Re: Phoneme analysis. (Motherhood statements?) (Long!)

Ruud Harmsen wrote:

Can you illustrate this with examples? Without them, I really don't
understand this. May be just me.

I'll do my best.
Schematically (I don't know British English):

"roar" , "roaring"

The hypothetical language. There is no syllable onset [r].

1. Insertion analysis (best analysis)

/O/, /O+IN/
/r/ is not a phoneme.
Constraint on distribution of /r/: No constraint (naturally).
Process: Insert [r] before vowel initial syllable when
certain vowels precede (with other conditions?).
[O], [O.rIN]

2. Deletion analysis

/Or/, /Or+IN/
/r/ is a phoneme.
Constraint on distribution of /r/: Not in syllable onset.
Process: Delete [r] in the offset of a syllable.
[O], [O.rIN]

The real language. There is a syllable onset [r].

3. Insertion analysis

/rO/, /rO+IN/
/r/ is a phoneme.
Constraint on distribution of /r/: Not in syllable offset.
Process: Insert [r] before vowel initial syllable when
certain vowels precede.
[rO], [rO.rIN]

4. Deletion analysis (best analysis, IMHO)

/rOr/, /rOr+IN/
/r/ is a phoneme.
Constraint on distribution of /r/: No constraint.
Process: Delete [r] in the offset of a syllable.
[rO], [rO.rIN]

Greg

.

Relevant Pages

  • Re: Explanation of Maximum Entropy
    ... > best distribution pay at least a little attention to the data? ... MaxEnt isn't very useful for the coin toss example. ... non-trivial constraint derived from the data that would not determine ... parameters of a Beta prior, it would prescribe you _symmetric_ ...
    (sci.stat.math)
  • Re: Generate a sequence of random numbers that sum up to 1?
    ... fumanchu wrote: ... It seems to me that the only constraint on the randomness of my results ... where the very distribution of each value is ... but my intuition suggests it's unlikely to ...
    (comp.lang.python)
  • Re: uniform distribution of P points on a (D) dimensions sphere
    ... be distributed "uniformly" in dimensions, ... constraint mentioned in your first paragraph should be equally "optimal", ... Given a solution to that problem, you can scale each vector by ... distribution of the resulting points is rotationally invariant. ...
    (sci.math)
  • Re: Fitting characters to a grid
    ... have to comply with the rules of the grid, so you will not have a truly ... random distibution on the second row - rather a constrained ... distribution. ... the degree of constraint will ...
    (microsoft.public.excel)