Re: Theory about the evolution of languages

Jim Henry said:
> "Mark P. Line" <mark@...> wrote:
>
>> Jim Henry said:
>
>> > Esperanto has at least one prefix (mal-) and several modifier
>> > particles (ne, jam, ankaux, ecx...) that can precede just about any
>> > word.  A language that has non-syllabic clitics with the same meaning
>> > or role as those is readily imaginable, though I don't know of an
>> example
>> > offhand.
>
>> Where can the 'ne' go in more complex sentences? Can it really go before
>> any word at all? (I'm disinclined to believe that *any* segmental form
>> can
>> go just anywhere in any natlang, but of course most bets are off in
>> conlangs.)
>
> I was obviously overgeneralizing.  I think that these particles can
> in theory go before any kind of word, but not in any possible
> context.
Well, before we lead ourselves too far astray, we should recall that I was
saying that clitics are not forms that can go "just anywhere". I was
talking about natlangs, of course, and as somebody else pointed out, most
bets are off in any event when looking at conlangs.



> I've numbered your sentences for commenting on them later.
> Most of them seem to be grammatical but nonsensical;
> "ne" is forbidden in these contexts by pragmatics or semantics
> rather than syntax rules, if I understand the use of those
> terms correctly.
We've probably crossed the boundary of theory-neutral discussion here.

I'm an empirical linguist of the corpus-based variety, so for me, a form
(lexical item or construction) occurs in a corpus with a certain frequency
-- perhaps exactly zero (which doesn't imply that it would never occur in
a different or more inclusive corpus) or perhaps more or less frequently.

So for me, a clitic can go "just anywhere" if I fail to find any way to
formulate coherent distributional constraints that describe where the
clitic does occur in the corpus. (This is oversimplified, but it might
serve for now.)

You probably didn't intend to disagree strongly with that (and certainly
not with respect to natlangs), and we both realized you were
overgeneralizing, but this might still be an interesting thought
experiment nonetheless.

(Remember that we're really talking about Esperantissimo, a language that
is just like Esperanto except that 'ne' is a proclitic.)


>> 0. (Se vi havas aux konas informado-rilatajn TTT-ejojn en la reto,
>> bonvolu
>> informi min pri tio.)
>
>> 1. Ne se vi havas aux konas informado-rilatajn TTT-ejojn en la reto,
>> bonvolu
>> informi min pri tio.
>
> Nonsensical.  I'm not sure that "...ne se..." is impossible
> in any context, but I can't think of a case where "Ne se..." would
> be a meaningful sentence opener.
So 'ne' doesn't occur before sentence-initial 'se' (and there's no
coherent way to describe what it might be intended to mean in such a
context if we did happen to run across one). That's a distributional
constraint, and it tells us that 'ne' can't go just anywhere.


>> 2. Se ne vi havas aux konas informado-rilatajn TTT-ejojn en la reto,
>> bonvolu
>> informi min pri tio.
>
> Also doesn't make sense, though "ne" can negate a subject pronoun in
> other contexts.
Ditto, then.


>> 3. Se vi ne havas aux konas informado-rilatajn TTT-ejojn en la reto,
>> bonvolu
>> informi min pri tio.
>
>> 4. Se vi havas ne aux konas informado-rilatajn TTT-ejojn en la reto,
>> bonvolu
>> informi min pri tio.
>
> "ne aux" (and "ne sed", "ne kaj" etc.) would be wrong
> in all context I can think of.
<event class="beat" ref="foobar">
  <object role="patient" class="horse" state="dead">
That's what I thought. Clearly, 'ne' can't go just anywhere. :)
  </object>
</event>


>> 9. Se vi havas aux konas informado-rilatajn TTT-ejojn en ne la reto,
>> bonvolu
>> informi min pri tio.
>
> Wrong. I can't think of a case where "ne" between a preposition
> and its object would make any sense.
<foobar/>



>> 10. Se vi havas aux konas informado-rilatajn TTT-ejojn en la ne reto,
>> bonvolu
>> informi min pri tio.
>
> Probably ungrammatical.  I don't think "ne" can stand between
> the article "la" and its head noun, unless it is negating an adjective
> that precedes the noun, or is used as a prefix:
<foobar/> :)


-- Mark