Re: [tied] Phonetic iconicity

From: Piotr Gasiorowski
Message: 5236
Date: 2000-12-30

----- Original Message -----
From: Glen Gordon
Sent: Saturday, December 30, 2000 8:03 PM
Subject: Re: [tied] The Googoo Hypothesis must be squashed

>By what criteria do we possibly measure such a thing? The size of vocabulary of any given language is virtually limitless, so we can hardly arrive at any meaningful percentages.
The only thing I can do here is refer you to the research done so far. Let us not try to reinvent the wheel. Your statement that no meaningful statistics can be derived from lexical studies is patently false. The very reason why statistical generalisations are so useful in science is that very large sets (say, all molecules in a container) can be characterised *only* in that way. Who says every single item has to be taken into account? A representative sample will do. And I was talking only about recently coined words (in order to circumvent problems caused by cognacy). I can assure you their number is not as vast as you make it look.

>...Judging by your Batta example, we've gone off the original topic of phonemic symbolism and into what would appear to be "frequency symbolism" involving the entire vocalism of the word. I can see how this makes sense since high frequencies go with small animals/humans and low frequencies go with large animals/humans. Large men with feminine
voices are the meat-and-potatoes of many a comedy sketch because it's so unexpected and thus draws in a quick laugh. So you're talking about the quality of frequency not phonemes themselves. Why didn't you say so in the first place?
I was talking about *classes* of sounds ("front vowels", "low vowels", "coronals", "dorsals", etc.). A class is characterised by a common distinctive feature which has a certain articulatory definition and a set of acoustic correlates. The latter include things like vowel formants.
What do you mean by "frequency"? The fundamental frequency, i.e. the pitch produced by the vocal folds? Indeed, it can be used in tone languages to symbolise size. Such "tonal iconicity" exists e.g. in Yoruba, where (predictably!) high tones mean "small" and low tones mean "large"; e.g. gbo`ro` 'to be wide' versus gbóró 'to be narrow'.
The frequencies of formants, on the other hand, are not independent variables but functions of the articulatory configuration. For example, [i] requires the raising of the whole body of the tongue and its forward movement -- an articulation that leaves very little space in the oral cavity. *That* is the iconic thing, not the fact that the formant frequency (F1) is low for [i] or that F1 and F2 are extremely far apart for that vowel. I didn't use the term "phonemic symbolism", and I *wouldn't* use it, since as you rightly point out, phonemes are abstract and have no meaning. But *phonetic* symbolism is perfectly possible, since both articulatory gestures and the sounds they produce are physical and can be iconic.

>I've seen IE *kaka- reconstructed so what's the story on that then?
Don't ask *me*, Glen. I don't trust such "protowords" at all. I prefer *sok-r/*sk-o:r/*sk-(e)n-, a heteroclitic noun with enough morphology to guarantee its reality. I've seen PIE *kuku- as well and don't suport it either as a comparative reconstruction (even though it is a priori probable that the IEs had an iconic term for "cuckoo" like most peoples familiar with the bird's call). I'd believe in *kuku- or *kuku:- if I ever saw it transformed by early sound changes in the usual manner, say, *kuku:- (with final stress) > Germnic *xugu: > Old English *hugu > Modern English *how or Old Church Slavic *kUky, Gen. *kUkUve -- which is precisely what I've never seen. One reason why iconic words make lousy comparanda is that they rarely develop according to regular "laws". Once in a while sound changes threaten to distort the shape of an iconic word so much that must be abandoned or "refreshed" if it's to remain usable as an icon. Middle English cuccu [kuk'ku: ~ 'kukku:] replaced Germanic [jE:k] but it can't be the direct ancestor of Modern English ['kuku:] via regular development, or we would have ['k^kou] or [k@'kau].

>>Babies often go "mama" by way of asking to be breastfed (cf. Latin mamma 'nipple'), and mothers tend to think that the sound refers to them. It's a "situational" universal, not even an example of sound symbolism.

>Then why did you mention it if it has nothing to do with phonemic symbolism? I agree that it's a situational universal.
I mentioned it as an example of non-arbitrary sound-to-meaning relation -- something that is induced by extralinguistic circumstances. It is not phonetic symbolism, but it is *like* phonetic symbolism in one respect that is of interest to us: it produces false cognates.