about the problem in cognitive terms -- instead of 'hand wavy' ones -- it becomes both logically incoherent and computationally intractable. Indeed, Wittgenstein famously mocked the concept of reference in the Philosophical Investigations:
"This is connected with the conception of naming as, so to speak, an occult process. Naming appears as a queer connexion of a word with an object. --And you really get such a queer connexion when the philosopher tries to bring out the relation between name and thing by staring at an object in front of him and repeating a name or even the word "this" innumerable times. For philosophical problems arise when language goes on holiday. And here we may indeed fancy naming to be some remarkable act of mind, as it were a baptism of an object. And we can also say the word "this" to the object, as it were address the object as "this"--a queer use of of the word, which doubtless only occurs in doing philosophy."
Yet here we are, sixty years later, still recycling the same bad ideas. I admit -- it puzzles me.
For a technical discussion of some of the problems with 'referential' models, see Ramscar et al (2010) Section 1 (910-912).
Ramscar, M., Yarlett, D., Dye, M., Denny, K., & Thorpe, K. (2010). The Effects of Feature-Label-Order and their implications for symbolic learning. Cognitive Science, 34 (6), 909-957 : 10.1111/j.1551-6709.2009.01092.x