LINGUIST List 3.330

Thu 09 Apr 1992

Disc: Rules

Editor for this issue: <>

Directory

1. Martti Arnold Nyman, Rules of Language vs. Rules of Grammar
2. Martti Arnold Nyman, re Rules: three-level ontology of phenomena
3. Swann Philip, 3.250 Rules
4. David Stampe, What theories are about
5. , Saussure & rule-government

The discussion re Rules has caught on to the effect of running
into danger of dissolving. There's a lot of threads I'd like to
follow, but it may be in order to re-start from Jon Aske's original
posting on the existence of rules (3.231):

> At the risk of seeming contentious, I must take exception to Jose
> Ignacio Hualde's statement that "nobody would doubt that the rule
> of plural formation for Spanish words ending in a consonant is to
> add /-es/; but new borrowings may not undergo the rule, as in
> poster-s (*poster-es), cf. the integrated dolar-es 'dollars'."

At this point let me advert to the following conceptual differentiation:
Rule of Language (Lrule) vs. Rule of Grammar (Grule).
 Lrules are rules of linguistically correct behaviour. Lrules don't
cause behaviour. Because Lrules exist as institutional facts (on
which see Dyvik's posting in 3.281 & 3.297), they impose social control
on behaviour. Accordingly, 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). The speaker may also choose to speak
incorrectly (ungrammatically); in such a case, the others know that
the speaker is speaking incorrectly and are able to act appropriately.
 Grules are generalizations over Lrules (or, more precisely, over
rule-sentences which refer to Lrules; see below). Grules are theoretical
descriptive entities expressed by means of some system of notational
conventions (eg. the TG formalism). About Grules the linguist has
no intuitive knowledge before s/he has carried out the descriptive task.
 When we say that the competent speaker follows a rule, we mean that
the speaker follows a Lrule.
 When we say that the competent speaker knows the rules of her/his
language, we mean that s/he is able to follw (or not to follow, if
s/he so chooses) the Lrules of her/his language.
 Incidentally, I claim no originality to the views expressed above.
 They are compatible with what has been put forward by Esa Itkonen
 in (eg.) _Linguistic Theory_and_Metascience_ (1978). And I believe
 they are more or less in accordance with what Helge Dyvik has presented.

Now, when Jose Ignacio Hualde states that new borrowings may not
undergo the rule of Spanish plural formation, he must be referring to
a Grule: What he obviously means is that the theoretical generalization
 (1) Plural --> es / [Consonant] __
doesn't extend to new loans such as _poster_. Insofar as it is the task
of the linguist's grammar to generate all and only correct/grammatical
(word, morpheme, phoneme) strings of a given language, the Grule (1)
is falsified by *_posteres_ (which it generates) and by _posters_ (which
it fails to generate). Notice that the falsification is made not on the
basis of some spatio-temporal occurrences but in virtue of two rule-sentences:
 (2) a. *_posteres_ (= _posteres_ is incorrect);
 b. _posters_ (= _posters_ is correct),
which get their force from the social existence of a correspondig Lrule.
(Notice that I'm taking Hualde at his words; however, I'm aware of the
possibility that there may not be an established norm/Lrule concerning
the plural of _poster_. In that case, of course, I'm presenting my point
for the sake of argument.)

> This may seem like a minor point, but for those of us who are
> interested in determining what a speaker's real psychological
> linguistic knowledge is, saying that there is a rule X that
> speakers know, but that they just happen not to apply that rule
> to new words, or in a certain number of occasions, doesn't make
> much sense.

