I should begin by confessing that I am not by any means the linguistic genius that my last letter may have made out. Actually, I've often envied the hispanophones on the list, because I don't really have any ACADEMIC language other than English.
My Chinese depends largely on it being the language in which I am nagged by wife, and I often have to ask the graduate students to disambiguate the Korean in my data. Even my English is not always limpid; in the last letter I said that light vowels were yin and dark ones were yang in Korean and I meant to say the opposite (I get "left" and "right" confused a lot too).
I should also confess a personal interest in disproving the existence of phonemes. When my mum did her Ph.D. in first language acquisition, I was her chief experimental and observational subject. Her thesis was about dichotic listening; she used it to demonstrate that children sample sound at about the rate you would expect if they were working with syllables, not phonemes.
Mike counters by saying (or at least implying) that mankind has used language for so long that we have evolved categorical perception of sounds along phonemic lines. This is consistent with other things he has written about the co-evolution of the brain and language. It is also consistent with some research on neonates that my mother just after I was born, which I also, in my modest way, participated in.
I think there's a better explanation, though: we designed language to fit the kinds of perceptual categories that we had already evolved. I think this explanation is more consistent with Tomasello's insistance that the time scales involved in the phylogenesis of language do not permit the kinds of large mutations that would be necessary (and also with his desire to never admit a genetic solution where one based on cultural transmission will do). It is also consistent with the extremely divergent phonological systems of the world (something disguised by the use of IPA but pretty clear to me) and Roman Jakobson's ideas about complementarity (that is, we start with EXTREME vowels and gradually fill in the middle).
My argument is slightly different though. Predicatably, I guess, I consider phonemes to be a kind of developmental palimpsest, with different categories of articulatory gesture, ranging from visible physical gestures (like /m/ and /b/), built over with purely abstract categories that have no basis in physical actions, much less genetic mutation. "Vowels" and "consonants" are purely abstract categories, although like many of our abstract, mental categories they have a very rough physical basis ("open" vs. "closed").
It is very well known that babies teach their parents to speak using /m/ and /b/ in almost every language on earth. They get bored with repeating open naseo-pharyngealized vowel sounds (that is, crying) and they try the most obvious way of stopping them, using the lips.
When they are inhaling this produces "m" and when they are exhaling this produces /b/. True to Vygotsky's description of how a sign first is a thing in itself, and then becomes a thing for others, and only in the final analysis, when the "social" meaning of the gesture has been moved to the front and becomes intentionality, can it be consdiered a sign for oneself. This explains why most of the world's languges have names for mother and father beginning with precisely these sounds.
But from there things get much more complicated. The reason why Sejong's linguists had to use abstract signs meaning "yang" and "yin" and "heaven" and "earth" for the vowels is not simply because there are "bright, light" vowels and "murky, muddy" vowels (notice how the same phonaesthestic vowel distinctions seem to apply in English!). It's because we cannot really see or feel how vowels are articulated.
The "tongue position" charts we were all taught about turn out to be largely innacurate or irrelevant, and the way a vowel is pronounced depends very very much on the consonants in the immediate environment (which is why in fast spoken English almost all vowels tend to the centre of the mouth, and become "schwa"). When I teach my students, I don't talk about tongue position at all, I just set up a grid showing "kissing" (that is, lip position) vs. "biting" (jaw position). Much more useful, and probably much closer to the way babies attack the problem of imitating and differentiating those pesky back vowels.
I think that by and large the acquisition of phonemes follows Vygotsky's "from the outside inwards" rule, with /t/ and /d/ acquired before /k/ and /g/ and all of these acquired before /h/. But /h/ is a big problem! Why should /h/ be considered a consonant? Unlike all the other consonants, it has no distinctive articulatory gesture--you simply use the kissing and biting configuration of the following vowel when you make /h/.
I think that the difficulty of this point (for children) is why we have so many Humpty dumpty rhymes in English ("higgledy-piggledy-pop", "hickory-dickory-dock", etc.). The fact that /h/ is a consonant is learned, not innate, and the child registers of our language (the set-up wizards that make it learnable for children) are designed to teach this key point. The child is clearly led, by the salience of the initial sound, to reason that if /d/ is a consonant, well then so is /h/. I think that the same thing explains why our "vowel harmony" rhymes (e.g. "hip hop", "fiddle faddle", "fee fie foe fum") appear to obey the rule of "closed open"--the most consonanty vowel /i/ goes first and is followed by more indubitable ones, so that the child reasons by distribution. Similar rhymes exist in Korean and Chinese. Pinker, of course, claims all this is innate.
