Re: Cowan on morphology

[Mr Andrew Rosta <ucleaar@UCL.AC.UK>]

Chris writes:
> And writes:
> > A more rational design would make words self-segrating, but not morphemes.
> >Even if morphemes fail to self-segregate, they would still serve as clues
> >to the word's meaning.
> >
> As I see it, and as a very much peripheral Lojbanist I am more than likely
> wrong, the problem is not that morphemes may not self-segregate. but that
> they may segregate differently from the intension. Although the meaning of a
> word is accessible from a dictionary, we normally only consult a dictionary
> if we are unsure - we may be wrong but unaware of it.

This simply makes the words countermnemonic.

> I know this does not seem like the same problem, but I think it is. If we
> wish to put together an unlikely string of words in Lojban, then we may, and
> we can then attempt to puzzle out the meaning. If we wish to put an unlikely
> combination of morphemes to gether, than we should be able to. If
> 'legitimate' words pre-empt unlikely words by trespassing on their space,
> then we lose the ability to create the unlikely words.
>
> Remember -- colourless green ideas sleep less furiously in Lojban than the
> do in English.

You can still have compounds with strange meanings. If it can be
morphologically parsed in some alternative, more plausible, but
wrong, way, then the compound is that much harder to learn. But
personally I don't have much problem with the word _understand_,
even though if one didn't know the word one would surely guess
it means "stand under".

Perhaps if there were an elegant solution to the self-segregating
morphology problem then it would be mildly useful, to avoid the
problem cases Chris envisages, but these mild sources of awkwardness
don't justify the baroque ghastliness of Lojban morphology. If
the morphology of Lojban were exported to Klingon, it would be
most appropriate.

----
And