Theiling Online    Sitemap    Conlang Mailing List HQ   

Phaleran and C'ali updates: weak-crossover

From:Thomas R. Wier <trwier@...>
Date:Saturday, November 13, 2004, 9:02
Hi all.

I mentioned earlier this week that I'd try to get out some information
about weak-crossover in Phaleran and C'ali.  Since this is a rather
obscure question about variable-binding and anaphora, I first need
to say something about how it works in English.

Weak crossover refers to a kind of ungrammatical long-distance
dependency between some kind of variable (like wh-words, pronouns
or quantifiers) that is structurally high in the tree-structure
and the NP it is coreferential with:

Strong crossover:  *"He-i sees John-i" (where "he" and "John" are
                   coreferential)
Weak crossover:    *"Who-i does his-i mother like?" (where "his" and
                   "who" are coferential)

It is in fact more complicated than a simple structural analysis
like the above suggests, even in English.  Linear-order and syntactic
hierachies of grammatical relations are also relevant in deciding
such binding relations, and the appearance of tree-structural
hierarchies is often epiphenomenal from these.

In general, in most languages, the hierarchy of grammatical relations
and linear-order will align in determining coreference conditions:
since subject outranks object, and most languages have subject
preceding object (SVO, SOV).  However, in nonconfigurational languages
like Phaleran, there is no VP to create structural hierarchy, and so
linear order alone may determine weak-crossover effects:

(1) a. Atheruluo[i] sallei[i] gethasyonti?
      'His[i] father saw who[i]?'
    b. *Sallei atheruluo gethasyonti?
      'Who[i] did his[i] father see?'

Note that this crucially *does* refer to an asymmetry between a
binder and bindee, i.e., a unidirectional relation.  It is just
not a syntactic relation. That this is so is made clear when
related elements are interleaved:

(2) a. pyaka    pongeiyullu    atherul
       head.ABS sword-3Sg-ERG  father-3Sg.ABS

       sallei(lu)  yenauþnuntia
       who(ERG)    pain-CAUS-3SgPfPxRe-S-Q

       'Whose[i] sword made his[i] father's head hurt?'
    b. *pyaka    pongeiyullu    sallei(lu)
       head.ABS sword-3Sg-ERG  who(ERG)

       atherul           yenauþnuntia
       father-3Sg.ABS    pain-CAUS-3SgPfPxRe-S-Q
       *'Whose[i] sword made his[i] father's head hurt?'

Although some part of 'B' formally precedes some part of 'A'
(i.e., 'sword' precedes 'father'), all of 'A' formally precedes
some part of 'B' (the variable 'who' of 'B' follows the last
part of 'A', 'father').  Thus, this construction is grammatical.
Just switch the last two elements and it becomes ungrammatical,
as in (2b).

C'ali, because it is configurational, cannot work in this fashion.
In fact, C'ali is quite the opposite in all crucial respects, since
it is the grammatical hierarchy, and not linear precedence, that
determines binding relations:

(3)a. t[ai-oxthe-r         saxmë-n  olmissi-qa siyu    ?
      3SgDat.I-love-3Sg.II man-OBL3 wife-AGT1  who.OBL
      "The man loves whose wife?"
      ('love' here being an inversion verb with dative
      subject and agentive object)
   b. olmissi-qa siyu    t[ai-oxthe-r         saxmë-n      ?
      wife-AGT1  who.OBL 3SgDat.I-love-3Sg.II man-OBL3
      "Whose wife does the man love?"

In both (3a) and (3b), the binder is the subject, and this
outranks the bound object, even though the object precedes the
subject in (3b) by a rather normal processing of focusing in
questions.  If we reverse this, however, we see that we get
ungrammatical sentences:

(4) a. *t[ai-oxthe-r         saxmë-n  siyu    olmissi-qa ?
        3SgDat.I-love-3Sg.II man-OBL3 who.OBL wife-AGT1
        'Whose man (=husband) loves the wife?'
    b. *olmissi-qa t[ai-oxthe-r         saxmë-n  siyu    ?
        wife-AGT1  3SgDat.I-love-3Sg.II man-OBL3 who.OBL
        'Whose man (=husband) loves the wife?'

This is a highly unusually circumstance:  there are vanishingly
few languages in which linear precedence has no role whatsoever
in binding restrictions.  The only way to fix this situation is
a kind of cleft construction:

(5) saxmë-n  siyu    siti* t[ai-oxthe-r          olmissi-qa ?
    man-OBL3 who.OBL that  3SgDat.I-love-3Sg.II  wife-AGT1
    'Whose man (=husband) is it that loves the wife?'

Strange, but true.  Historically, another alternative was extraposition
of the operator:

(6) t[ai-oxthe-r         saxmë-n  olmissi-qa -- siyu?
    3SgDat.I-love-3Sg.II man-OBL3 wife-AGT1     whose
    'The man loves the wife -- whose was it?'

* (This subordinator was actually borrowed from Phaleran, and can
only be used in informal contexts. Formal contexts still require
the extraposition method.)

 =========================================================================
Thomas Wier	       "I find it useful to meet my subjects personally,
Dept. of Linguistics    because our secret police don't get it right
University of Chicago   half the time." -- octogenarian Sheikh Zayed of
1010 E. 59th Street     Abu Dhabi, to a French reporter.
Chicago, IL 60637