Re: Cowan on morphology

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

> And's logic escapes me.  If meaning is not predictable, then self-segregating
> morphology has no value at all???  And it ADDS complexity to the language
> and the task of acquiring vocabulary.

It has a little value, but is not remotely worth the cost of Lojban's
morphological complexity.

It undeniably adds to the complexity of the language. It adds to the
complexity of learning gismu. It is not the only possible solution
to the task of learning compounds: compounds could be formed by
regular principles without necessarily being self-segregating.

> Based on what it replaced - the old Loglan system of cramming words together
> in whatever way seemed useful, the Lojban system is immensely better AND
> easier both for learning vocabulary, AND for inventing new vocabulary on
> the fly - aprocess that will occupy Lojbanists for many years while the
> vocabulary remains much smaller than that of English.

What is needed for inventing new vocab on the fly is regular rules
for compounding, not necessarily self-segregation.

> Having the self-segregating morphology means that you  need to memorize
> the morphological roots that are unique, but as I have argued often before,
> this is not that onerous a task since the optional root values are from
> a limited set of forms derivable from the gismu, and you can always use
> the expanded form that is unambiguously associated with the gismu for any
> listener (THE thing to do when you are writing or speaking to an audience
> that may not knwo the rafsi well enough to dissassemble your creation, or
>  deducethe meaning from context).

One-to-one communication where each participant makes allowances for
the competence of the other is only a small fraction of communication
in a language of a modern culture, Lojban included, probably. When
Nik or Colin, for example, post in Lojban to this list, they don't
use expanded rafsi, even though they know that significantly fewer
readers of the list know rafsi than know gismu.

> However, with that morphology you have significant clues as to meaning, and
> moreover, because of the self-segregating quality, you know that a large
> body of meaniungs is excluded.  i.e if the final morphological term is a
> rafsi for klama (come/go), you know beyond all doubt, that the word relates
> to going/coming of some type, and not to being blue, beating your wife, or
> compiling a computer program.

You don't know beyond all doubt. Even if the last rafsi is identified as
klama's, the lujvo could refer to sapphires, to pick a random example.

Say the non-self-segrating, but simple & regular rules for compounding
require that for the final morpheme you strip off the initial
consonants of the gismu. In this case, the last morpheme of a compound
ending in _-ama_ is going to stand for either _klama_ or _jbama_
(bomb). Now I would gladly sacrifice tthe complexity of the
self-segregation (i.e. having to learn the rafsi & tosmabru rules, etc.
etc.) for the price of having to guess from context whether the
compound refers to a kind of going or to a kind of bomb (assuming I
was unfamiliar with the word).

-----
And.