LINGUIST List 3.336

Sat 11 Apr 1992

Disc: Rules

Editor for this issue: <>

Directory

1. Avery Andrews, reality of rules
2. , Rules
3. Avery Andrews, rules
4. , theory & rules

Re: "Larry G. Hutchinson" <hutchincs.umn.edu> (Linguist 3.323)

>Slogan: FACTS do not become EVIDENCE (for or against a theory) unless the
> theory is about those facts.

I think this is a very bad slogan, and much prefer:

 facts (may) provide evidence about the things that
 cause them.

The trouble with the first slogan is that it doesn't provide any sort
of motivation for people to seek large-scale integration of different
theories, of the sort that you need to get real sciences.

I agree that generative grammarians tend to be somewhat over-eager
to ignore embarassing evidence, but I don't notice that this problem
is confined to `extralinguistic' evidence.

Re: Dyvikhf.uib.no (Linguist 3.323)

>Imagine a future discovery to the effect that the production of NPs is
>*not* correlated with a single 'facet of mental structure', but is
>distributed among various 'facets' according to principles bearing no
>relation to grammatical properties. Or, alternatively, imagine that these
>correlations vary significantly from one speaker to another. Would any of
>this lead us to revise our grammar for English? Would it lead us to
>conclude that the category NP does not really exist after all?

For me, and I hope, other `Chomskyites', the answer is definitely yes,
provided, that these different facets are concerned with `NP's (now
shown to be some sort of peculiar epiphenomenon, of unclear status)
appearing in different positions. In a current modular theory, the
`NP-rule' is already split up into various components that do
other things as well (X-bar theory, government theory, LP-theory or
whatever your favorite scheme is), but the same bunch of components
is operative and interacting in the same way in all the NP-positions,
& so constitutes a sort of `complex facet'.

We are then, making the prediction that this particular discovery
won't be made, and if somebody thinks they've made it, they ought
to look again. Notice the stiff challenge that our hypothetical
discoverer has to meet: he has to come up with a reason for all these
different, causally disconnected, facets of mental structure (one for
each NP position) to handle the same subsequences of words in the same way in
each individual language, even though the way the subsequences are
handled in different languages is different.

>Theories about the mental correlates of grammar *presuppose* grammatical
>theories, but they *are* not grammatical theories. From the point of view
>of such theories, grammars do not explain - they define what is to be
>explained.

I don't find this a viable viewpoint for grammatical theory, because
the actual shape of grammatical systems is determined by all sorts of
different factors (brute history for example being a dominant and
*extremely* troublesome one in phonological systems), and there is no
point in launching into the elaboration of grammatical theories without
doing one's best to sort out the roles of the know causal factors.

A grammatical theory, as a theory about the mere shape of the data in
somebody's file-card collection, is an utterly uninteresting object.

Avery.Andrewsanu.edu.au
 (currently andrewscsli.stanford.edu)

Martti Arnold Nyman <MANYMANcc.helsinki.fi> (Vol-3-330. Thu
09 Apr 1992) makes a distinction between Rules of Language
(Lrules) and Rules of Grammar (Grules). Lrules are said to be
"conventional social norms of which competent speakers have
intuitive knowledge". Grules, by contrast, are "theoretical
descriptive entities expressed by means of some system of
notational conventions". About Grules "the linguist has no
intuitive knowledge before s/he has carried out the
descriptive task".

It is still a matter of doubt in my mind what value there is
to the distinction, but the concrete question I want to focus
on here is what Nyman understands by "intuitive knowledge", as
this is alleged to somehow differentiate Lrules from Grules.
For instance, how do we tell it apart from nonintuitive
knowledge? Does the latter exist at all? Does the linguist
have nonintuitive knowledge of the Grules before "carrying out
the descriptive task", and intuitive knowledge doing so? How
does the distinction relate to explicit/implicit knowledge
distinction?

The question arises more generally, ie with respect to the
distinction between Grules and Lrules itself: how does one
tell the two types of rules apart? For instance, if one finds
that a speaker observes the wh-island constraint, or forms
yes-no questions by having recourse to hierarchical structure
rather than linear precedence, is (s)he observing a Grule or
an Lrule? How do we know? Unless these questions can receive a
satisfactory answer, I see no grounds at all for making a
distinction between Grules and Lrules.

G. Vanden Wyngaerd

Re: Martti Nyman (Linguist 3.330)

>Lrules are conventional social norms of which
>competent speakers have intuitive knowledge. It's in virtue of this
>(shared) knowledge that a competent speaker is able to speak correctly
>(or grammatically, if you wish).

Is an Lrule supposed to be a fact about a particular sort of
utterance, such as `say [would like a cookie] when you want
to make people think you want a cookie'? Or is something
else involved?

>What he obviously means is that the theoretical generalization>
> (1) Plural --> es / [Consonant] __
>doesn't extend to new loans such as _poster_.

One thing that has been left unspecified in this discussion is
whether the `new' here is pleonastic. The situation is obviously
quite different if borrowings never come to follow the rule, than
if they simply take a certain amount of time to start doing so.

Avery.Andrewsanu.edu.au
 (currently andrewscsli.stanford.edu)

Without much doubt, the most striking aspect of the conventional object of
linguistic study is _communication_. Communication is the space BETWEEN
speakers. A theory built on this alone will lack explanatory adequacy. It is
never very clear why, in a discourse between A and B, B should be sufficiently
motivated in its inferior role to participate! A theory of the dotted lines in
the 'circuit de la parole' will fail particularly on
two issues:

 1. it will assume constantly successful communication and will omit
 behavioural facts, such as 'false starts', which have too readily been
 excluded from the purview of linguistic theory.

 2. it will ignore the act of utterance and will therefore be unable to
 account for the default status of the antecedent in
 'I dutifully tried..., because that was the idea of one's
 being there.
 or 'On est des voisins'
 or 'Having left daughters only, the property was sold for the
 immense sum of...'
Please, if you are, at this very moment to correct The Independent newspaper,
Le Bon usage, and Boswell, pause to consider that these are 'characteristic'
errors which bid fair to become part of the _langue_. And our theory ought to
include just such variation and change.

A theory of linguistic behaviour, the only proper linguistic theory, cannot
enclose both the determination of the rule-governed ideal speaker-hearer AND
the delineation of a speech-community. The speaker-hearer does not spring
fully-formed into a social grouping not of the speaker-hearer's own perceived
making.

The basis of linguistic theory can be nothing but the ideal speaker-hearer in
its perception of social relationships (wherein lies the trigger of language
("langue") use and extension) and its definition of reality through the use of
language ("langage").