Right. It doesn't make sense to say that "there is a rule X that speakers
know, but that they just happen not to apply that rule to new words".
Contrary to what is suggested by syntax, the two occurrences of "rule"
in this quote aren't coreferential. In the first clause, "rule X that
speakers know" can be only the Lrule exemplified in (2) above; Grules, like
(1), can't be the object of shared intuitive knowledge. In the second clause
(ie. "happen not to apply that rule to new words"), "rule" must refer to
Grule (1), 'ontologized' in the mind/brain of the speaker. Clearly, if
speakers wouldn't apply (or follow) the Lrule, they would be uttering
some incorrect form (eg. *_posteres_); so, the meaning here is obviously
that the correct/grammatical form _posters_ results from not applying
Grule (1), qua psychologically realistic rule, to the word _poster_.
 But is Grule (1) psychologically realistic? By Chomsky's standards,
not, I would guess, because (1) would impute ungrammatical behaviour
(uttering [posteres]) to the speaker. But this can't be the whole story.
What does a competent speaker of Spanish know? Well, s/he knows (inter alia)
the Lrule(s) that _posters_ is correct and _posteres_ is incorrect. While
the linguist's grammar represents this knowledge by excluding all incorrect
forms, this is conceivably not the speaker's way of storing and retrieving
her/his knowledge. Besides knowing what is (in)correct in her/his native
language, the speaker is also more or less conscious of the structural
possibilities of her/his language. It is on the basis of the latter type
of "knowledge" -- I'm reluctant to call it knowledge -- that the competent
speaker is able to infer forms, part of which may be socially incorrect
(ungrammatical) yet structurally regular and natural. Such inferences are
based on analogy. For example, in English, *_foots_ is incorrect as the
plural of _foot_, yet it is more than a mere error: it's a piece of
overregularization, and as such it has a principled basis. The same goes
for quite a many self-invented asterisked forms that linguists invoke
in order to bring home some theoretical point.

> For example, as I have argued elsewhere (BLS 16, 1990), the
> different rules [Grules/MN] that have been proposed for Spanish stress, in
> spite of being about 95% accurate (about 5% exceptions), do not
> seem to have any real existence in the speakers heads apart from
> the words in the lexicon themselves. So when the speakers are
> presented with new words they have never seen before they do not
> go to the rules to tell them how to stress them, but rather they
> go to the lexicon and extract the pattern right from there.

Here the word "lexicon" is used in a systematically ambiguous way.
When speakers "go to the lexicon and extract the pattern right from
there", they're using their ability to analogize. It is precisely
the analogizing ability that is psychologically real.

Martti Nyman, Dept of General Linguistics, Univ of Helsinki, Finland

In 3.305, John Coleman writes:

> I would suggest that the various aspects of language: biological, mental,
> social, ..., and in my view Ideal too are complementary, rather than
> contradictory. After all, what is the motivation for attempting to
> promote one such aspect as more fundamental or important than the others?

I think this is a healthy position, and it was in this spirit that I
talked (3.266) about three-level ontology in terms of Karl Popper:
physical phenomena belong to Popper's World-1; mental, to World-2;
and social, to World-3. None of these phenomena can be reduced to
something else. They are complementary.

> And the limitless productivity of
> linguistic constructs such as centre-embedding doesn't fit well with a
> theory of language which doesn't admit Ideals.

Well, Ideals too belong to the Popperian World-3, but I still object to
viewing human language as a mathematical object. It's true that mathematical
methods can be fruitfully applied for descriptive purposes, but this per se
doesn't define human language as a sub-case of formal languages.
 For example, centre-embedding is NOT limitlessly productive in human
language. (On productivity, see eg. David Crystal, Introducing Linguistics,
Penguin English 1992, p.59.) On the contrary, centre-embedding tends to
be avoided altogether, except in classical Latin literary prose, where
it occurs, though not limitlessly.

Martti Nyman, Dept of General Linguistics, Univ of Helsinki, Finland

Here are some comments on the discussion regarding the status of
linguistic rules.

(1) When an argument runs for as many decades as this one has, there
are (at least) three possible reasons: (a) it's a philosophical
problem that cannot be untangled by appeal to empirical evidence; (b)
it's a terminological problem due to mutual misunderstanding:
different people (insist on) use the same words with different
meanings; (c) it's about a genuine scientific question which will
eventually be resolved by empirical evidence. I would guess that here
we are dealing with a lot of (b), some of (a) and probably none of
(c): so in this note I will consider only (b).

