LINGUIST List 3.497

Mon 15 Jun 1992

Disc: Natural Phonology

Editor for this issue: <>

Directory

1. Richard Ogden, natural phonology
2. John S. Coleman, 3.494 Natural Phonology
3. Rick Wojcik, Re: 3.483 Natural Phonology

Geoff Nathan says that natural phonology is "the only theory that has
any interest in any level of representation below the level of the
surface phoneme".

I would liketo say that this is not true. Firthian Prosodic Analysts
were always interested in minute phonetic details and were never
interested in phonemes as phonlogical objects. They constructed
analyses without the mediation of such units, directly from the
'phonic material', which was usually far more detailed than ever
appears in most 'modern' phonological analyses.

Richard Ogden
University of York
England

Geoffrey.S.Nathan writes

> [Natural Phonology] is the only theorythat has any interest
> in any level of representation below the level of the surface
> phoneme

This is untrue.

> (most discussions of post-lexical rules do not include
> ANY of the interesting (inneressin) things that happen in real
> speech production, even though they are rule governed and language
> specific.

A matter of opinion about what are the interesting things that happen
in real speech production. I venture to predict that Nathan cannot
name more than one or two "interesting things that happen in real
speech production" accounted for by Natural Phonology but not
addressed by someone else in some other framework.

--- John Coleman

Message 3: Re: 3.483 Natural Phonology

Joe Stemberger writes:

> The main problem seems to be with the notion of Natural Process. There
> seems to be an implication that there are processes that should be common
> to all children. Stampe says that there are phonological items/sequences
> that are hard to pronounce, and that the natural processes get the children
> over these difficulties.

Bear in mind that the historical basis for this observation goes far beyond
Stampe. The seminal work was Jakobson's Child Language, Aphasia, and Phonolo-
gical Universals, and even Baudouin de Courtenay had some remarks about the
universal similarities found across languages in babytalk. Don't forget that
these observations were also the basis for markedness theory in generative
phonology. So more is at stake here than just Stampe's theory.

> But there is very little uniformity across children in terms of the
> processes that they use. There is a fair amount of uniformity in terms of
> what children find difficult. For English-speaking kids, at least, we can
> say that syllable-final consonants are difficult, that unstressed syllables
> are difficult, that voicing in final obstruents is difficult. But children
> seem to "solve" the same problem in different ways.

This is fully compatible with what Natural Phonology says. One might expect
more similarity if children were required to solve exactly the same articula-
tory problems in exactly the same sequence, but the theory does not claim this.
The same is true in L2 acquisition--not all foreign accents are the same and
some people have more difficulty suppressing native processes than others.
This doesn't mean that English speakers, e.g., all start out with different
phonological systems. All NP says is that the pristine set of processes is the
same, not that children all arrive at the same phonology via the same route.
I suspect that part of the problem lies in the fact that different vocabulary
presents different articulatory problems. Does everyone learn the same
vocabulary at the same stage of acquisition?

> Consider final voiced obstruents. Some children devoice them. Some children
> delete them (while not deleting final voiceless obstruents). Some children
> prenasalize or postnasalize them (or, rarely, nasalize them completely).
> Some children epenthesize vowels after them. Some (rare) children replace
> them with a reduplicated syllable; e.g. PICK is [bik], but PIG is [bibi].
> This list is probably not exhaustive.

Natural Phonology does not take the position that a given phonetic target is
attacked by one, and only one, process. The NP literature goes to
great lengths to say just the opposite. For example, those sounds which
Jakobson found to be rarest from a cross-linguistic perspective are those
sounds that are impeded by more processes than the common ones. But perhaps Joe
is arguing that there is a purely random distribution among children as to
which consonants and vowels will be mastered first. Jakobson's observations
were absolutely contrary to what language acquisition specialists observe
today. Is this true, Joe? If not, then what is the most likely explanation
for the general trends observed in child pronunciations? (I agree that
Jakobson could not explain variation, but I am asking for acknowledgment that
there was *some* validity underlying his observations.) Stampean theory, unlike
Jakobson's, can explain both the generalizations and the variation.