LINGUIST List 3.371

Tue 28 Apr 1992

Disc: Rules

Editor for this issue: <>


  1. , grammar and theory
  2. Avery Andrews, Rules
  3. , Thrainsson/Andrews/Itkonen on Rules

Message 1: grammar and theory

Date: Sat, 25 Apr 92 18:53:53 BSgrammar and theory
From: <>
Subject: grammar and theory

I should like to comment on the discussion in 3,357 by Dyvik of Andrews' views
on grammar and theory. In the course of this discussion Dyvik claimed that

>I take a grammar to be a theory...of an infinite set of possible
>utterances, predicting not only their acceptable shapes, but also aspects of
>their possible interpretations...

If this suggests that grammarians should look askance at "high back nasal" or
at such a quasi-minimal pairing as _loge_ /_l'auge_ the future looks bright.
But when Dyvik states that

>A grammatical theory predicts constraints on possible differences between
>grammars, and defines analytical tools and descriptive devices with which to
>write...grammars [differentiated by a grammatical theory]

we recognise that too much is being asked of a single component of linguistics
: grammatical theory. There must be an order in the two tasks.

Andrews, as cited by Dyvik, has put this usefully:

>>there is no point in launching into the elaboration of grammatical theories
>>without doing one's best to sort out the roles of the known causal factors.

Depressingly, for Dyvik this appears to be an impossibility

>we know next to nothing about...causes [of grammatical systems.

This cannot be true, if I understand Dyvik right, for psychology and sociology
have not been without important insights on offer to the linguist as to the
acquisition, variation and loss of grammar.

Dyvik rightly sees that here must be a dual task, for

>accounting for "grammatical systems" is one thing, and finding mental causes
>is another

Dyvik correctly rejects the notion that

>...[mental] causes can be found through inspection of grammatical systems

This would be to do our job backwards. The duality of our task is too often
concealed by the terminolgy that we get away with.

>From these valuable discussions, which have arisen from our uncertainty about
how "real" grammatical rules can be, _linguistic theory_ is missing as the term
for the theory of language, concerned with behavioural and utterance data.
Grammatical theory needs to be limited to name our important concern with the
theory of grammar, the discipline concerned with the analytical and descriptive
devices which will select the models which will enable linguists to account
for the data of linguistic theory.

Isn't reality, anyway, "created" through language?

Bill Bennett.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Rules

Date: Sun, 26 Apr 92 11:03:33 PDRules
From: Avery Andrews <andrewsCsli.Stanford.EDU>
Subject: Rules

Re Dyvik on rules, Linguist 3.357

>>A grammatical theory, as a theory about the mere shape of the data in
>>somebody's file-card collection, is an utterly uninteresting object.
>This may be true, but I do not see its relevance. I cannot believe that I
>have come across as a die-hard, corpus-confined inductivist - for one
>thing, I do not think that institutional facts, which I have been going on
>about, carried much more weight with that stereotyped prugelknabe of modern
>linguistics than it does with my present opponent(s).

Well, I wasn't arguing with you there, but with somebody else who was claiming
that theories are about their evidence, rather than the causes of their
evidence, which is not something I noticed you saying.

>I take a grammar to be
>a theory, not of a finite set of file cards, but of an infinite set of
>possible utterances, predicting institutional properties of them

There's a difference between grammars, as descriptions, and grammatical
theories. A grammar, as a description of a social norm, can be and often
is put together on a theoretically eclectic and pragmatic basis, using
whatever means seem best suited to convey a knowledge of the social norm
being described (I even know someone who used Chomsky's Syntactic
Structures formulation of Affix-Hopping to teach a bright but not very
well educated B.E.V. speaker how the Standard English auxiliary
system worked--all attempts at an intuitive explanation failed, but
with the formal rules, the guy could compute the required forms, &
write his papers). Perhaps you want to think of a grammatical theory
as a toolbox for producing such grammars, but I think this is a
mistake--a theory is not a collection an useful methods (however
useful important, and even interesting they may be), but an attempt to
explain why certain things are true.

It is therefore senseless to stare at some facts about social norms,
and try to produce a theory about them, without appropriate reference to
whatever is already known about the relevant causal factors, such as
history, U.G., and whatever else may be relevant.

