query: left-right asymmetry in order variation

[Jon Aske]

Matthew Dryer thinks that I am missing the point of Bingfu's message in my
reply.  I don't think I am, but just in case he is missing mine, I will try
to express it more clearly so that we can try to ascertain whether each one
of us is missing something, or whether we just disagree, and how much and
about what.  I must say, however, that I think our positions are much closer
than what it might seem at first sight.

It is true that the first task of a typologist is to uncover "descriptive
generalizations" about linguistic patterns.  But, two caveats here.  First
the generalizations cannot be just about grammatical/formal patterns.
Constructions, just like simple morphemes or lexical items, are pairings of
meaning and form, and "descriptive generalizations" should contain not just
grammatical categories and arrangements of grammatical categories, but
semantic and pragmatic information as well, such as semantic roles and the
pragmatic functions of the parts, as well as of the whole construction.  The
second caveat is that a second, but not secondary, aspect to typology, at
least as I see it, is achieving an explanation, or understanding, partial
and incomplete as it may be, of why such patterns exist.

With that in mind, let us look at the adjective position example.  It is
clear that the 'basic', 'syntactic' position for most adjectives in Spanish
is postnominal.  This is probably the closest that one could come to an
ordering convention, or "syntactic requirement", in Matthew's terms.  In
languages such as English, the 'basic' position is prenominal.  These facts
might seem to be to a large extent arbitrary (at least from the point of
view of a synchronic analysis).  Then, in addition, some languages allow at
least some adjectives to be on the other side from the 'basic' side at least
under some (semantic and pragmatic) circumstances.  (More on that later.)

The next question the typologist must ask is, of course, whether there exist
cross-linguistic asymmetries in adjective placement, such as (1) whether
more or less languages are Adj-N than N-Adj, or (1a) whether the degree of
preference for Adj-N or N-Adj is correlated with other properties of the
language, such as "basic word order", and so on; (2) whether Adj-N languages
are as likely to have a secondary position for adjectives as N-Adj languages
are; and so on.

If asymmetries are found, then the next question, I believe, is: why do
those asymmetries exist?  As a functionalist-typologist, I believe that the
search for explanations for the patterns we find is an intrinsic part of
doing typology.  The reasons may be of many different types: ease of
processing, diachronic/accidental, desire for symmetry, iconicity-related,
and so on.  The motivations may or may not be cognitively, or
psychologically, real, but often they are arguably so to a degree, or we
just may have no way of knowing.  More often than not, a pattern is the
result of a compromise among a number of different "competing" motivations,
each with different strengths and priorities, acting on a particular
language having certain characteristics (and thus favoring certain outcomes,
in Jean Aitchison's terms).

One may argue that at the stage of looking for symmetries and asymmetries,
we don't need to concern ourselves with the reasons (explanations,
motivations) for the asymmetries.  I would disagree with that on principle.
It is true that for some asymmetries we may not know enough about what is
going on to make sense of them and classifying them and quantifying them is
about all we can do.  But for many others, we know enough so that we can
begin to let our knowledge of the semantics and pragmatics of constructions
to guide our questioning, and to the extent that this is possible, I believe
we must pursue it.  And I don't think we should ever simply ignore the
meaning or function of particular constructions, and their parts, on
principle, such as the meaning/function of the Adj-N construction in a
language like Spanish in which N-Adj order is somehow 'basic', or the
meaning/function of the N-Adj construction in a language like English.

So, let us look at Spanish adjective position variation, and secondary
adjective position more generally.  It seems to me that departures from the
basic adjective position are typically quite clearly and transparently
motivated.  Thus pre-nominal positioning of adjectives in Spanish, which is
basically an N-Adj language, is associated primarily with a pragmatic
function which I will vaguely label "emphasis".  Thus adjectives that are
found in prenominal position in Spanish are often 'redundant' adjectives,
mentioned just for emphasis (e.g. "white snow", "ancient ruins").  Plain
restrictive (the most common) and contrastive adjectives on the other hand
are postnominal.

Now, fronting (early placement) of contrastive elements is a well-known
iconic principle which works at the clause-level as well, probably in all
languages, so that presumably in all languages in which the focus of the
assertion is typically postverbal, the so-called VO languages, a contrastive
focus can be placed in preverbal position under such circumstances (for
emphasis, as in exclamations, e.g. bloody soaked I-was).  Surely, the use of
such a motivated secondary position must still be conventionalized by the
language and different languages differ as to whether and to what degree
they allow this.  In Spanish, some adjectives have specialized meanings, and
even specialized forms, when they are prenominal, for instance.

On the other hand, I think that the main motivation for post-nominal
adjective position in Adj-N languages, is of a rather different nature,
having to do with weight and complexity, e.g. a man set in his ways.  Now
this motivation for 'postposing' is also found at the (asserted) clause
level.  (The converse of this motivator may also help account (to a much
lesser extent than the already mentioned motivation) for the existence, or
conventionalization of (short) pre-nominal adjectives in N-Adj languages
such as Spanish.)

