Re: Phoneme distribution

Jeffrey Henning wrote:
>
> > It is quite common to see tables of how different letters are distributed
> > in typical texts of a given language, but have there been any similar
> > investigations regarding phonemes of the spoken language?
1.  Record a large corpus of language
2.  Transcribe it phonemically.  Or, optionally --
    transcribe it orthographically, and do a search-and-replace with a dictionary
    which contains phonemic representations.
3.  Count characters.

Same principle....

> Not of one language.  I've seen analysis of the frequency of phonemes
> *across* languages (the UPSID data).
>
> > There are a few things I am particularly curious about, such as whether
> > the various phonemes of a language typically follow some sort of pattern.
> > If, for instance, any given phoneme typically occurs in between 1% and 45%
> > of the words of a language (this example is, of course, total nonsense).
Oh, yeah.  I mean, this is behind the fact that certain letters occur
more often than others...  If the language were written phonetically, it
would be the same as a letter count, and letter counts turn up patterns
of frequency.

Within a langugage, by the way, unmarked phonemes tend to occur more
often than marked ones.  Cross-linguistically, ejectives are fairly
marked compared to plain stops, so you can predict that you're gonna
hear a lot more plain than ejective stops within a language that has
both.  Same for voiceless vs voiced nasals, and all other
marked/unmarked contrasts...

Speaking in broad and fallible generalizations of course. :)

> Yes, this can happen.  Which is why when I was releasing an update of
> LangMaker/Win language creation software I added the capability for people
> to weight particular phonemes to occur more often.
Is this a recent new update?

> > Or if diphtongs (when a language has such) are usually much less frequent
> > than single vowels, etc. (Hmm... perhaps a diphtong would count as a
> > morpheme rather than a phoneme? Oh, well, you know what I mean.)
>
> Don't know about this one.
Me neither, but counting the diphthongs vs pure vowels in the sentences
above, pure vowels seem to come out way ahead.  And American English is
supposed to be a diphthong-heavy language.

Ed

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com