>If it is, then so is generative grammar of the variey Andrews advocates.
>For how could a grammatical theory that in *practice* confines itself to
>the traditional kind of data - informant judgments of well-formedness
>properties and meaning properties - possibly be anything more than this?

Well, for example, in my 1982 paper on Icelandic case, I suggest a rather
flashy formulation of the regular accusative case-marking rule (mark an
object accusative if there is a `direct' subject), but also point out
that the actual reason for the existence of the generalizations that this
rule captures might just be the history of the language. So one should not
be hasty to dismiss some theory of case-marking just because it didn't
provide an elegant formulation of this rule (though, of course, there
would have to be some formulation that got the facts right). I could
perfectly well imagine that data from acquisition, speech errors, history,
etc., in Icelandic itself or other languages, could be relevant to deciding
what was realy going on with case-marking, and be evidence as to whether
my formulation or something more clunky was the right one.

More recently, in a 1990 NLLT article on morphological blocking, I cite
some acquisition data on `overtensing errors' as evidence that what might
be taken as an infelicity in some LFG analysis of English verb inflection
(they require special machinery to produce `did John leave?' instead
of `did John left?' is actually not a problem.

I agree that in practice linguistics is mostly pretty autonomous, but if
people aren't clear about the principles, the practice will never get
any better.

>Does this mean that a grammatical theory should be a
>theory of only those aspects of grammatical systems that have a certain
>type of causes? Isn't that at least impractical, given that we know next to
>nothing about such causes

A grammatical theory needs to provide resources for describing
whatever kinds of grammatical facts there are to be described, but it
it should only seek to provide `economical' and (Chomsky-style)
`explanatory' accounts of things that are clearly telling us something
about the nature of mental structure, that is, lack plausible
alternative causes. Common sense can do quite a lot to tell us what
we should attempt to treat in this way, though there are obviously
going to be many cases where things can't be settled from grammatical
evidence alone (as Chomsky has always insisted is the case in
principle). With a clear idea of what the causes are causing, we might
be able to find out more about them, but this requires that we distinguish
what is plausibly is being caused by mental structure from what isn't
(such as, perhaps, complex morphophonemics).

>Where we disagree, is in the assumption (not shared by me) that such causes
>can be found through inspection of grammatical systems alone.

This is not what I have been saying, but only that, given certain (sometimes
neglected) cautions, the structure of grammatical systems provides some
evidence about what's in the head. `Some' is the crucial word here.
If the structure of grammatical systems provides no evidence about what
is in the head, as people such as Scott Soames and Jerry Katz have recently
argued, then there is for example no point in psycholinguists trying to
find more evidence to corroborate or challenge the results of linguistics.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: Thrainsson/Andrews/Itkonen on Rules

Date: Sun, 26 Apr 92 23:25:00 EDThrainsson/Andrews/Itkonen on Rules
From: <>
Subject: Thrainsson/Andrews/Itkonen on Rules

Although I reject the Itkonen arguments for linguistic rules
being any less empirical than the claims of such sciences as,
say, biology ("All crows are black" would not be given up
in the face of a "counterexample" such as a crow freshly
covered in white paint), I must say I am distressed by the
recent arguments of Thrainsson and Andrews.

The point that is being made is that a sequence like "man the"
would not constitute an NP even if it occurred in the speech
of an English speaker. I reject this line of reasoning for
the following reasons:

 (a) If someone uttered these words either by mistake or
 as a deliberate joke, it might well be that the
 intended structure was an NP.

 (b) If we still refuse to admit (a), and say that,
 whatever he intended, the speaker nevertheless
 did not succeed in producing an NP, then I fear
 that that WOULD make linguistics normative in
 the way that Itkonen claims, for then we would
 be saying that, regardless of what we observe,
 the rule is that an NP SHOULD consist of a determiner
 FOLLOWED BY other stuff (such as a noun).

 (c) The argument seems to me to stray dangerously
 close to the position that linguists study
 grammars which are in people's heads, not
 the utterances which these alleged grammars
 are responsible for. If so, then again, Itkonen
 can all too easily win, because he can correctly
 point out that on this approach there is apparently
 NO empirical basis for linguistic claims.

In conclusion, I think that linguists should be prepared to
make claims about observable facts, even if these claims are
not always correct and even if we are forced to make such
vague distinctions as that between "normal" and "jocular"
or "normal" and "erroneous (i.e., unintentional)" utterances.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue