Re: attributive predicates in rinya

Daniel wrote:

>I'm sorry it has taken a while replying to this mail. I had
>to do some serious thinking. You ask some excellent questions,
>Marcus.
Thanks.

>Marcus wrote:
>
> > > Let's say we have the phrase "The brown dog runs". "Be brown"
> > > is an inherent state and its argument is marked as PAT. "Run"
> > > OTOH is a controlled event and its argument is thus marked as
> > > AGT. Now, the syntax of Rinya doesn't allow the subject argument
> > > of a relative clause to be different from the one in the main
> > > clause.
>
> > Do you mean the subject arguments have to be the same in AGT/PAT
> > or that they have to refer to the same entity?
>
>After a lot of thought, I've decided that they have to refer to the
>same entity.
>
> > Does this hold for sentences where the subjects of the two clauses
> > are not co-referential, like "The dog that John saw ran."?
>
>I think Rinya will use passives like this:
>
>(1)  *areamie thelin  ly           Jynin    qhentie.
>       run:PST dog:AGT REL:PAT:DIFF John:AGT hit:PST
>       'The dog that John hit ran.'
>
>This above sentence is impossible. It just doesn't make any sense
>at all. Instead one will have to use the passive to rearrange the
>arguments like this:
>
>(2)  areamie thelin  ly           qhentau  Jonumbe.
>      run:PST dog:AGT REL:PAT:DIFF hit:PASS John:INSTR
>      'The dog that was hit by John ran.'
>
>That is, the relative clause "that John hit" must become "that
>was hit by John".
>
>OK. But how about a sentence with an indirect object being
>relativized?
>
>"The book that John gave to me was red." is ungrammatical
>in Rinya. This must become:
>
>(3)  riw      le           mestau    imenya  Jonumbe    le  timie.
>      book:PAT REL:PAT:SAME give:PASS 1SG:DAT John:INSTR 3SG red:PST
>      'The book that was given to me by John was red.'
Sounds good. But that makes me curious about the other object.

"The man who John gave a book was ugly."

Does passivization apply in these cases too? I ask since in some languages
this argument cannot be promoted to subject.

>So far everything seems to work out fine. It's when it comes to
>the genitive that problems arise.
>
>Consider the sentence "The book whose writer was eloquent was red".
>"The book's writer" is rendered in Rinya (The genitive is marked
>by the dative):
>
>eriwil     riwenya
>writer:PAT book:DAT
>'A writer of a book' Lit. 'A writer is to the book.'
>
>(4)  *riw      lenya   eriwil     edrinie      le  timea
>       book:PAT REL:DAT writer:PAT eloquent:PST 3SG red:PRES
>      'The book whose writer is eloquent is red.'
>
>The problem is that the relative pronoun _lenya_ refers to the
>book, but it is _eriwil_ 'the writer' which is marked as PAT.
>Somehow this seems wrong. I thought maybe this then would always
>be ungrammatical in Rinya and one had to paraphrase this somehow.
>Examples:
>
>(5)  "The book who had an eloquent writer was red."
>(6)  "The book which was red had an eloquent writer."
>
>The problem here is that there is no word "to have" in Rinya.
>
>"I have a book" = "A book is to me" = _riw imenya_ 'book 1SG:DAT'
>
>The dative is used for have-phrases as well. Example (5) would
>then look like:
>
>(7)  riw      le           eriwilenya edrinie      le  timie.
>      book:PAT REL:PAT:SAME writer:DAT eloquent:PST 3SG red:PST
>      'The book who had an eloquent writer was red.'
I'm confused. Are you saying that the simple sentence "The book had an
eloquent writer" would be:

riw      eriwilenya edrinie.
book:PAT writer:DAT eloquent:PST

It looks to me like "eloquent" is acting as the predicate of the sentence,
since it takes the tense marking. I would have guessed it meant something
like "The book is eloquent to the writer", or something like that.

>Compare this to (4). Either you get _lenya eriwil_ or you get
>_le eriwilenya_. This leads to confusion. I'm not sure if
>either of these sentences will be understood. Try looking at
>the morpheme-by-morpheme translation gloss only and I think
>you'll see what I mean.
>
>What should I do?
Well, some languages don't allow relativization of possessors, so you could
include Rinya in their ranks. That isn't a very helpful suggestion, is it? :)

Seriously, how about possessor raising of some sort? When you raise the
possessor to be an argument of the verb, the former-possessor and possessed
have distinct roles, so could be treated accordingly. I'm going to take a
guess at the proper Rinya for my examples, but if I'm wrong, just follow
the glosses.

The sentence:
erivil     riwenya  edrinie
writer:PAT book:DAT eloquent:PST
'The book's writer was eloquent.'

