Mark P. Line, On 02/02/2006 22:03:
> Jumping in here briefly without having been able to make time to give
> Dick's draft a proper read, in between sneezy fits ...
> 
> 
> And Rosta wrote:
> 
>>Syllables. As far as I ignorantly know, the evidence for syllables is
>>pretty scanty. The evidence for onsets and rhymes is strong, but what is
>>the evidence for saying these jointly form a unit that must be formally
>>recognized by the model?
> 
> 
> I think it's pretty well established that the English lexicon is partially
> coded by number of syllables, and I don't know of any other way to
> describe the effect without positing syllables.

Counting nuclei gives the same result as counting syllables. I reckon that when linguists and laity say "this word has 3 syllables", they have actually been counting nuclei.

>>"There is nothing unrealistic about assuming that a child memorizes each
>>syllable individually." For Japanese, Swahili, Italian -- OK; but for
>>*English*?? Not only are there zillions of syllables, English speakers
>>don't even have any but the feeblest intuitions about which are the
>>syllables any given word is made of.
> 
> So it would depend on whether you're wanting to describe English, or
> English speakers' intuitions about English. (Unless you happen to still
> subscribe to that old saw about these really being the same thing...)
> 
> If a cognitive model works out right and uses syllables, then an apparent
> lack of metacognitions about syllables wouldn't bother me too much, any
> more than the lack of metacognitions about VOT or vowel length bothers me.

Fair enough. If syllables exist, they are a highly theoretical notion, not accessible to our intuitions; and a claim that speakers memorize syllables must be supported by a clear definition of what counts as a syllable (e.g. which strings of phonemes do and don't constitute syllables).

--And.