LINGUIST List 7.919

Fri Jun 21 1996

Disc: Written Syntax & Lg Equality

Editor for this issue: Ljuba Veselinova <>


  1. benji wald, written syntax & lg equality

Message 1: written syntax & lg equality

Date: Fri, 14 Jun 1996 03:07:00 PDT
From: benji wald <IBENAWJMVS.OAC.UCLA.EDU>
Subject: written syntax & lg equality
Warning: this is one of my long postings. Print it out. Take
your time. Underline whatever is worth commenting on. I'm going
to unload a lot of ideas about syntax and written language, and then
comment on "*it's* raining" and the Yoruba serial verb construction
for things like the English preposition "to".

While I, like most linguists, am inclined to dismiss such ideas as
those expressed in Ryan's posting out of hand, I welcome it for the
moment, because it gives us the opportunity to get into an issue
in linguistics that is of great interest to me: the issue of the
effect of the written medium on syntax. Since it is inconceivable
that societies of the size, complexity, and most specifically of the
*nature*, that we are familiar with nowadays could have developed
without the aid of the written medium, it is understandable that many
people, though mostly non-linguists, might conclude that whatever is
embedded in the "written language" is the "cause" of that form of social
evolution. Of course, the durability of a written message and its
potential to be disseminated over a wide area without the wild changes
associated with mouth-to-mouth dissemination, most notable in the
phenomenon of "rumor", is one reason, perhaps the main one, that a
certain amount of uniformity and stability (such as it is) is possible
within an extremely large and diverse society. This is a feature
of the written medium, regardless of language. However, there are
other things to be considered -- of a more specific linguistic nature.
And it is expectable that people can be confused about whether specific
features of particular written languages are responsible for the
non-linguistic features of the societies associated with those linguistic
features. (Chickens and eggs, carts and horses will be discussed in due

For the most part, those who have tried to counter the notion that
a specific type of language is responsible, i.e., causally related in
some way, for a specific type of society have dwelt on lexicon.
Lexicon is obviously an important consideration, and its relation to
a particular type of society (but perhaps as "effect", not "cause")
is so transparent that most people recognise this connection very
quickly when it is pointed out to them. They have no problem with the
notion that any language can adopt, import, coin, what-have-you, any
amount or kind of lexicon as the *need* arises. Thus, lexicon can
be excluded from "inherent" inequality of languages. The only
grudging recounter to this counter could be something like, yeah,
but who thought of it (= some lexical item associated with some
valued form of culture) first? It still leads away from lexicon as
an inherent source of inequality among languages.

Arguments one way or the other on the basis of syntax are quite
different. It is not usual for linguists to counter the idea that
there is a relation between a particular syntax and a particular
kind of society with the same argument, i.e., the argument that
languages can borrow syntax as the *need* arises. It is not clear
to linguists what it would mean for the "need" to borrow syntax to
"arise". Of course, syntax does get exchanged among languages in
contact situations, but the idea that it does out of any *inherent*
linguistic "need" is quite controversial, and there is very little
agreement about what counts as evidence or "proof" in this domain.
Instead, syntactic borrowing usually leads to issues concerning the
nature of bilingualism.

The only part of this domain where there is a widespread agreement,
though still controversy when it comes to details, is the difference
between, on one hand, an elementary pidgin (which is a stressful and
unbearably long-winded and repetitive kind of language both
syntactically and lexically, for telling stories, or any other task we
demand of a "normal" language), and, on the other hand, a well developed
stable pidgin or, of course, a creole (where a "creole" is a first
language, presumably like any other in all essential respects, whether
we understand this term as restricted to a nativised pidgin, or various
other languages which have come to be called creoles for social

For such reasons, with few exceptions, we find messages on this subject
only paying lip service to the notion that all languages are "equal"
or "potentially equal" when it comes to syntax (and morphology).
Is this a dogma? Or is there a reason to believe this?

Well, to begin with, we do not note that speakers of any language
have any more difficulty than speakers of any other language in
telling stories -- and I don't see why we can't include in "telling
stories", developing etc. scientific hypotheses, connecting the
incident involved in some law case to some legal precedents, and
whatever else might be included in our complex societies. Now anyone
who has ever been in a situation where they need to ask something but
there is no common language can readily appreciate this, even without
being so ambitious as to tell or hear a story. Such a person can
invidiously eavesdrop on a voluble conversation without understanding
anything but realise from the social interaction that a lot is being said,
and the interlocutors are doing it without the least discernible effort.
You can notice this about any language whatsoever as long as it is not
one you understand -- except for that elementary pidgin, which you would
notice causes the interlocutors communicative problems and limits their
social interaction in ways that are obvious to even the most obtuse

Of course, it has been noted for a long time, starting with Slobin's
work on first language acquisition, as far as I know, that, in fact,
some languages take longer to acquire than others (because of their
morphology it seems). Although this at first gave linguists pause
about the "equality" of languages, it is not clear, in the final
analysis, what this has to do with the issue of inherent equality of
languages, at least for "mature" speakers. Besides, if one were to
argue that time necessary for learning up to "maturity" is a measure of
the "efficiency" of a language, it seems to me that written languages
would count as the LEAST efficient. That, I think, is a rather perverse
application of the notion of "efficiency", especially since, at least as
used, written languages are in certain relevant ways MORE efficient
than spoken languages.

