Print

Print


For me, I'd want a self-segregation algorithm that maximizes utterance brevity. So, here's a question I invite you to consider (because I'd like to know the answer...):

1. Assume, for the purposes of experiment, that the frequency curve for English words (i.e. frequency on one axis and on the other axis the number of words with the given frequency) is universal for all language. (Ignore derivational morphology and homonymy.)
2. Assume that the minimal phonological unit is the syllable, and that a self-segregation algorithm classifies them as word-initial/noninitial or word-final/nonfinal.
3. If the engelanger's goal is then to achieve maximal brevity of utterance, then what word-length curve (number of syllables per word, against number of lexemesa of that length) gives the shortest utterances? (E.g imagine you had to recode English text, word-by-word, into the new phonology.) (Assume self-segregation and no homonyms.)
4. The answer for (3) will vary according to the number of syllable types. But is it possible to state a rule (of thumb) that generalizes over different numbers of syllables types, such as "generally, n% of syllable types should mark word-finality"?

I hope the question makes sense and is of interest...

--And.

Gary Shannon, On 09/09/2010 16:49:
> Words are made up of any number of CV syllables where C is a glottal
> stop, a single consonant, or any one of a number of permitted
> consonant clusters (as yet unspecified). The first syllable may have a
> null consonant, i.e. V only.
> 
> The first vowel of a word is any vowel other than 'a'. All of the
> remaining vowels of the word are the vowel 'a'. For example:
> 
> 	diva, ropa, upasana, purampada, toskala, osa'atanda ...
> 
> The accent falls on the non-a syllable.
> 
> Compound words are easily recognized and parsed since they will have
> more than one syllable with a non-a vowel. For example:
> 
> 	otampaposata can only be otampa + posata since that is the only
> partition that satisfies the morphology of roots.
> 
> 	sikalakoranita -> sikala + kora + nita
> 
> If a compound joins a vowel to a vowel, e.g. chupa + otaka, a glottal
> stop is inserted: chupa'otaka.
> 
> I can't imagine a simpler system than that.
> 
> Now all I need is the simplest possible grammar to go along with this
> and I can build the simplest possible conlang. :)
> 
> --gary
>