LINGUIST List 7.633

Tue Apr 30 1996

Disc: Ungrammaticality

Editor for this issue: Annemarie Valdez <avaldezemunix.emich.edu>

Directory

1. benji wald, more unG

Please don't think I'm trying to get the last word on the topic of
"ungrammaticality". It's just that now that I'm thinking about it,
what other people are saying on the List (and off) is just stimulating
me more, for example, Karl's and Alexis's published along with mine in
response to Larry Hutchinson.

First, Karl, Alexis and I are in complete agreement that
"grammaticality" is a theoretical matter. To say that a sentence, or
whatever, is "grammatical" is to say that it is accounted for by a
grammar -- and a grammar is a theoretical construct, the creation of
one or more linguists. They said this better than "me". But they did
not go on to say that "grammars" are supposed to exist in reality, in
the "mind", or, more precisely, some phenomena exist outside of
linguists which we attempt to grapple with by the theoretical concept
of "grammar". This, then, is one source of confusion. It's the one
that Socrates in Plato's dialogues was warning his victims about ad
nauseam: don't confuse the shadow on the wall (the one you can
perceive) with the "real" thing (Plato's archetype) -- the
not-directly-accessible thing that makes the shadow. In optics this
might be interpreted as not to confuse the impression "light" makes in
our mind when transmitted through our eyes with the phenomenon of
"light" (particle or wave etc?) which "exists" (like the tree falling
in the forest) before it comes into contact with our eyes.
Ironically, in the case of grammar, the "real" thing is in our minds,
whatever "minds" are -- Chomsky would no doubt simply say "grammar" is
part of our "nature". Without this humbling constraint we can
construct any kinds of grammars we want, and admit/refuse any
sentences we want as un/grammatical because they're not in the
grammars WE constructed (as prescriptive grammars do). Of course,
very few will appreciate OUR grammars if those grammars don't
correspond to their experience as well. So, that's where the issue of
ungrammaticality just won't go away.

Chomsky put it nicely somewhere or other in observing that with the
(relatively) small amount of input that children receive, compared to
what they do with it, it must be "nature" that they come out with the
SAME grammar (we're talking about the same (dialect of the same) language
here, and "same" up to a certain point, that point being a problem when
you look for it). In other words, it seems to be innate that we
agree as much as we do. And if we want to be literal about the
implications of Chomsky's argument, our disagreements are unimportant,
he used to use the word "uninteresting" -- but he can hardly say that
anymore without sabotaging his own invention. In any case, we have a
paradox in that in some contexts the term "grammar" refers to something
"real" and presumably natural and innate in some ways, and in other
contexts, theoretical constructs about language data and the nature of
language that are -- uhm, theoretical constructs of language data and...
In practice it is not always easy to keep these two contexts of discussion
apart. A rule of thumb that often works, but not always, is that,
paradoxically, (in English) the "real" thing is often called "*a* grammar"
but the theoretical construct is called "*the* grammar (of whatever)".
And I can't blame you if you retort, they're both theoretical anyway.
But try to appreciate the difference. Somebody, try to say this better.

Moving from philosophical/scientific considerations to linguistics, a
comment I received from a reader of my last message is interesting,
and I think constructive. It's about how to characterise the
ambiguity in: I heard the girl playing my song.
The reader said that he had construed the ambiguity as, to paraphrase
what he said:
 (a) I heard my song being played by the girl
 (scope of "heard" "the girl/the song/playing"
 but I changed the topic from "girl" to "song")
 (b) I heard the girl who play+INFL my song
 (scope of "heard" "the girl" NOT "the song/playing")

In particular, the (b) interpretation he characterised as
containing a reduced relative clause, while the (a) interpretation does
not. This is a different way of saying what I said, but, as I told him,
I specifically avoided this characterisation, because I did not want to
prejudice the discussion in the direction of assuming that the
"ambiguity" is "strictly" syntactic. To characterise the (b)
interpretation as a "reduced relative clause" (rather than
paraphrasable as such) is to run the risk of imprinting in the
mind of an innocent student that such interpretations are purely
syntactic, and is more in line with missionary activity than with
encouraging students to understand the more fundamental issues that
grammar attempts to deal with.

