Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: more English orthography

From:Roger Mills <romilly@...>
Date:Wednesday, May 17, 2000, 20:02
>Nik Tailor wrote: >>I suppose it depends on which definition you use of "phoneme". The >>definitions I learned and use do not allow two phonemes to share an >>allophone (at least in the same environment). >
Marcus Smith replied:
>That is a definition I haven't heard before.>
Ah me, times do change! Nik's definition is also the one I learned-- pretty much "Classical (American) Phonemics" as it was Pre-Chomsky. It's still a useful tool for beginning analysis, and still works for some languages, like Spanish; but it could not handle neutralization. Theoretically, you were supposed to be be able to deduce Phonemics from Phonetics, _and vice-versa_......
> What would you do about the case >of Japanese /d/ and /z/, both of which become [dZ] befure /u/? There is no >phoneme /dZ/ in the language, so while this is certainly a neutralization
of a
>contrast, I would have to say that it is also a mutual allophone.>
Gnashing of teeth, and fudge, fudge, fudge. IIRC you probably had to posit /dZ/ as a phoneme of limited occurrence. (BTW don't Jap. /d/, /z/ also > [dZ] before /i/?-- and, to pick a nit, isn't it /d/,/z/ > [dz] before /u/? So you CAN have minimal pairs [hypothetical] kutsu : kudzu, or hatSi : hadZi; also [tSa] occurs IIRC.) The lack of [dzo, tsa, tse] et al. would just be a mysterious gap in the phonological system, and that would be of interest to the historical linguist but irrelevant to the analysis of the modern language. It's true that, under any school of analysis, there will be loose ends, like "sounds" (i.e. phonemes) of limited occurrence. My pet, Buginese, has contrastive glottal stop /?/ only in word-final. When certain suffixes are added, /?/ changes to -r-, -s- or -k-.........
>>To put it another way, how would you distinguish between phonemic >>neutralization and allophony, or would you?>
You were only allowed to go by the native speaker's production, without recourse to your knowledge of written forms or history (ditto your informant's knowledge of these: (a) he's most likely wrong or folk-etymologizing (b) as a linguist, you know more than he does. Yeah, right). So we came up with awkward lists like: /'tEl@fown/ 'the form in isolation', /tEl@'fan-/ 'allomorph in the environment of suffix /-Ik/', /t@'lEf@n-/ 'allomorph in the environment of suffix /-i/'; similarly for the Buginese case, **/apa?/ 'the form in environments XYZ', **/apar-/ 'allomorph in env. suffix /-W/'. Generative phonology fixed this problem, at least, by positing underlying forms in which the errant vowels/consonants were "there" underlyingly, then changed/deleted etc. by various rules on their way to the surface. That implies that speakers must have some sort of underlying knowledge of their language's history-- a very big and in most cases incorrect assumption. It could be carried to extremes, like IMHO Chomsky & Halle's SPE which IIRC included the Great Vowel Shift as part of the rules of modern English. And I recall an analysis of Spanish where some underlying forms looked positively Indo- European! But never mind....
>I see the two terms as covering different domains. As I've always
>it (phonology is not my thing, so maybe I've got it all wrong), "allophone"
>for describing variation within a phoneme's possible pronunciations; >"neutralization" is for describing relations between phonemes. >Marcus>
That seems essentially correct. All theories solve some problems, but produce others. I'm lamentably unfamiliar with the most recent theories of phonological analysis, but would suspect a priori that they don't have all the answers either.