The words we read seem to carry meaning. We believe that there is a message encoded in the text. And since we see it written there in black and white, we are usually quite confident that we are able to decode it. Let’s see why this can not possibly be the case.
The belief that words have meanings is intuitively appealing to both laypersons and linguists. The meaning of a particular word like tree comes so naturally that it seems beyond doubt. After all, everybody knows what a tree is.
Unfortunately, there are about 7.4 billion distinct ideas of tree. Many of these are similar, but some less so: imagine a biologist studying trees and a desert nomad who has never seen one. Or even a biology student vs a biology professor.
These differences are what make arguments possible, from domestic to legal and political. Every verbal argument can be viewed as being over the meanings of words. As an attempt to limit the range of possible meanings, people create dictionaries, terminology standards and legal definitions. It is easy to see how these fail:
- There can be, and frequently are, conflicting definitions in equally legitimate sources. If the goal is a single right answer, conflicting sources can not both be right at the same time.
- The sources are compiled by humans. Sometimes these people have some kind of claim to representativeness, sometimes not. Even if they do (e.g. a democratically elected parliament approving legal definitions), their abilities of both data acquisition and definition writing are humanly finite, and may be further limited by political compromises or technical restrictions. Also, the definitions themselves are often results of discussions just like the original debate. So, referring to authoritative definitions only delegates the debate to another group of people, instead of solving it in an objective way. The authoritative sources state what conclusion did that other group reach, not what the word means.
- Definitions too are expressed using words, the meaning of which is exactly as open to debate as the original meanings. Some of them will have definitions of their own, but as one proceeds down the chain of definitions, the problem only gets worse. Regardless of the length of the chain, at one point it inevitably either becomes circular (referring back to the original debated meaning) or resorts to using non-defined concepts, assuming them to be common knowledge. Again, everybody knows what a tree is.
No they don’t. Even highly educated lawyers, authors and critics still continue the debate, despite adhering to the same legal or dictionary definitions of terms under debate. If words had meanings, how would this be possible?
Background from information theory
It is much simpler to let go of the belief that meaning resides in the word, especially since it appears that retrieving it from there would not even be mathematically possible. This can be explained in two equivalent ways.
For lossless coding to be possible, the coded version (words) must contain at least as much information as the version it encodes (thoughts).
Consider all the trees in the world, and how much information would be needed to describe them all. Or consider the effort needed to describe what each person alive considers to be a tree. Describing ideas of tangible objects is even relatively easy, compared to things like kindness. Huge amounts of information. Compare that to information contained in the word tree: eight bytes at most.
Without additional information sources, it is not possible to take an item containing eight bytes and retrieve from it all the richness of knowledge mankind has about trees.
For lossless coding to be possible, the set of available code words must be at least as large as the set of items to be coded.
The Second Edition of the 20-volume Oxford English Dictionary contains full entries for 171,476 words in current use. This can, depending on the counting method, be extended to several million (if we also count proper names), but the number of available words is still in the millions at most. Compare that to the number of possible thoughts that people are capable of, which is basically limitless. Each word corresponds to n thoughts, where n is usually much larger than 1.
Without additional information sources, it is not possible to start from an item from the smaller set and know which member of the larger set it refers to.
|Words can not possibly encode meanings, because there are more meanings than there are words.|
For understanding to take place, there must be an additional source of information, and this source is the reader’s experience. Information about the writer’s thoughts is lost in the writing process. To achieve a sense of understanding, readers add it back, not based on what they read, but based on their own prior experience and knowledge. The experience of each reader is different.
- No text means anything on its own. Texts only mean something for particular readers, and they mean different things for different readers. The observation that they may mean similar things for similar readers is not a refutation of this. First, the readers must be similar in order to achieve a similarity of understanding, and second, even very similar readers will still find small differences in their retrieved meanings, if they look closely enough.
- Texts mean different things even for the same person at different points in time. Try re-reading a book you liked when you were younger. The meaning you arrive at will be different, although the text is the same. The reason is that your experience has changed.
- Since meaning is constructed by the reader, the reader is also responsible for which meaning they construct. Any references to text meanings (“The law says that..”) are either misguided or dishonest. The text doesn’t say or do anything. It’s not even alive. The reader does it all.
- Arguments about meanings are pointless, as is ironically proved by the very existence of the arguments. If there would be such a thing as the literal meaning of a text, then it would be sufficient to find it out, and debating would not be needed.
- Since arguments do take place, they do also get settled, often to the satisfaction of all parties concerned. But it is important to note that the method can only be brute force: since there is and can be no objectively correct understanding of a text, the winner is always determined by which side has a louder voice, higher position or more friends.
- Although texts do not objectively mean anything, this is not end of communication and collapse of the society as many have feared. There still is a convention about how to understand texts, determined by the same brute force arguments. If you want to conform with the group, you choose to understand texts the way the group does. As best you can, of course, given your experience or lack of it.
- The same goes for writing. There is no way to achieve lossless transfer of information to the readers. If the goal is to be understood, the writer can only craft the text in such a way that the target audience will most likely be able to add back the lost information. If that fails, there is no point in claiming that the writing was correct, only the readers understood it incorrectly. The writing may have been correct (whatever that is), but certainly it was not successful.
- Since words do not have meanings, these can also not be represented in dictionaries. What is usually represented is a description of the lexicographer’s understanding of the words (i.e. the higher position or louder voice argument). The qlaara dictionary uses the number of friends argument instead, documenting how people have used the word so far.