Re: New Site: New Language: Tyl-Sjok

Hi!

<frustrated look> I wrote a posting yesterday.  Took me an hour.
It was lost in the jungle of bits.  I switched on posting backup
now.

It was supposed to be an answer to the following mail.  I'm going
to try to remember what I wrote.

daniel andreasson <daniel.andreasson@...> writes:
> Henrik Theiling wrote quite a while back:
>
> > with thinking about `what if a language' suggestions. They
> > are not influenced by any theory, so they think more freely
> > and philosophically.
>
> You have some nice friends! Mind if I borrow some of them? :)
No, no, they're mine!! :-))


> That is so neat! As minimalistic as can be.
Actually, I might over-minimalise a bit currently.  I don't dare to
add rules.  I even cut the grammar to one composition rule + particles
(and there are only very few types of particles, too).  But at least
the vowels aren't minimalstic.  I though that so many conlangs have
few phonemes that I wanted something with some more vowels.


> > That's why 'I love you' is not rendered with ablative but by a
> > second patient applied on the embedded structure `to love you'.
> > So both `you' and `I' follow `to love'.
>
> Aha! [ Okay, I'm out of exclamation marks. I think I've used my
> share for this mail by now. :) ]  Could you give some more examples
> of this embedding, with interlinears and all?
Sure!  That's going to be lengthy.  But you asked.

Let's start with the sentence above:

    jet  je  jo.
    [jet.VERB je.PATIENT].VERB jo.PATIENT
    *love you I
    `I love you.'

So `love' has two patients here.  The reason why they are added in
this order is control: `the love to you' is the predicate that
controls `I'.

The embedding above is the same that is used for relative clauses.
Starting with the sentence:

    father peter.
    `Peter has a father. / peter's father.'

The literal translation is very weak because Tyl-Sjok has no distinction
between verbs and nouns, but it would be something like:

    *(it) fathers peter. / peter is fathered.

Note the ordering again: it is basic on control: the father controls having
a son (at least sometimes).

In a relative clause, you simply use that whole phrase to refer to something
in the relative clause.

    [father peter] strong.
    `Peters father is strong.'

In contrast to English, the entity described in the matrix clause is
realised in the sub-ordinate clause and the whole sub-ordinate clause
is used instead in the matrix clause.

The referent in the sub-ordinate clause is usually unmarked, because
it is clear from context in most cases.  To give an awefully ambiguous
example, take the following:

   good father son.

This has various interpretations.  I'll give them with a version that
is unambiguous to show the use of particles.

  a) Second part is embedded:  good [father son].

      1) good father REF son.
         `The father's son is (inherently a) good (person).'

         REF marks the exported referent that is used in the matrix clause
         from the embedded clause.  Usually, REF also marks that the marked
         entity is in the sub-ordinate clause (the order of embedding is also
         unmarked most of the time).

      2) good START REF father son.
         `The son's father is good.'

         START marks a phrase start.

      3) good have father son REF.
         `It is good that the son has a father.'

      4) good father have son REF.
         `It is good that the father has a son.'

   b) First part is embedded: [good father] son.

      5) good father END son.
         `The son has a good father.'

Well, usually, many of these ambiguities don't occur: if `strong'
instead of `good' was used, the interpretation as a full sub-clause
would not be possible: *`It is strong that the father has a son?'.

If `son' was substituted by `Peter', it would have been clear that the
referent `Peter' isn't really the default.  Sentences like `Peter,
having a father, is good.' would be circumscribed.

(Currently I've introduced a pitch-based prosody is marked to
disambiguate embedding structure, but I'm not sure whether that
is good.  It may vanish/be replaced/whatever.)

All sub-ordination works by direct embedding.


Time
----
There are three verbs that define when, until, etc. something happens:

    happen-when
    happen-until/end-before
    happen-after/start-at

The phrase that happens is in agent position (it is in control of the
piece of time it occupies).  The time is in patient position.

E.g.

    I read book happen-after eat end.
    `I read a book after the meal.'

Place
-----

Works exactly like time using three verbs:

     take-place-at
     come/go-to/arrive-at
     come/go-from/leave-from

Coordination
------------
There is one particle, called SEQ, that links two phrases that have no
sub-ordinate relation.  It is usually translated as `and', but may
also be `but', `or', `although'.  It's use is straight-forward:

    sleep SEQ dream I.
    `I sleep and dream.'

    sleep SEQ dream resist I.
    `I sleep but still dream.'

    sleep resist SEQ dream I.
    `Although I sleep, I dream.'

    ...

Gapping
-------
Not done yet. :-)


New Words
---------
A particle NULL marks an unmentioned referent that is described
by the surrounding phrase.

    NULL walk             -> `the walker'
    NULL eat              -> `the eater'
    eat NULL              -> `that which is eaten: meal'
    use NULL              -> `means, manner'
    happen-when NULL      -> `moment'
    take-place-at NULL    -> `place'
    NULL (move) use feet  -> `pedestrian'
    move use NULL         -> `that is used for moving: e.g.: road'
    ...

Another particle marks a non-referenced unmentioned entity: SKIP:

    love NULL             -> `the one who love: lover'
    love SKIP NULL        -> `the one who is loved: the beloved'

Long Distance References
------------------------
There are a lot of particle constructions for back-reference.  Two verbs
are used.  I introduced the constructions quite recently:

    to make (resultative verb)
    to become (inchoative verb)

The can be composed with the above particles:

    make NULL     -> `the last entity that something was done to in a phrase
                     that had an agent'

    NULL make     -> `the last mentioned agent'

    become NULL   -> `the last entity that something was done to in a phrase
                     that had no agent'

    become SKIP NULL -> `the secondary patient in a last phrase that had no
                         agent'

    ...

    Using the REF-particle, whole sentences can be referred to:

    REF make      -> `the last action that was performed'

    make REF      -> `the whole last sentence that had an agent'

    REF become
    become REF    -> accordingly

Example:
    Peter drink black tea SEQ good NULL make.
    `Peter drinks hot tea and he is (a) good person.'

    Peter drink black tea SEQ good make NULL.
    `Peter drinks hot tea and it (the black tea) tastes good.'

    Peter drink black tea SEQ good become NULL.
    `Peter drinks hot tea and it (the tea) tastes good.'

    Peter drink black tea SEQ good REF make.
    `Peter drinks hot tea and that's a good thing.'

    Peter drink black tea SEQ good REF become.
    `Peter drinks hot tea and it is good that it is hot.'

For longer distance discourse reference, `he, she, it', the 3rd person pronoun
may be used.

I could generate more, but I think it's enough for one posting now. :-)

**Henrik