I also mentioned in my reply that I was aware of the ambiguity because
I once read an article on some US varieties of Spanish in which the
author observed that the Spanish equivalent in those varieties have
adopted the same ambiguity as English, whereas in monolingual
varieties of Spanish the -ing phrase in (b) has to be expressed by a
relative clause. que tocaba mi cancion, NOT tocando mi cancion. My
observation about Larry's example remains that tense interpretation is
free in (b), as it is in a relative clause, "who played/is
playing/will play/plays", but not in (a).

I also mentioned that more was involved than just tense in why
students fail to recontextualise the example into a (b) reading,
since, in fact, ONE (b) reading coordinates tense between "heard"
and "playing". As long as I'm on the topic, my deduction is that
the (b) reading is much more context-dependent than (a). If you
mean "I heard the girl who is playing my song" then that's the more
"effective" (NB not "*grammatical") way of saying it. If you're saying
this to somebody in the context of both of you being somewhere where
this girl is playing your song when you say it, it's still unlikely
outside of a larger discourse context, because something has been
elipsised.

WHAT did you hear the girl (now) playing your song do/say/...(ing)?

It's pragmatic that "I heard the girl" is not much easier to construe
out of further context than "I heard". NB "You heard what?"/"what
did you hear?", "you heard the girl what?/??what did you hear the girl?"
(Aha!) So, that adds to the problem of recognising the (b)
interpretation in the blinding light of the (a) interpretation.
By the way, Larry has not yet confirmed to me that (a) was indeed the
interpretation his classes fix on, so maybe this is all a fantasy.
(I don't think so.)

To anticipate some comments I want to make about Karl's message,
it occurred to me that an experiment for the intro syntax students
might be (not for the first class): Give them a grammar which
will generate the structures for (a) and (b) above, as well as for
many other things, and a list of sentences which can be generated
from the grammar, some ambiguous in terms of the grammar, like
our example, and some not. Then tell them that some are ambiguous
and some not, and ask them as an assignment to generate all the
sentences explicitly, including all the ambiguities they can find,
IN TERMS OF THE GIVEN GRAMMAR. (Otherwise some may generate
an infinite number for any sentence. I heard the girl=Alice,
Betty,Carol, Diane..., which misses the point of what grammars/
languages do/are). For extra-credit, report the results to the
LING.LIST. with regard to the example. (LING.LIST pledges not to
help them do the assignment.)

Finally, I'd like to invite Karl Teeters to elaborate on what he
meant by:

"...let me just
point out that the problem here is that in fact most actual utterances
are ungrammatical (especially in Syntax I classes). That is why our
Master Leonard Bloomfield taught us that the fundamental assumption of
linguistics is that "some utterances are the same".

What I'm hoping he doesn't mean is that "most actual utterances are
ungrammatical". This reminds me of one of Chomsky's early
overstatements. In his zeal to impress upon psycholinguists the
innate basis for language learning he summoned among his arguments
the notion that most utterances are grammatically degraded. --it
following, then, that children as language learners were getting
very little guidance in examples of what's grammatical in the
languages they were acquiring. Ergo, they must have an innate
language learning/grammar-forming capacity. The arguments is:
A. Garbage in
B. good stuff out!
=. There must be a purification system built in.

I think the argument I alluded to above, the one about coming out with
the SAME grammar, would have been sufficient. However, I do remember
hearing or reading somewhere that Chomsky does believe that if a group
of human infants were isolated from human society, and somehow managed
to survive, they would develop some form of communication (I don't
suppose it would have to be verbal) that would reflect the grammars of
natural languages. Actually, that might be easier than learning
natural languages, "degraded" in performance, according to Chomsky's
perception. (I doubt he takes much time to observe how people actually
speak; his position and schedule would not encourage him to do so. If
I'm right, we can appreciate his brilliance in taking advantage of the
stammering, stuttering and groping for coherence among his opponents
in live discussion, as I'm about to explain.)

