ellis@spar.UUCP (Michael Ellis) (12/02/85)
> reilly > Perhaps it is an ad hoc move to claim that /s/ is syllabic in >English. But that still does not explain what the /s/ is doing >there in the first place. What I mean is that if one accepts the >Sonority Hierarchy as defining the canonical syllable as an increase >up to the vocalic nucleus and a decrease in sonority (by sonority >hierarchy, i mean stop < low fricative < high fricative (eg /s/) < >nasal < liquid < glide (r,y,w) < vowel constitutes a hierachy of >sonority--except for nasals, this corresponds basically to how open >your mouth is!), then you have to face the question: what is the >s doing there in words like sky, scrunch, etc. Assuming that syllable nuclei occur at points of highest sonority, this approach is fairly successful at predicting the general features of phonetic structure in most languages: /trat/ or /tart/ are indeed much more likely syllables than /rtat/ or /tatr/, for example. Similarly, `carl' /k<a>r>l/ is more likely to be monosyllabic than `collar' /k<a>l<r/, which often `breaks' into the disyllabic /k<a>l<0>r/ However, the frequency of English /sk-/, /st-/ and /sp-/ is not the only problem with `sonority hierarchy'; where are all those initials like /ts-/, /pf-/, /fn-/, /dn-/ that `ought' to exist? The only such sounds that are common in English are /tsh-/ and /dzh-/ (CHew, Joke). A reverse problem occurs in finals; if /tsh-/ is a `good' initial, then /-sht/ should be a `good' final, yet /-sht/ is rare in most languages I know. Likewise, the frequency of /-ps/, /-ts/ /-ks/, (and /-bz/, /-dz/, /-gz/ in English) is as inexplicable as that of /sp-/, /sk-/, /st-/. I think it's fair to say that /s/ (and /z/, in English) is an unusual sound in most (european) languages. >Selkirk says you can dispense with the major features (consonantal, vocalic, >syllabic) by using sonority, and I think she is right. If you have the time and interest, I'd enjoy hearing more about this. -michael
reilly@iuvax.UUCP (12/08/85)
Sure, I have lots to say on all these things. Frequencies can tell you lots of things. First, they'll tell you that onsets (beginnings of syllbales) differ vastly from codas. consonantal clusters in onsets are constrained to differ MORE in sonority than codas. This is why you find neither pf- nor fp- frequently in languages. And when you DO have sounds like tsh and dzh in English, they are called "affricates". This means that the timing of these events are different (an old debate: do you have one or two segments) and no phonologist in the world is going to give you a theory of timing. you're going to have to get more phonetic to investigate that angle. Notice that sk- sp- and st- do form a sizable enough contrast along the scale of sonority, just as mpa- does with nasal homoganics, even thought the sonority is increasing the "wrong" way. Nasals involve a very slow, temporally imprecise gesture (the velum), and are more common, i bet, in codas where languages in questions have consonanatal codas (a lot don't-- English is VERY exotic in its wealth of consonantal clusters in codas.) You can further see things just by looking at the list of permitted clusters in onsets and comparing them to permissible codas in English. The permissible codas far outnumber the onsets. Why should this be so, unless onsets are more constrained by principles than codas? And yet it is well know that the distribution of features is more random (informative) at the beginning of words (& therefore onsets) than the ends of words (and therefore codas). So you have a seeming paradox. If you are interested, i can send you a copy of the paper i presented to the lsa last christmas which deals with precisely these issues (and to boot, shows how Selkirk's ideas fit in). It is on file, i can send you a copy. will reilly linguistics, IU