On 25 October 2010 15:28, And Rosta <[log in to unmask]> wrote:
> Peter Bleackley, On 25/10/2010 10:17:
>> staving Jörg Rhiemeier:
>>> The sound change of the type /sp/> /b/ looks unlikely to me;
>>> I'd rather expect something like /sp/> /f/ etc.
>> The idea came from the fact that when my children were first learning
>> to talk, they would often render /st/ as [d].
> So did my son. Or rather: in adult English in his environment (contemporary
> outhern England) onset /t/ is [t_h] and onset /d/ is mainly [t] and
> sometimes [d], and "/st/" (i.e. IMO /s/) is [st]. He had the [t_h]:[t]
> contrast, so when he had trouble with "st", it came out as [t] (= /d/, I
> would argue). (When trying extra hard, he'd come out with [ts].)
I'm wondering about that (the argument that onset [t] is /d/ in English). As
a native French speaker, my /t/ is [t], and my /d/ is [d]. If English /d/
was really [t] at word onset, I would have expected to mishear if for /t/,
at least when I started learning English. Yet I never did (despite the fact
that I originally couldn't hear the aspiration of [t_h]), and I don't know
any French student of English who ever mistook an English onset /d/ for /t/.
English onset /d/ for me never sounded different from French onset /d/.
This leads me to think that English onset /d/ must be at least partially
voiced (the Wikipedia page about voice onset time:
http://en.wikipedia.org/wiki/Voice_onset_time agrees with me here), and the
contrast between onset /t/ and /d/ in English is not *only* a question of
aspiration (rather, English onset voiced stops have a negative VOT, rather
than zero, and voiceless ones a positive one).
Now, the question of whether to analyse the [t] in "st" onsets (which I
assume has a VOT at or near zero) as /t/ or /d/ depends on the value of the
VOT at which point a fortis stop becomes lenis (I use those terms in order
not to get sidetracked by the terms "voiced" and "voiceless").
For French people, clearly, any stop with a VOT at or near zero, or positive
counts as a fortis stop (hence both the French [t] and the English [t_h] are
heard as /t/ by French people), while any stop with a negative VOT (but not
necessarily coinciding with the onset of the stop) is heard as lenis (hence
both English onset partially voiced "[t]" and French [d] are heard as /d/).
But how is that then with English people? Does the switch between fortis and
lenis occur at a positive VOT (in which case [t] in "st" should indeed be
analysed as /d/, and aspiration is indeed the salient feature of onset
stops) or at a zero or negative VOT? (in which case the [t] in "st" should
be analysed as /t/, assuming it has indeed a zero or near zero VOT). Has
anyone ever done an experiment using synthesised speech with onset stops at
various VOTs to test at which point an English person stops hearing a /t/
and starts hearing a /d/? And is the [t] in a "st" onset indeed completely
voiceless or partially voiced?
This is a place where phonetics could have a big influence on phonology, and
indeed on diachronic phonology. Indeed, if the [t] in "st" is partially
voiced, a sound change /st/ -> /d/ becomes suddenly more plausible,
especially if the people who lose the onset /s/ feel that an onset stop is
voiceless if and only if its VOT is strictly positive. If not, then I can't
see it happening that easily (the only other way I could see it happen then
would be if an /sCV/ onset took on an epenthetic vowel, lost its /s/, and
then stops got voiced between vowels).