What's a good way to understand the difference between phonetics and phonology?

[zaba]

I "get" the difference between underlying and surface forms -- and I know what the respective fields are, per se, but how can I best understand the difference between phonetics and phonology in terms of human knowledge?

[Daniel]

Phonology is how sounds interact in a grammatical system. It's language specific and arbitrary (to some degree).

Phonetics is how sounds are produced by the body, conveyed through the air, and perceived by the ear.

Certainly there is some overlap between them, but the basic difference is obvious: Phonology is grammatical/linguistic, and Phonetics is physical.

[zaba]

Phonology is grammatical/linguistic, and Phonetics is physical

Sure! But how does one "get" from one to the other?

[MalFet]

Sure! But how does one "get" from one to the other?

Phonology describes the human capacity to interpret gradient sounds as categorical symbols. In other words, the ability to get from one to the other is what defines phonological knowledge.

[zaba]

Thanks!
So phonology is the study of the organisation of sounds in a language; where:
(A) "sounds" = phonemes
or
(B) "sounds" = phones

[zaba]

Thanks again! Also, by

gradient sounds as categorical symbols

you mean "phonemes" (= gradient sounds)
and "phones" (=categorical "symbols")

Correct? I just want to make sure I'm understanding accurately.

[ibarrere]

you mean "phonemes" (= gradient sounds)
and "phones" (=categorical "symbols")

Other way around. The stream of sound (phones) is gradient and ever-changing, the human language apparatus identifies categorical symbols (phonemes) out of that stream of gradient phones.

[Daniel]

Phonemes are perceptual categories. Whatever actual phonetic form they take, we perceive them as part of category X.

Phones are just sounds. "Gradient" means somewhat variable. So an individual instance of a sound is not gradient. It's just some precise acoustic/articulatory information. But in general a phone like /t/ is gradient because no two sounds are identical. Phonemes, however, are also gradient in a sense: they're categories that contain infinitely many subtle variations in pronunciation, still within the same category.


Phonemes are perceptual categories.
Phones are analytical categories used by linguists (and perhaps orthographers).
Phonemes are phones (fewer phoneme categories, more phone categories). Allophones are subphonemic variations in phones-- not perceived to be different, but different enough for the analyst to note it.



Think of phonemes as geometric shapes you recognize ("triangle"). Think of phones as shapes relevant to mathematicians ("acute triangle"). And think of gradience/variation as the real-world instances of those shapes in general (perhaps a "yield" sign). (Allophones then would be variation in context, so perhaps different sizes of street sign depending on the distance from which observers were looking.)

[zaba]

Phonemes are perceptual categories. Whatever actual phonetic form they take, we perceive them as part of category X.

This sentence just blew my fragile little brain. Can you give me an example to make this more concrete?

[zaba]

Phones are just sounds. "Gradient" means somewhat variable. So an individual instance of a sound is not gradient. It's just some precise acoustic/articulatory information. But in general a phone like /t/ is gradient because no two sounds are identical. Phonemes, however, are also gradient in a sense: they're categories that contain infinitely many subtle variations in pronunciation, still within the same category.

Think of phonemes as geometric shapes you recognize ("triangle"). Think of phones as shapes relevant to mathematicians ("acute triangle"). And think of gradience/variation as the real-world instances of those shapes in general (perhaps a "yield" sign).

OK, this is super interesting. This example is fundament to my understanding of what's up so I want to make sure that I don't misunderstand something. According to this analogy:

phoneme = the broad geometric category (triangle, octagon, square)
phone = geometric analysis of a category member (right-triangle, acute triangle)

So far so good?
It gets a bit fuzzy when you suggest that we consider

gradience/variation as the real-world instances of those shapes in general (perhaps a "yield" sign)

. Isn't the phone already a real-world instance of a phoneme? Aren't both phonemes and phones gradient?

If you could clear this up for me, you'll have successfuly taught me more about phonetics and phonology than my brainiac intro prof. did in the last semester! 

Thanks a lot!

[Daniel]

This sentence just blew my fragile little brain. Can you give me an example to make this more concrete?

If you hear R and L as a speaker of English, you perceive them as two categories. Japanese speakers perceive them as only one category, not contrastive.

[lx]

The phones [pʰ] and [p] are not contrastive in English. They are part of the category (phoneme) /p/.
In Hindi, the phones [pʰ] and [p] are contrastive. Changing one in certain contexts gives you a completely different word.

When phones are not contrastive, they're allophones (they are subsumed under the same phonemic category).

This means the way you represent phonemes is language-dependent.
The way you represent phones is language-independent.

In no matter what language you speak of, there are phones [pʰ] and [p]. But, we cannot say that there is a /pʰ/ in English (phoneme), but we can in Hindi.

Hindi:
Phones: [pʰ] [p]
Phonemes: /pʰ/ /p/

English:
Phones: [pʰ] [p]
Phonemes: /p/

[Daniel]

Phone: a sound category, determined by linguists
Phoneme: a perceptual category for a speaker of a language
Allophone: a phone that is NOT contrastive, within the range of a phoneme

[zaba]

Phone: a sound category, determined by linguists

I'm getting it. Thanks! Earlier you said that a phone was an analytic category -- but now you say it's a "sound category". Can you elaborate a little bit?

[Daniel]

Same thing. For example, we might say an "m" sound. It's a sound category determined by linguists via analysis-- an analytical category.

Think of linguists as auto mechanics-- one of the first things a mechanic might do is separate vehicles into different classes, such as trucks vs. cars. This says nothing about how people actually use the vehicles.