From Geoffrey Sampson's excellent Writing Systems: A Linguistic Introduction (p. 36):
In the first place, the units of the 'first articulation' of a language tend to be relatively apparent to native speakers of the language without special study—a child does not need to have learned to read and write in order to be able to split up a spoken sentence into separate words—while the units of the phonological 'second articulation', particularly phonological units smaller than syllables, are not obvious. A child has to be taught to hear the word cat as 'ker-a-ter', and such learning is far from easy for many; there is nothing self-evident or natural in the splitting of a speech-chain into separate vowels and consonants.
Leaving aside the question of whether it's true that word-segmentation is easy and natural (I think it's a bit more complicated than that, although I agree there's a vast difference between word- and phone-segmentation), what's this business about "ker-a-ter"?
I would pronounce "ker-a-ter" as something like [kɚ aː tɚ]. Unfortunately, that doesn't mean anything. Turns out the problem is that I speak an r-full dialect of English, and Sampson doesn't. According to his site, Sampson is an Englishman, and apparently he speaks an r-less dialect, because when writes "ker-a-ter" he's thinking of the pronunciation [kə æ tə]—that is, C-A-T, pronounced phone-by-phone.
Separated by a common language, indeed. Isn't this what the IPA was invented for?
He's talking about spelling out the _letters_ by pronouncing them, isn't it?
Posted by: des von bladet | June 11, 2004 at 08:59 AM
The locations of word boundaries in written English correspond 90%+ of the time to points that can be found using statistical methods. I know of no comparable result for phonemes and several good arguments why no comparable result should be possible for phonemes.
Posted by: Scott Martens | June 11, 2004 at 10:41 AM
When I saw the title of this post, I imagined the word character in a dialect of English that simplifies /kt/ clusters.
And wouldn't most non-rhotic speakers put a linking /r/ between the ker and the a?
Posted by: Q. Oheevr | June 11, 2004 at 11:04 AM
He can't be talking about spelling out the letters, or else it would be something like "see-ay-tee". It's pretty clear from the context that he's talking about learning to segment a stream of speech into phones (or maybe phonemes).
The point about the linking /r/ is interesting. If the dashes are read as pauses, then /r/ wouldn't be inserted, if I understand how those dialects work. To find out for sure, we'd have to sneak up and eavesdrop on an r-less person while they're reading the post to themselves.
Posted by: The Tensor | June 11, 2004 at 11:52 AM
Stressed schwa, and ash, both in open syllables: these don't occur in English apart from this form of spelling-out, so there's no clearer alphabetic way of representing them. If there's a consonant between it's the glottal stop.
Posted by: entangledbank | June 12, 2004 at 07:29 AM
I would have gone with "kuh" and "tuh" for the stressed schwas, actually. I can't think of a way to represent ash unambiguously in an open syllable in English orthography.
Posted by: The Tensor | June 12, 2004 at 11:17 PM
I find the "uh" representation confusing, because it doesn't occur in English at all except as an interjection of indeterminate sound. I tend to read it as the vowel of "cut" deprived of its coda, and as that's very variable within English accents, it doesn't seem to be much use. The Dilbert cartoons mock "ind-uh-viduals", and as the normal value for the I in that position in an American accent is a schwa, I presume the exaggerated spelling must refer to something other than that.
Posted by: entangledbank | June 13, 2004 at 01:17 AM
What's the difference between letters and phones in lay understanding? In the scenario Sampson presents, surely Johnny is learning to read:
What does "cat" say, Johnny? (Points) [k@] - [a] - [t@] ... [kat]!
Posted by: des von bladet | June 14, 2004 at 04:10 AM
Ooh, Tigger! You've come THIS close to one of my favorite bugbears as a recovering Language Maven---the way so many Americans say the Nazi Minister of Propaganda's name to rhyme with "gerbils". I've always suspected that originated in some helpful Englishman having written "Gerbels" as a pronunciation aid (what do you call that kind of dialect-dependent pseudo-phonetic spelling?), which, when said in RP, is pretty close to the way a native speaker of German would say the name, but when said by most Americans has an "r" that shouldn't be there.
Posted by: Alex the Not-troll | July 25, 2009 at 09:41 PM