Re: THEORY: And wrote:

Rob Nierse:
> And wrote:
> >> In this languages it is misleading *not* to describe the vowels
> >> in oral and nasal
> >
> >But the point is whether to treat a-oral and a-nasal as basic
> >phonological elements, or whether to treat |a| and |nasal| as
> >basic elements
>
> So, how do we make this decision?
List the rules that state the basic elements. List the rules that
state their combinatorics. Combine the two. Compare the result
to that you get for alternative analyses. The best analysis is
the simplest.

(We also need a list of rules that state the phonetic realization
of a given phonological structure, but I'm abstracting away from
this.)

> >> In Dutch I make a difference between 'pot' and 'bot'
> >> Do I use different phonemes ('p' vs. 'b') or do I use
> >> different versions of one sounds, i.e. a bilabial stop?
> >> I think the first. Or is my thought the result of education?
> >
> >The question is whether the elements out of which phonological
> >structures are composed are /b/ and /p/ or (say) |labial-stop|
> >and |voice|. For a language like Dutch, the latter does not
> >lead to much simplification level, at least at a superficial
> >level of analysis
>
> Is simplification the only argument to divide phonemic elements up
> into smaller parts?
Simplification as a major ingredient of capturing the underlying
logic of the system.

Going beyond the scope of this discussion, a major reason for
recognizing subsegmental elements is to state generalizations
about phonetic realization and about combinatorics.

> So, say, Dutch lacks the voiced velar stop. Does this mean we
> can analyse Dutch as:
>   voiceless stops: p t k
>    and nasal versions of voiceless stops ('nasalisation?) m n N
> versus
>   voiced stops: b d
>
> It is simpler, but somehow I think it is tricky
The gain is marginal. By treating nasality as a separate element,
you dispense with m,n,N. By treating it as combinable with p,t,k
instead of b,d,g you avoid the problem of having to stipulate
that g must combine with nasality. But countervailingly, your
combinatorics have to allow for elements to be simultaneous
instead of sequentially concatenated. So there's not much to
be gained from the suggested analysis in this instance. And all
that's before we even think about which version gives us the best
story for phonetic realization.

--And.