I want to say something pretty simple but with staggering consequences. Language is only inference. Now to show those terrible consequences i will take a long detour.
If i am walking outside and i see a dark, fluffy cloud and hear thunder, it might happen that i come up with the idea of going inside ASAP. It wouldn’t make me the cleverest person in the world if i did think thusly.
But let me propose another example, one that is less familiar. Imagine i am on a big white room where the floor is composed of tiles. Some of the tiles are red, some are blue but most are white. I start to walk around, and when i step in a red tile i receive an eletric shock. I continue to walk around, and again i step in another red tile and again get a shock. As times goes, every time i step on a red tile i get shocks and every time i step on a blue tile i receive a Slurpee from a big shiny hand in the sky.
You might imagine that, pretty soon, i will draw the conclusion that red tiles mean shock and blue tiles mean bonus. Actually, i have given this example to show that even lab rats can make inferences given the proper circumstances.
Now language is only inference.
For example, when you say “bring me a cup of water” and moments later i show up and give you water, i did not understand that “water” is a transparent liquid composed of Oxygen and Hydrogen, i did not understand that by “cup” you meant a recipient capable of holding liquids with internal volume of about one third of a liter.
I only and merely thought that it would fit to perform the action of bringing you a cup of water. Just as a dog extends his front leg when you say “give me the feet”, without understanding what do “give” or “leg” mean.
Actually, i might be capable of saying many words regarding “what does water mean” (that is, giving the definition of water). And i also might have a deep understanding of the chemical properties of H2O. Nevertheless, neither is important for the communication process whereby i have inferred that you wanted me to bring a cup of water.
When more than one person is in the big white room with colored tiles, for example, it might come to pass that i infer that when you yell and jump and make faces you mean that i am about to step on a not-good tile. It is still inference, even if you had been trying to influence me into it.
Now after millennia, this human trick of making specific sounds into specific circumstances has produced the side-effect that we have large hierarchies of words and word-sequence-techniques that are collectively called languages. And languages have many uses, but they do not change the nature of the process they’re based on. Language is still inference. Only.
Just for starters, my hypothesis would imply that word-by-word exchanges (that is, definitions) are basically only possible inside already very structured fields of experience, or in other words when the person you are talking about already almost knows what you are talking about. Dictionaries are useless before you know most of what is in them.
It also forces many linguistic concepts to be revalued (ex. “conotation”, “denotation”, “reference”, “iconicity”, and so on).
But what most casual readers will find shocking is that my hypothesis also implies that during normal conversation the two persons speaking do not share a common meaning for the words.
When i say “i love you” we do not share a common idea of “i” and “you”, even if we can point correctly (as a well-trained dog could) at the person supposedly referenced. But now when i say “love” it becomes clear that there is not one “meaning” of the word, but instead each person have different experiences related to human relationships, and amongst the set of experiences each person divides what is love from what isn’t in a independent process. The more important the word, the less likely there will be common experience to enforce similarity of inference.