Iversen wrote:Cainntear wrote:That's something I've tried to replicate in every language since: get the archetype of the phoneme first, knowing full well that phonemes all interract to lead to various allophones. (...) I'm focusing on doing the right phoneme first so that I can leave the allophones until later.
If phonemes behaved like letters then I would also tend to support this,
In what ways do they not?
but sometimes the allophones of one phoneme overlap with the allophones of another - as with the unstressed Russian о's that sound exactly like unstressed а's. Is there actually any difference in pronunciation between *малоко and молоко in ordinary sloppy Russian? The definition of a phoneme tells you that using one phoneme instead of another has semantic consequences, but I do get slightly worried when a phone can be interpreted as the allophone of two different phonemes. It's like having two hotel rooms with a door between them, and then someone has left that door open.
Yes, and using one grapheme instead of another has potential semantic consequences -- if I miss a line out of an "m" I get an "n". You can argue that this has no semantic consequences because there is no such word as nunber, but it has
potential semantic consequences, because a miswritten "sum" can look like "sun".
This is an example of where graphemes can be similar enough to cause problems. One of the key differences between graphemes and phonemes, though, is that it's moderately easy to alter graphemes to make them clearer -- now, with universal schooling, you can switch people's handwriting quite drastically within a single generation. Doesn't work that way with phonemes, though.
And it didn't always work that way either. In the earliest days of the English language's use of the Latin character set, there was an issue bwith handwriting where words had sequences of multiple half-height ascenders in a row -- words like sun, summer etc. I'll demonstrate the problem with printed characters: sıııı, sııııııııer. Scribes in the middle ages wrote a variant of U in such circumstances where they connected the top of the U to be like a lot of modern handwritten As, although the lower-case A was usually written in a similar way to how most fonts render it today.
For this reason, a lot of moderately modern scholars looked at old manuscripts and decided there was a song called "Somer is a-comin in", but that's a U, not an O... ie "Sumer is a-comin in". Oh yes, and that other O wasn't an O either: "Sumer is a-cumin in." The difference is that the written convention of connecting the top of the U was so pervasive that people ended up believing that "come" was written with an O. There's a pervasive misunderstanding now that "come" was once pronounced with an O and the spelling has stayed the same despite the pronunciation changing, but as I understand it, that's not the case (which I do find hard to accept, given that Hochsdeutsch goes with "komen", but I'll defer to my lecturers on that one!) -- it was always a U phoneme, and it used to be a U grapheme, just it was an "allograph" that looked confusingly similar to the O grapheme.
For the most part, orthographies tend to be driven by a desire to avoid potential confusion of graphemes that phonologies are pushed towards making phonemes more readily distinguished from each other...
This is different from the situation where a certain sound can be written in several ways - like with ι, η, υ, ει and οι in Modern Greek, which all are seen as representing the same phoneme, while they were pronunced differently in older stages of the languages. When I have to write in Greek I do hear the i-sound, but I also have like a faint shadow of a ghost-sound telling me what letter to use - but this is a problem that only concerns the writing, not the phonematic system. I may have the same problem with the first vowel in молоко/*малоко, but here I should in principle have to know which phoneme the sound represented - though in this case one of the options luckily doesn't seem to exist. Though maybe the choice of phoneme doesn't matter because I can hear that someone is saying the word for milk, and I know how to spell it, should the occasion arise.
Let me finally take a third case which supports Cainntear's stance: Italian vowels. Sometimes their e's and o's are open, sometimes closed, and it's all about which dialect you are listening to. But in practice a tourist isn't expected to know how they are pronounced before actually settling down in a certain region - and then you just have to listen to find out how the local people in that place pronounce their e's and o's - if you care at all, that is. Otherwise you just take a bold guess and do your best. The Italians are used to people pronouncing their vowels wrongly in different ways.