One way that written language is more efficient than spoken language
was already mentioned in one message to this list on this subject.
I have to apologise to whoever noted that, because I am not able to
find that message at the moment in order to give due credit. Anyway,
I previously made the same point against Ryan's ideas when I responded
to his same comments (and more) on the historical linguistics list.
It is that the same amount of "information" can be processed more
quickly in reading than in speech. (By "information" I am obviously
excluding all the social and emotional information which is
easily and automatically conveyed in speech but not in writing,
the phonological and prosodic information that must be converted into
lexicon and syntax in writing, where punctuation and available written
symbols fail -- but is often not even attempted in writing anyway.)

This processing time element could be construed as bestowing a certain
kind of efficiency to written language that spoken language does not
have, as long as we restrict our interest to "representational" content.
Part of this is due to the "permanence" of the medium. It is known that
there is greater redundancy in speech than in written forms of the same
message. Elinore Ochs and John Gumperz are among many sociolinguists who
have provided evidence for this observation. Apart from simply the way
information is processed visually, including linguistic information, there
are obvious appeals to the perspicacity of the written medium in
expressions like "the former/latter" and "respectively", which are not
particularly effective forms of "textual" deixis in speech. It requires
more repetition or circumlocution to represent the same kind of reference
in speech.

So, in my histling posting, I "conceded" one point to Ryan -- and
took it away from him by observing that it had nothing to do with
any argument he presented. It was that there may be a relationship
between literacy (not literate language when spoken) and complex
civilisation to the extent that in complex civilisations we have
the "value" that goes:

 "hurry! hurry! get the information! got it? good. pass it on!"

Time! (Time is also a big element in examinations: how FAST can the
examinee do the task? And in psycholinguistic experiments, where all
kinds of conclusions are drawn from reaction time). Time and speed is
highly valued in complex societies. Fortunes are gained and lost because
of split-seconds. Deadlines are hallmarks of the kinds of societies Ryan
has in mind. What is the greatest achievement of such societies; the
accurate clock?

I hate to ask this, but do you learn more in an hour lecture (assuming
that it's pure lecture) or in an hour of reading? (Or -- why is that
a difficult question?)

My own particular interest in the effect of the written medium on
syntax is more directly relevant to processing considerations. It
has to do with the kinds of condensations which favor complex syntax,
which combines multiple clauses and/or reduces them to phrases within
a single sentence in writing, in contrast to the more frequent
use of multiple "sentences" in speech (where "sentence" is a more
problematic unit in speech, as we can see, for example, in the way
children acquire punctuation in writing -- but I'm not going to get
too technical here, much less talk about the problmes of transcribing
speech). There are many reasons favoring condensation in writing, and
even the invention of such artifices as "respectively", not least of which
is inevitable limitations on length for written units, books, articles,
headlines, etc, and, of course, analogues in the electronic medium in
storage space (notice how every new model PC has more and more space
on the disk, making you wish you waited a couple of months before
buying yours -- whenever you bought it) -- everything costs money, or
runs out of availability. (But "talk is cheap" -- and it works in the

Beyond that, I am interested in whether there might be a universal
tendency toward the development of complex syntax in writing, as opposed
to speech, simply because the medium allows and encourages it, for reasons
of "economy" in all its senses, and because "time" and "fast" is so
important in complex societies that exploit literacy to the fullest.

One fast point here. Most languages have certain kinds of redundancy,
agreement rules for example. Perhaps such redundacy is historical
accident, but linguists find it difficult to resist speculating that
the redundancy serves a regulatory function in speech because of
attention lapses or whatever. Such rules, of course, are not dispensed
with in writing, though they could be if that was the only function they
serve (as in "telegraphic" or "headline" writing). Complex syntax,
squashing sentences together into clauses, and reducing whatever possible
to phrases, is a different order of economy in any case. This can be
appreciated by its spoken spin-off at a conference, debate, or whatever,
where you are limited to ask one question of the speaker and have to
cram as much into it as you really want to say before you get to that
one question mark that you are allowed. I assume such spoken strategies,
and the reasons for them, are parasitic on complex literate-based
societies. Am I wrong?

I won't pursue this here, because I am, as usual, abusing the economics
of the electronic system -- and who reads all the way through these
long posts I write anyway? But I put it out there for shorter "bytes" of

Having said that, I must say that what I have just said about
written language and syntax stands in contrast to the naive
propositions stated about specific syntactic features in Ryan's
message, which show a profound lack of understanding about what
syntax is and how it works. (I'm not trying to dis him. I'm
glad he gave me an opportunity to mention these topics on the
list. We've discussed them before, but I'm putting more detail here.)

