Re: Thoughts on Word building

Hi!

Taka Tunu writes:
> My post was precisely meant to discuss derivation vs. compounding.
Yes, of course. :-)  I read your post. :-)

> <<<
> > Chinese is full of very different kinds of compounds, where the meaning often
> > cannot easily be derived from the parts -- the new meaning is ad-hoc and then
> > lexicalised, and not predictably derivable. It usually does not make
> > use of derivation, but of compounding.
> > <<<
> > I have an opposite experience with Sino-Japanese vocabulary and for
> > my conlang I personnally use the very consistant compounds that do
> > exist. But you are free not to and to pick the least consistant
> > ones.
> Hmm, hopefully someone has figures, otherwise discussing what is more
> frequent is probably leading to nothing but, errm, discussion.
> >>>
>
> I don't get your point here. Please clarify.
We'd need numbers counting compounds vs. derivation in Chinese and
Japanese.  Without them, we should not discuss who is right about
which one is the major concept.

> It is needed to take into account that Chinese compounds are
> oriented as "verb-object" AND "modifier-modified" AND
> "verb-subject." Hence "love+country" ("to love one's country") and
> "king+country."
Yes.

> A conlanger could make up "country+love" and "king+country" (or
> reversely "love+country" and "country king", depending how his
> conlang works.) ...
But thes is not derivation, then, do we agree?  This as compounding.

> My own list has entries for "sibling", "compute (data)", "machine",
> "terrain", "diagram", "institute", "learn/teach", "country", "love",
> "king." My conlang compounds "S-V", "V-O", "modified-modifier."
Ok.  'Compounds'.  Not 'derives', then.

> <<<
> There are some derivational suffixes, I think, e.g. 'zhe3' in Chinese
> for the agent ('-er' in English): 'xue2zhe3' - 'learner'. Searching
> Cedict does not seem to reveal any exception of this referring to an
> agent and '-zhe3' being suffixed to anything but a verb. Anyway, I
> think that that is an exception and that the majority of words are
> ad-hoc compounds. But we'll need figures for comparison.
> >>>
>
> "Agent"? What's that? :-)
> I use "person", "workman", "specialist", "master", "disciple", etc.
But these are again compounds.  I thought you said we're talking about
derivation here (see the beginning of my paragraph).  I'm confused.
Are you saying these are derivation?  Or do we agree these are
compounding?

Chinese has these, too, of course, but only 'zhe3' could be classified
as a derivation, since it's a) bound, b) clearly defined what happens,
c) is restricted to being added to a certain verb class (verbs).  The
ones you list are not bound, can be used alone, and are not restricted
to certain word classes.  And the completed meaning has to be guessed,
it is not clearly derived.

E.g. let's use 'person' - 'ren2', and compare this to some other
compounds:

   ai4-ren2      - lover; spouse      bi1-ren2     - threatening
   verb + noun                        verb + noun

Same structure: verb + object, different effect: the first one is the
noun specifying the object, the second is a compound adjective.  So
'ren2' does not strictly derive the meaning in a predefined way: it
has to be guessed and is ad-hoc.  'ren2' can be used alone, too, so it
is clearly compounding.

   ai4-gou2      - patriotism
   verb + noun

Compare this is 'ai4-ren2'.  It is also verb + object, and we cannot
say that 'ai4' works in the same way, so 'ai4' is also compounding,
so we cannot classify 'ai4' as a derivational prefix either.

Both 'ren2' and 'gou2' can be used alone and compose ad-hoc meanings
instead of in a well defined way.  Both 'ren2' and 'gou2' can also
attach to nouns, so they are not restricted to certain contexts.

   nü3-ren2   - woman
   wang2-gou2 - kingdom

Japanese is slightly different here, since '-jin' is bound -- alone,
it would be 'hito'.  So here, the same Kanji has two readings and one
of them is bound.  So '-jin' is definitely closer to a derivational
ending than 'ren2' in Chinese.

>...
> There are kanjis for "institute", for "official building", for "master", for
> "disciple", for "learning" and "superior" and so forth, so "school" and
> "university" are kind of at reach out there for a conlanger, if you clarify
> first what you mean by "school"? An institution? A building? A corporation?
>...
Ok, we disagree.  Can we agree that a) they are certainly a source
very worth looking at and b) conlanger might want a different list for
some reason?

If that's a minimum we could agree on, I'd add for myself:

   ad a)
      Be aware these are not derivational endings, but form compounds.
      So when seeking lists of derivational endings, keep this in mind
      and possibly narrow down and clarify usage.  And if you really
      want bound morphemes, even think of a bound form for each Kanji
      used in derivation in your conlang.

   ad b)
      I'd say a typical reason could be historical development where
      the conlangers engelang demands a systematical grid of atoms or
      something.

>...
> One question underlying this thread is where to stop breaking down the
> vocabulary into "roots" or "stems" or "affixes" and reversely, where to stop
> multiplying them? Criteria may vary: realism, idealism, "conciseness", poetry,
> etc.
Yes.

> That is personal taste, granted, but shouldn't conlangers be happy to find that
> some natlangs have already "rounded up" their own vocabulary, not into a
> 30,000-entry dictionary, but a merely 2,000-entry one?
YES!  Of course.

> As a matter of taste, I find this "kit" invaluable because it's
> well-tested and still in use and I don't understand why this is not
> taken as much into consideration than the other useful ones like the
> "Basic English" lexicon or what else.
Of course, it's a very good source of ideas.  I use it myself all the
time when trying to make up vocab.  I just say that in general, it is
compounding, not derivation.

**Henrik