On 10/04/09 15:07:28, caeruleancentaur wrote:
> --- In [log in to unmask], R A Brown <ray@...> wrote:
> > 
> > In _catsup_ we have two syllables: cat-sup /k&t.sVp/. The /t/ is 
> the
> > coda of the first syllable and /s/ is the onset of the second.
> > 
> I find it interesting that catsup is derived, through the Malay
> 'kechap' (/kE.tSap ?), from the Amoy Chinese 'koe' + 'tsiap'.  What
> was originally an initial affricate has become for us coda and onset.
> But 'ketchup' is divided differently, /kEtS.Vp/.

This syllabification has caused some disagreement amongst contributors. 
The argument here is that because /tS/ is glottalised, it must be in 
the previous syllable. The counterargument is basically the onset 
maximisation principle, that as many segments should be in an onset as
the language will support. Both make some degree of sense, although I 
think the OMP should win the day:

Renown English phonetician John C. Wells argues that intervocallic 
/tr/ [1] must be a coda because it is glottalised, i.e. a word like 
"Patrick" has to be syllabified as /p&tr.Ik/ because it is 
[p_h&tr)_?Ik] [2]. This leads us to the I think unhappy conclusion that 
/tr/ can be an onset and it can be a coda, but only if a vowel follows. 
Other phonologists have claimed that in non-rhotic English varieties, 
/r/ is likewise permitted in the onset, but can only be in the coda if 
a vowel follows, so that "arid" (say) is [&r.@d]. I forget the exact 
justification for this, but it's at least partly to let the linking [r] 
go in the coda. I personally think this is a funny desire, given that 
the entire reason for a linking consonant is to make sure every 
syllable has an onset! (if no word was before, the onset is of course 
[?]; if a word ending in a consonant was before, the onset the 
consonant that was before --- this is especially clear in varieties 
like RP which alternate between dark [5] in codas and light [l] in 
onsets; a word like "little" ends in [5] unless a vowel-initial word 
follows, when it's [l]).

[1]: in this post, for simplicity, I'll be using [r] to refer to the 
     English approximant rather than the trill, even in the narrowest 

[2]: for Wells, /tr/ is phonetically an affricative, i.e. the release
     is directly into the [r]. I prefer to say it's [tS)r], which
     nicely gets around Wells's problem. The difference between our 
     descriptions might be due to differences between RP and AusE.

[5]: Your pedantry/unreasonable desire for unambiguity has now got you
     stuck in an infinite loop, because I was referring to a voiced
     velarised alveolar lateral approximate, IPA [ɫ], X-Sampa/CXS [5].
     You'll get out of it just as soon as you use common sense! But you 
     already know that if you've got this far.

Anyway, as Ray Brown and Paul Kershaw have so rightly observed, 
fretting about syllabification is probably a waste of time. I think 
there's no reason not to assume a single utterance of a single word 
couldn't have more than one syllabification in the utterer's head.

> BTW, "An Etymological Dictionary of Modern English" says that 
> 'catsup'
> is incorrect.

Yeah, but so's "ketchup". The stuff's called "(tomato) sauce". Or, on a 
dog's eye (meat pie), "dead horse". When I worked in the restaurant in 
Ikea there were lots of kids who either didn't know what "ketchup" 
meant or clearly took pride in showing off their knowledge to their 
siblings/friends. (Yeah, the reason I replied to this post was to make 
sure you know I'm Australian ;)