One quick comment is that post hoc propter hoc arguments about the
relation between the syntax of certain languages, and the fact that
those languages happen to be used by complex societies for
reasons that have little to do with the syntax of those languages,
is a confusion of cause and effect which Ryan himself claims to be
be expressed better by the syntax of some languages than others (e.g.,
the "expletive" subject in "it's raining", as if that makes it
easier to ask "WHAT's raining?", leading to some advance in the
science of meteorology -- but I ask the List, is there some language
where syntax prevents asking the question: WHAT's raining? --
Or what's the difference if you have to ask WHY does (?it) rain?
I hope nobody out there's gonna tell me "what's raining?" is

Final comment, about serial verb constructions, where Ryan
suggests that using the verb "give" as a "preposition" (my terms,
he spoke of "separate sentences") somehow "limit(ed) the growth or
competitiveness" of ... (African) societies that employed them",
at the same time arguing, in some way I do not understand,that it
did not have the same effect on Chinese society, where a similar serial
verb construction is used. His assumptions about relevant African
societies are wrong, but I want to keep the discussion linguistic.
He did not quite get the facts straight, let alone the analysis. Let's
take Yoruba, the arch-serial verb language (and from time immemorial
spoken in a highly complex and, yes, competitive! society -- any
Yorubas out there want to comment?). Here's a fact:

 nwon ta a *fun* mi "they sold it *to* me"
 3p sell 3s *give=to* 1s

(I'm ignoring tone marks and vowel diacritics)

It does not mean "they sold (it) and gave it to me" (e.g., it doesn't
necessarily imply I actually took possession of it yet, any more than
the English translation does.) The fact that *fun* can also be used as
a verb "give" is not relevant to analysis of the construction.
Furthermore, Yoruba does not descend from a language in which it was
necessary to use two "sentences" to express the idea exemplified by this
serial verb construction. It most likely evolved from a language more
like Bantu, where a post-verbal inflection was used to indicate the
relation between the "indirect object" and the verb, cf. Swahili

 wa-li-ni-uz-*ia* "they sold (it) *to* me"

 (object "it" usually goes without saying in Bantu. Yoruba and
Bantu are both Benue-Kwa languages, much more closely related than
many other Niger-Congo languages, and there are a variety of languages
in their extensive area which suggest the various paths of historical
transition among these two relatively different syntactic types.)

The Yoruba serial verb construction arose as the language (general to
its area) evolved into a more analytical type. (Similarly, it seems
with Chinese from Sino-Tibetan). The full range of uses of Yoruba
*fun* is somewhat similar to the range of uses of the APP verb suffix
in Bantu, and goes far beyond "dative" uses, as indeed "to" does in
English. The idea that the serial verb construction exemplified above
for Yoruba either consists of or derives historically from a two-sentence
construction, is a naive idea that mountains of literature on serial verb
constructions refute. For good measure, we might as well note that
Yoruba is the most advanced among the serial verb languages in its area
in having a peculiar constraint that one and only one object can be
unmarked (for "case"). Which object that is is determined by the
lexical verb of that (ONE) sentence/clause (get it? one sentence!)

Thus, another fact is:

 o *fun* mi *l'*-owo s/he gave me money
 3s *give* me *"at"=ACC*-money

Notice that as a verb "fun" (give) takes the dative as the unmarked
object. That's a lexical fact, and one historically relevant to the
choice of "fun" as a preposition, or "case-marker", in the previous
serial verb construction. Some Benue-Kwa languages still have verbs
of giving which allow both an unmarked dative and accusative object
(to use those terms). The above example represents that there are no
ditransitive verbs in Yoruba anymore. l' < ni may have once been a verb
meaning "be at", but in context in Yoruba it marks the "accusative" or
"direct" object when the verb requires a different object. (Thus,
the single lexical verb "teach", among others exhibit the same

The grammatical logic is approximated in English pairs like:

 he shot me with a gun/he shot a gun at me
 she loaded the car with suitcases/suitcases into the car

Only approximate since there is a meaning difference in the English
pairs. The point is that only one object can be "unmarked" for
case. Yoruba is different from the English examples in that a
different verb would have to be used to change which object can
be unmarked. If this has anything to do with the difference between
Yoruba society and English-speaking society we're back to our old
friend the lexicon, not two sentences vs. one. Thus, another fact is
Yoruba can change focus differential between two objects by
changing the verb, so that
 3s put money "give"=to 1s "s/he gave the money to me"
expresses the same alternation as between the dative and prepositional
translations in English.

It will not be sufficient to refute such naive ideas about syntax
as Ryan expresses with a general principle alone. But to the extent
that the general principle holds, it will be possible to refute any
particular example of such naive ideas with specific analyses. This
is why I did not dismiss the idea out of hand, but changed it into
something more interesting. -- Benji
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue