[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >


[Date Prev] [Date Next] [Thread Prev] [Thread Next]

RE: [engelang] Self-segmenting words & the treatment of names



Lee:
> >> What's a good way to handle names in a self-segmenting
> >> engelang that constitutes an optimal balance between (a) preserving
> >> self-segmentation, (b) distorting the original name as little as possible,
> >> and (c) concision.
> 
> How important is self-segmentation?  When JCB started in the 50s it
> was a big deal, but today my watch is a more powerful computer than
> he ever used.  It seems to me that if present state-of-the art speech
> recognition software can perform word breaking about as well as a
> human using a reasonable amount of resources, then making that part
> of a language design is a waste of time.

As far as making things intelligible to a computer goes, yes. And
anyway, in actual speech, most speakers would probably violate 
phonetically various of the phonological properties that might
effect self-segmentation: so, for example, I reckon that a computer
equipped with nothing but the rules for Lojban self-segmentation
and the raw acoustic signal would in practise be unable to identify 
many word and morpheme boundaries.

But any computer powerful enough to understand one language is probably
powerful enough to understand any. So the goal of designing a language
that is especially suitable to being spoken to a computer is not
a particularly compelling goal, for me.

So to me, the virtue of self-segmentation is that it is a necessary
ingredient in effecting wholesale absence of structural ambiguity:
it is necessary in otder to guarantee a one-to-one (or many-to-one)
match between phonological strings and sentence meanings.

--And.