However, it is easy to lose sight of the fact that what linguistics presents is a model - a map. Map vs. territory is an important thing to keep in mind. Some hobbyist linguists with a background in conlanging sometimes forget about this dichotomy.
The model I am about to present probably is not all that useful. It is, however, an attempt at highlighting what linguistic models leave out. It is also an attempt at modelling language using a very powerful tool from mathematics - probabilities. We will find certain rather powerful things about distributions in general, and we may possible learn something about language just from observing things about probabilities.
There are a few things I wish to draw attention to, and which I will try to provide 'uselessly convoluted' methods to model in general, then try to see what results we can obtain by reducing the convolution. The things I mainly want to draw attention to are:
- idiolectal variation
- linguistic parsing as something that often is haphazard and somewhat random, affected by associations the lexemes and phrases trigger
- how associations work, and how important they are as a complement to the more regular kind of 'meaning' we tend to think of when thinking of words
- the implications of the brain being a neural network, especially with regards to the previous point
- language as being a complex system of which its participants only have a partial copy in their mind
Some of these might be quite obvious, but sometimes when discussing language, we still fail to acknowledge these issues. So, a greater description might be justified.
No comments:
Post a Comment