Monday, December 9, 2013

Conlanging by Fiat: Unjustified Claims for Typological Quirks

There is a thing that bothers me about the approach to linguistics that some conlang descriptions rest on.

Not because it relies on bad linguistics (which it does), but because it relies on bad epistemology and a generally bad approach to understanding how things works and what makes them what they are.

The example I will give is phonology-based, but similar kinds of mistakes appear in grammatical descriptions.

I guess I have to explain this in greater detail . When we discuss phonology, we are making several related kinds of statements about the sound system of a language - we are making a claim about what actual phones appear in the language, the distribution of these phones, and the parsing of these phones in terms of what phonemes they belong to - basically how the brain will analyze the sounds it hear, and finally there's a more meta thing that no one ever thinks about. This final thing is how would a brain that doesn't know a single language assign the phones they hear to phonemes in a way that is natural for the human brain. Most conlangers skip that step, and sometimes make rather unjustified analyses that not only are unlikely and unnatural, but basically unlearnable except by being explicitly taught the way they are designed - which 'natural' speakers generally wouldn't be, most people acquire their phoneme system by unconscious analysis of the language they hear around them.

2 comments:

  1. I can see how this can occur if someone doesn't have enough knowledge of how other languages work or concepts like natural classes, but could you give an example, perhaps?

    ReplyDelete
    Replies
    1. I've been trying to come up with some examples for a while, and kind of gave up - this post's been sitting in my draft folder for ages already, waiting for that one golden example. Either my examples get too obscure or unclear, or they're just ripoffs of some conlang draft I've seen on some forum somewhere. Whenever I get a good example I'll post it.

      One kind of thing that sort of falls under this is when people have an allophone belonging to several phonemes - oftentime, this seems to be justified by fiat rather than by any systematic pattern in the language. Essentially, the description of the phonology doesn't provide any other reason to think that the allophone is shared than that the author says it is.

      Delete