Thus these different motivations for departures from the norm may help us
understand the asymmetries we find.  Maybe they won't, but we can't ignore
the fact that they might.  In fact, I believe our working assumption should
be that they may have something to do and it is our obligation to prove or
disprove whether they do.

Let me be more specific about the Spanish adjectives.  As I said, in Spanish
ordinary adjectives are typically postnominal, unless they are 'emphatic'.
A few common adjectives are placed quite commonly either before or after,
sometimes without any apparent change in meaning (ordinal adjectives), and
sometimes with a big change in meaning (antiguo = ancient; former), at least
in some contexts.  Some of these adjectives which can commonly go in either
place have a reduced form in prenominal position (some ordinal adjectives,
e.g. primer(o/a) "first"; most possessive "adjectives", such as "su(yo)"
his/hers/your/their"; , buen(o) "good"; gran(de) "big; great").  The
examples that Bingfu gives all contain some of these adjectives.  Some of
Bingfu's examples are bad because the unreduced form is used in prenominal
position (cf. prenominal suyo in * los amables suyos amigos; this example is
also bad because of the order of adjectives is disallowed, short, prenominal
possessive adjectives always coming first, presumably because of scope
reasons, and not being able to coexist with an article).

I am being specific about these examples because I think there is a real
danger that typologists looking at grammars of languages they do not have a
good grasp of will draw unwarranted conclusions from a very limited number
of examples.

But let me return to the more general issue here.  I am pretty sure that the
differences between Matthew's position and mine are not that great.  It is
obvious that both of us acknowledge the existence of both (1) functional
factors governing alternations and explaining preferences and asymmetries in
language, at least to some extent; and of (2) what he calls "syntactic
requirements".  Of course, what he calls "syntactic requirements" I would
term non-transparent or, more commonly, semi-transparent, (more-or-less)
grammaticalized, or conventionalized, or over-extended, or over-regularized
functional motivations.  It seems to me perhaps Matthew tends to emphasize
the ways in which functional motivations cannot account for the facts,
whereas I tend to emphasize the ways in and the extent to which they can.
Thus I am perhaps more likely to look for motivators whereas Matthew is more
likely to give up when the motivations are not apparent and obvious.

Thus I believe, for instance, that clause-level constituent order cannot be
understood without recourse to the notion of the pragmatic, or information,
functions topic and focus.  I believe that it is the failure to understand
these functions and their connection with word order is what has led
typologists to give up trying to understand word order and to resort to
purely syntactic notions such as basic word order which, although
constrained, are largely arbitrary.

Thus, I disagree with Matthew's view that, with respect to cross-linguistic
patterns of constituent order at the level of the asserted proposition, "the
preference in many languages is essentially a syntactic requirement".  I
believe that even those patterns which are somewhat syntacticized, the
motivations for the relative orders is hiding right under the surface and
that they are to a large extent transparent to speakers.

Matthew also mentions that:
| It is quite obvious looking at textual data for many VO languages
| that contrary to what Jon Aske suggests, many if not the
| majority of VO languages are as "adamant as English about
| having the object in immediately postverbal position".

I do not deny that, statistically speaking, nominal objects in VO languages
tend to be immediately postverbal.  Actually, I believe that there is a good
reason for that, namely that an overt direct object is typically the focus
of the assertion.  But what I meant in my previous posting, agreeing with
Bingfu's observation, is that English for instance doesn't allow focus
adverbials to be immediately postverbal when there is a non-focus object
present.  To give you a (somewhat artificial) example with a clear, obvious
focus:

A: WHEN did you buy that book?
B: I bought that book YESTERDAY.
B': YESTERDAY I bought that book (emphatic; rare perhaps)
B": * I bought YESTERDAY that book

Less rigid "VO languages", such as Spanish, and probably a majority of VO
languages, would allow B" without a problem.  I'm sorry if that wasn't
clear.

What I think has happened in English is that a common statistical tendency
in the language, namely for an overt object to be in postverbal focus
position, has become "syntacticized" to some extent, allowing few deviations
from that pattern (unless they are sanctioned by another syntacticized
construction, such as dative shift).  Languages such as Spanish do not need,
and thus do not have, a dative shift construction (nor do they have
English-like passives, for that matter, for a similar reason).

Anyway, I'll leave it at this.  I already talked more than my share.  I
welcome all and any constructive criticism.

Best,

Jon

-----------------------------------------------------
Jon Aske
Jon.Aske at salem.mass.edu - aske at earthlink.net
Department of Foreign Languages
Salem State College
Salem, Massachusetts 01970
http://home.earthlink.net/~aske/
-----------------------------------------------------
If you can't convince them, confuse them.  --Harry S Truman.