Be that as it may, he used as evidence of such degradation
recordings from such learned conferences themselves, where scholars
were trying to express complex ideas, sometimes for the first time,
sometimes trying to handle new concepts, and inevitably trying to
defend positions which had become their emblems against criticisms
they had not sufficiently anticipated -- or at least the impression
that those criticisms were making on other parties to the discussion.
(This is where Karl's aside about "esp in Syntax I classes" raised my
antennae.)

Labov, for one, thought that this was not sufficient demonstration,
and recognised that the propostion could be tested empirically
(where "empirical" means something; for Chomsky introspection was
a --> the form of empirical method, whence "intuitions", which
seemed at one time to be the "competence" side of "introspection",
were considered empirical data, and quickly were absorbed into
"competence" rather than "performance". This is crystal clear in C's
first 15 years of writings).

So Labov did, and eventually published it as "the Grammaticality of
Everyday Speech". His conclusion was that for the most part
everyday speech is grammatical, even without a few simple self-correction
rules signaled by markers, whose intent no doubt even small children
are able to acquire (and which may be an innate but not strictly
linguistic ability -- the ability to recognise the signal of a mistake,
i.e., of some behavior to disregard for hypothesis-building purposes;
I think, for example, we can easily recognise self-corrections in
another language, even if we haven't got the slightest idea what the
speaker is saying.)

Perhaps astoundingly, virtually all grammatical mistakes are self-
corrected. (Anybody can observe this informally -- and note that
people saying things which are grammatical but not what they meant
to say are much more likely to be uncorrected by them. "Really? I
said 'hot' -- I meant "cold"/I said 'they do'? I meant "they don't")

The frequency of self-correction (as needed to make an utterance
grammatical) was a function of how familiar speakers were with what
they were talking about, e.g., personal experiences had much fewer
self-corrections, of any kind, than explaining in words how
to tie shoelaces. It is interesting to note that ungrammaticality as
such increases in the public performances of people whose
business involves talking (in public), such as scholars, commentators,
spokesmen for officials etc. (esp, as theatre demands, when they are
led into covering ground for which they do not have a well-rehearsed
response). No doubt part of this is caused by stress and the problem
of saying things in a way which will preserve the public image of the
one required to speak. (In view of what we observe in everyday
situations we may even suspect that public performers sometimes
decline to correct themselves because that would call attention,
that they hope to avoid, to their slips in fluency. The same thing
happens when musicians give solo public concerts. If they make a
mistake they don't go back and do it over right, making things even
worse. Instead they just go on as if nothing happened -- and you will
find that much of the audience will not even notice or remember the
mistake, quite different from what would have happened if they stopped,
apologised and did it over.) With regard to input to child language,
I think that follow-up studies on Labov's claims have been done on
Motherese etc. but I'm not sure. Of course, many child language
studies have dwelt on Mother correcting (usually by repeating in
"correct" form) the utterances of children, and that has been used
to argue against Chomsky's claim -- but that is different because,
as C would no doubt say (justifiably or not), the corrections
only amount to a vanishingly small part of the grammars which the
children correctly learn. I also wonder if Pinker has commented
on this topic one way or another -- a way to gauge the current status of
the degraded speech myth close to its source. All of the above
discussion of the grammaticality of everyday speech is one of many
reasons why I agree with Larry Hutchinson's implied comment that
linguists would be well served toward their own ends if they realised
that they are in the "people business". You can't watch TV and
recognise on that basis how grammatical everyday speech is -- except
maybe for the scandal talk-shows.

So without meaning to put Karl on the spot, I wonder what he means,
and how it relates to what Bloomfield meant (when I read it in B, I
really thought mostly about how you can't pronounce the SAME word
EXACTLY the same way twice, esp obvious to linguists when working on
the phonology of languages in which variation crosses phonemic
boundaries in the linguist's own language -- Heraclitus's truism: you
can't step in the same river twice?) Is it that "I wants glass a of --
want a glass of water" is the same as "I want a glass of water"? (not
that the first contains a possible speech error to begin with). I
don't remember any context in which Bloomfield could be construed as
saying this, but I expect that Karl has an interesting interpretation.
 -- Benji