Becomes:
riw      erivil     edrinie
book:PAT writer:PAT eloquent:PST
'The book's writer was eloquent' = 'The book has an eloquent writer'

The possessor has become the subject of the sentence. If there is any way
to be sure of the subject of the sentence (other than semantics), such
tests will point to "book" in cases like this. Some languages that do this
(such as Chickasaw and Mohawk) insert some kind of "benefactive" marking
that agrees with the raised possessor to show what happened.

Since the possessor is now the subject of the sentence, you can treat it
like one.

riw      le           erivil     edrinie      le  timie
book:PAT REL:PAT:SAME writer:PAT eloquent:PST 3SG red:PST
'The book which has an eloquent writer is red.'

> > > (5)  The dog:PAT [ which:PAT:SAME browns ] falls quicks.
> > >            P     [   P              V    ]   V    V:ADV
> > >      'The brown dog falls quickly.'
>
> > What kind of morphology marks an Adverb? It seems to me that the proper
> > treatment of them could also ease the awkwardness a bit.
>
>Yes, I'm having troubles with this, mostly because I haven't a clue
>as to how languages without adjectives solve this. Do you have any
>suggestions?
My suggestion would be to add a morpheme that changes a verb into an
adverbial. This could apply to simple words or to entire phrases. Take, for
example, a suffix like -ig (borrowed from a Pima morpheme that appears in
adverbs based on predicates).

I walked to school tired-ig.
'I walked to school tiredly.'

I walked to school sing-ig.
'I walked to school singing.'

I walked to schoool sing-ig a song.
'I walked to school singing a song.'

Or to appply it to your example:

the dog:PAT [which:PAT:SAME browns] falls quick-ig.

> > > The solution is to insert a pronoun between the REL-clause and
> > > the verb, but *only* if the main argument is a PAT. If the main
> > > argument is an AGT it isn't necessary as we saw in (6):
> > >
> > > (7)  The dog:PAT [ which:PAT:SAME browns ] it falls quicks.
> > >      'The brown dog falls quickly.'
>
> > Have you considered using internally headed relative clauses for this?
> > You could mark the noun according to the role of the relative clause,
> > then mark the entire relative clause for the role of the matrix clause.
> >
> >   The dog:PAT browns:PAT falls quicks
> >   The dog:PAT browns:AGT runs quicks
>
>I'm sorry, my head has run out of brain. What is the matrix clause?
>I'm not sure you could explain this better than you already have
>but could you try? :)
Sorry. Matrix clause = main clause.

Suppose that the case marking is a postpositional clitic, like in Japanese
or Chickasaw. It therefore attaches to the last element of a noun phrase,
not onto the head noun itself. You would get something like

[The dog]:PAT falls.

A relative clause is part of the noun phrase, so the case marker would
occur following the relative clause:

[The dog [that I saw]]:PAT falls.

You can place the head of the relative clause inside the clause itself,
rather than having a relative pronoun. To return to your example:

[The dog:PAT browns]:PAT falls.

"The dog" is the subject of "browns" and so is marked as PAT. "The dog
browns" is the subject of "falls" so the whole phrase is marked as PAT.

One serious draw back to this approach is that you would have to abandon or
seriously modify your relativizer system, because there is no longer any
room for "lyn", "le", etc. I don't recommend this, because I really like
the system you invented. I just mention this as a possibility that could be
considered.

> > > 8.  main=PAT rel=AGT -> _lyn_
> > > 9.  main=PAT rel=PAT -> _le_     [ unmarked ]
> > > 10. main=AGT rel=AGT -> _lin_    [ most marked ]
> > > 11. main=AGT rel=PAT -> _ly_
One further comment on this system that hadn't occurred to me before. This
is conceptually very similar to switch-reference marking, like is found in
Chickasaw. Switch-reference is a property of the complementizer system (at
least in Muskogean languages). In GB syntax, relative pronouns are in that
system, so having switch-reference on the relativizer like you do seems a
very natural possibility.

Nice Work.

> > So, when the rel= AGT, the morpheme ends in -n, other
>
>...wise it ends in a vowel? Yup. This is because two morphemes
>were joined together: REL:PAT + DIFF:AGT -> le + in + UMLAUT ->
>-> lön -> lyn.
>
> > I like the system. Only additional suggestion: could we see the
> > Rinya next time? :)
>
>Happy now? :)
Grinning like a possum. :)




Marcus Smith

"Sit down before fact as a little child,
be prepared to give up every preconceived notion,
follow humbly wherever and to whatsoever abysses Nature leads,
or you shall learn nothing."
                -- Thomas Huxley

Replies