Quote:
Originally Posted by Casiopea  Phones nor phonemes are hypotheses, though, but they do factor into a theory about language; that language is systematic, that there are patterns. |
No, neither phones nor phonemes are hypotheses; I agree with that.
But take, for example, the range of sounds classified as "s" and the range of sounds classified as "z". You can describe the actual phone in physical terms using machinery and frequency analysis. Your avarage listener makes that distinction on a daily basis, distinguishing between "seal" and "zeal". Dialects differ in the area of overlap, and in some dialects there may exist no difference at all. The difference is relevant on a speaker as well as a hearer basis. So what's going on here? Do people have a single entity "s" or "z" they hear/produce? Do they have different concepts: "s-sound" + "voiced/unvoiced"? Do they follow childhood notions of imatating ssssnakes and beezzzzz?
Whatever people do, they have - at the very least - habits that make them behave in a certain way. They may have explicit/discoursive theories about their habits, too. ("S like sssnakes, z like beezzz.") The thing is that distinguishes a phonem is "meaning". And meaning has to be attributed. A scientist classifying phones into phonemes is not fundamentally different from a speaker/listener classifying phones into phonemes. Or maybe it is?
Alienating phonemes into a range of phones is a methodology of uncovering hidden rules of creation of meaning. But the rules are the product of human behaviour.
It is possible that scientific hypothesis clash with folk-hypothesis (as you may call them): In my mother tongue, German, I've learnt the difference between "ä" and "e", in writing. I went around assuming there's a difference in pronunciation, but frequency analysis shows that - in most modern dialects - there is none. "Bären" (bears) sounds exactly like "Beeren" (berries); only in situations where the difference is topical is there a real difference in the pronunciation (i.e. in schools, when making clear the supposed difference). It is very hard to exorcise the idea that there is no difference. I heard a difference that wasn't there.
Basically, science has taken a folk-hypothesis and falsified it: in most contexts, there is no discernible systematic difference between "ä" and "e". This hypothesis would not be relevant if we didn't make the distinction in the first place.
Quote:
Ooh. Now there's an interesting spin.  How so? 
Yes, these are questions that either are hypotheses in themselves or lead to hypotheses: "Are there universal features?", "How much can you infer from statistics?", "What is the role of interpretation in the creation and testing of hypotheses?"
|
You'll probably qualify for a noble prize or two if you find a consesus-compelling answer to this question:
How do speaker's/listener's expectations about language differ from scientific hypotheses about them?
In short: Apples falling to the ground do not depend on human cognition; speech does and so does the creation of hypotheses. Is it possible for humans to divise hypotheses about meaning without recourse to "empathy"? "Meaning" like "consciousness" has not yet been explained by scientists.
Quote:
Well, I'm still trying to figure out how data can be considered a hypothesis. Call me thick.  |
The idea is that what makes a phoneme a phoneme and what makes a hypothesis a hypothesis are similar phenomena in that they both depend on cognition. In short, I could hypothesise that what makes a phoneme a phoneme is a hypothesis on the part of a speaker/hearer:
1. Locating s vs. z on a scale.
2. Expecting a word on the sound identified.
3. Recognise word from a database.
4. Semantic testing in context.
etc.
Does this describe scientific methodology? Everyday methodology applied in speech recognition?
Where hypothesis can tranform into generative templates without thorough testing, self-fulfilling prophecies are to be expected. Because lay people can attempt to understand scientific hypotheses and make them apply in future behaviour.