More on left-right asymmetries of word order variation

[bingfu]

Dear fellow netters,

	I got several responses since I posted my query about the
left-right
asymmetries of word order variation.  The following interim summary may be
helpful for further observations.
		
			QUESTIONING OF DATA

	Two netters (Jon Aske and Víctor Vázquez Martínez) question
the Spanish data in my posting as follows.

        a.      el primer buen capitulo
                the first   good chapter
                'the first good chapter'

        b.      * el buen primer capitulo
                 the good first chapter

        c.      el capitulo primero bueno
                the chapter first   good

        d.      el capitulo bueno primero
                the chapter good first

However, their judgements are not completely consistent. Aske regards all
four, including (b) perfect; while Martínez regards (b) as good but (d) is
strange.  Well, I don't know much about Spanish and my Spanish data is
taken from:
Terker, Andrew M. 1980. Linear Order and Intonation in the Spanish Noun
Phrase. University of Washington dissertation. pp 109-111.

	If this example is not good, how about the following data from
the same book.

	 a.	sus amables amigos
		his likeable  friends
		'his likeable friends'

	 b.	* los amables suyos amigos
   		  the likeable  his     friends

     	c.	los amigos amables suyos
		the friends likeable his

    	d.	los amigos suyos amables
		the friends his      likeable

When both suyos/sus 'his' and amables 'likeable' precede the head noun,
only one order is allowed (suyos must take short form sus when it precedes
other modifiers in prenominal position); when both follow, the two
alternatives are both allowed.
	Hopefully more clear examples might be found in other languages like
Russian,where word order is freer and as I know there is still a unmarked,
dominant order among multipleprenominal modifiers.  Any Russian native
speaker can help check it out?


	FURTHER SPECIFICATION OF THEISSUE

	Let me specify my question further. It should be noted that there
are many left-right asymmetrical phenomena and I am now concerned only
with the asymmetries of "word order variations/consistency".
	Richard Kayne's 1994 book "The Antisymmetry of Syntax" is not
direct relevant.  My understanding of his book is that Kayne argues for
the universal head-initiality at a deep, abstract level.  My concern is on
surface construction.  In fact, I regard the left-right asymmetry as
essentially a superficial phenomenon, since what left-right asymmetry
really means is a temporal before-after contrast, which is
a real-time performance issue.
	Following kayne, Yuji Takano, in his UC Irvine 1996 diss.
"movement and parametric
variation in syntax", also argues for the syntactic antisymmetry.  But he
differs form Kayne in that he argues for head-finality as universal order
at an abstract deep level.
	It seems, then, the abstract antisymmetry is hard tojudge.

	My concern is also different from what Hawkins, Cutler, Bybee et
al. previously discussed.  Their left-right asymmetries are about the
skewing
distribution of some particular units.  For example, while all head-final
lanuages use suffixes, some head-initial languages, which are otherwise
expected to use prefixes, use suffixes too.  Hence, there is a general
preference for suffixes over prefixes in human languages. The two kinds of
asymmetries though related, are not the same. Discussions on the
asymmetries in affixations are provided in Hawkins 1988a, b, Hawkins and
Cutler 1988, Bybee et al. 1990.

	Takaku Tsunoda mentions that he found preposition stranding in a
fair number of languages, but he does not know of any clear instance of
postposition stranding. In his co-authored paper on adpositions (
Tsunoda, Tasaku; Yoshiaki Itoh; and Sumie Ueda. 1955.
   "Adpositions in word order typology". Linguistics, Vol.33,
   No.4:741-61.), based on the statistics of 130 languages, he points out
that stranding is attested in 8-10 % of the prepositional languages, but only
2% of the postpositional languages.  Stranding is surely a word order
variation issue and therefore
directly relevant to our current concern. The stranding phenomenon
suggests  that postpositions are more bound than prepositions in general.
This may be a by-phenomon of the tendency that "postposed grammatical
materials is more likely to affix than preposed grammatical materials", as
observed by Greenberg, Hawkins et al.  In other words, a postpositional
bound morpheme is more bound than a prepositional morpheme.  I would like
to cite another example.   The Russian instrumental preposition "c" /s/,
though not constituting a syllable, is still treated as a word and
written, perhaps even read, separately, but English postpositional
genitive marker -s, is treated as a
suffix, even in such case like "the king of England's daughter".  My guess
is that the tendency is related to some particular property of human
cognitive perception.

	In addition to the asymmetries mentioned in my previous posting, I
would like raise another two: cross-linguistic adjectival orders and the
orders of direct and indirect objects.
The great word order variation in noun-initial NPs seems not only occurs
among determiner, numeral and adjecive as stated in Hawkins' NP internal
order universal; it exists among various adjectives as well.  It has been
well observed that almost all the languages that have 'free adjectival
order' are those where adjectives normally follow head nouns, such as
Somali, Kurdish, spoken Arabic and modern Hebrew (Hetzron 1978: On the
relative order of adjectives. In Hansjakob Seiler ed., Language
Universals: 165-184.), as well as Romance (Sproat and Shih 1991,The
cross-linguistc distribution of adjective ordering restrictions.
Interdisciplinary approaches to language: essays in honor of S.-Y. Kuroda,
ed. by Carol Georgopoulos  and Roberta Ishihara, 556-593. Kluwer Academic
Publishers.)  Hetzron noticed this and wondered: "I do not know if it is a
coincidence or not that these languages exhibiting free adjective order
have the adjectives after the nouns."
	The orders of direct and indirect objects (hence DO and IO
respectively) also demonstrate some left-right asymmetry. If both DO
(direct object) and IO (indirect object) follow V, IO frequently stays
nearer to V than DO does, as exemplified in English and Chinese dative
shift.  In Mandarin Chinse, the order [V IO DO] is in fact used more much
frequently than the order V DO IO and is hence viewed by some grammarians
as canonical, basic order.  By contrast, if both DO and IO precede V, the
corresponding mirror-image oder [DO IO V] is never taken as a canonical
order, as far as I am aware.

	Hope to receive more feedback!
							Bingfu Lu
							USC