So you can see that there are some sounds (particularly the most physical sounds, e.g. wide open howl vowels and clear, unambiguous labial stops like /m/ and /b/) which might be part of our genetic endowment. But this does not mean they evolved for language purposes; it's important to keep in mind that speaking is an exaptation of organs evolved for the purposes of breathing and eating, and /m/ and /b/ are what they are precisely because of their relationship to breating and eating (which is why /m/ figures prominantly in ALL onomatoepoeic expressions for eating, such as "yum yum yum", and we have similar indexical expressions for spitting, such as "ptui" that start with a /p/ or /b/ sound).
Most of what we call "phonemes" are not like this, though. Instead, there are a wide range of purely abstract distinctions, apparently learned in opposition to each other, and at least some of them are language specific and quite arbitrary (for example, /w/ is a vowel in Korean, not a consonant).
They are also, as Vygotsky points out with his famous example of the volitional segmentation of the word "Moscow", late emerging. Children cannot deliberately pronounce the sound /sk/ in Moscow, because the phoneme is not psychological real until literacy has been acquired. Suspicious, no?
The basis of spoken Chinese is, as the writing system indicates, the syllable, and in fact there is no way to render Chinese syllables (or, for that matter, the syllables of any other language) into phonemes that is not at some level purely arbitrary (and therefore not inheritable). Hanyu pinyin (that is, "Romanization") was a very late invention (that is, the 1950s); my mother in law still does not know it, and that is why we cannot really communicate using e-mail (since my keyboard doesn't allow me to use Chinese with her).
Here is a little experiment you can try which should convince you a) that phonemic categorization is not an important part of comprehension and b) that syllables are.
When you say "cap" or "cab" (or "rack" and "rag" or even, to choose an example from my elementary English textbook, "can" and "can't") the syllable ends with a stop. This means that you cannot really hear it at a distance; if the speech flow is genuinely stopped, there is no way for it to influence the air flow, and no actual way to communicate it (this is why initials are more salient than finals, and why it makes more sense to have ambiguous sounds like /h/ in front than at the end!). But we can tell whether someone says "cap" or "cab" very easiliy--we listen to the length of the VOWEL.
Try saying "cap" with a slightly longer /a/ sound, like the one you find in "cab". If you really close your mouth over the stop /p/ (don't spit it out), then the person you are talking to will hear the syllable "cab", not "cap". Why? Because, of course, /b/ is a VOICED consonant, and all vowels are voiced. So when you have a voiced consonant, the vowel appears longer. When you have an unvoiced final consonant, the vowel has to stop. This is even true when the unvoiced final consonant is mediated by a voiced one, as in "can't". "Can't" is STILL a shorter syllable than "can".
I apologize to all Swedish magpies, Leif. Korean magpies also build their own nests, and I think the English expression might actually be a slander on the species, like when Chairman Mao tried to wipe out sparrows. Perhaps magpies have been framed for crimes really committed by cuckoos?
But the "magpie" adjective is still useful in the sense that Mike was talking about--as a way of talking about how, unlike in biological evolution, earlier structures remain and can still be discovered underneath the later ones. That is, magpies build their NEW nests out of OLD nests, and in Korea we can often see the remnants of the old nests in the foundations of the new. Last night as I was nodding off, I came across the following quote from Halliday, which kept me reading a good half hour past my bedtime.
“What I have been suggesting is that if we compare the two histories, the evolutionary history of the linguistic system and the developmental history of the learning of language by children, we find that the two are related epigenetically, that is, the development of the child’s power of meaning follows the evolutionary trajectory—in a way that is analogous to the epigenetic biological development of the organism. But I want to make it clear that the analogy is only partial: semiotic systems are not the same as biological systems, and one fundamental distinction is that the grammar retains the features of all its earlier historical moments. Children do not give up the commonsense grammar when they move into educational knowledge; nor does the clausal mode disappear from the system when the nominal mode takes over.” (p. 22).
Vygotsky makes this point too--higher functions are built on top of lower functions, and the latter can still be observed when the higher functions decay. This is one of the reasons why looking at languages, writing systems, and even the phoneme system as rag-bag palimpsests is so difficult. It is a bit like Schliemann discovering the ruins of nine cities build on top of each other--we still don't really know which one was Priam's Troy.
Seoul National University of Education
PS: By the way, Mike, your Goethe quote is found in labs, but also at the beginning of Bronfenbrenner's great book, " The ecology of human development".
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
xmca mailing list
This archive was generated by hypermail 2b29 : Thu Feb 01 2007 - 10:11:31 PST