(2) What does it mean to say that something is psychologically real?
In the present context it seems to mean "in the mind". So what is the
mind? For me, the mind must be limited to the actual or possible
*experience* of higher animals. In other words, the mental must, at
least in principle, be accessible to conscious awareness. In contrast,
many cognitive scientists postulate an *unconscious mind* filled with
symbols, logical inference engines, generative grammars and other
magical reincarnations of obstruse mathematical textbooks. All this
would be harmless speculation, if it were not for the fact that this
second mind is largely constructed by projecting inwards some of the
observed properties of the conscious mind (Chomsky is a classic
example). After a several fruitless decades, this approach is no
longer so popular, and connectionist models are taking over from
symbolic ones. For me, then, the only coherent use of the term
"psychologically real" is in the first sense.

(3) The word "rule" has many meanings, but the central one is that of
a prescriptive or procedural formulation of a social practice. Such
rules are (usually) propositional statements that can be learned,
stated, followed, ignored and so on. The formulation of a rule should
not be confused with its application, and neither process can be
isolated from their contexts. Less central, but related meanings,
included the use of rules to constitute or describe a practice. In
general rules can be considered as arbitrary conventions established
in a community by a process of negotiation and diffusion. In the
formal sciences, procedures for calculation or syntactic manipulation
are often referred to as "rules": and it is via this usage that rules
found their way into cognitive science and became part of the supposed
machinery of the unconscious mind. This is stretching the original
central meaning of the word to the point of contradiction, and has
inevitably led to confusion.

(4) As several people noted, there is no problem in talking about
rules (or rather, formulations of rules) in the mind - provided that
both terms are defined as I have suggested above. Nobody would
seriously question the psychological reality of a spelling rule
presented by a teacher to their class. Likewise, the rules invented by
grammarians are certainly present in their minds. In such cases,
psychological reality can be established by empirical investigation.
What is established, 'though, is the merely the reality of the rule
*formulation*. How (or even if) the rule formulation gets applied to
generate behaviour is a very different question, to which nobody has
an answer. One possible answer is that the question is a meaningless
abstraction that will go away when we know more about the neurological
infrastructure of language processing (and this would also apply to
"rules" that are imagined to exist in the postulated unconscious
mind).

Philip Swann
TECFA-FPSE
University of Geneva

In 3.323, Larry Hutchinson <hutchincs.umn.edu> writes that "FACTS do
not become EVIDENCE (for or against a theory) unless the theory is
about those facts." For example, "facts about speech errors are not
evidence with regard to phonological theories unless those theories
are about speech production."

But isn't it an empirical question what a theory ought to be about?
If the nature of phonology were due in part to the nature of speech
production, then shouldn't a theory of phonology to that extent also
be about speech production? And similarly speech perception, and
auditory memory, to the extent they make phonology what it is?

Or to take a stronger case, if the form of phonological rules echoes
exactly the ways the conditions on their application in speech (their
inner hierarchies, their interrelations, etc.), as I argued in my 1972
dissertation, then isn't a theory of phonology that is not also about
speaking simply wrong?

Do we decide what theories are about, or does nature?

David Stampe <stampeuhunix.uhcc.hawaii.edu>, <stampeuhunix.bitnet>
Dept. of Linguistics, Univ. of Hawaii/Manoa, Honolulu HI 96822

In my desire to respond appropriately but diplomatically to an interesting
insight (JJ36 in 3.309/3) into the way that the post-saussurean school deals
with 'the father of modern linguistics', I failed to give detail of my claim
that 'early' Saussure had already set up for himself important linguistic
tenets. His note 6 (Geneva, ms. fr. 3951) reminds us that:
> Nature gives us man organized for articulate _langage_ but without articulate
> _langage_...The individual, organised for talking, can arrive at the stage of
> using his apparatus only by way of the surrounding community...
> The social fact of _la langue_ may be compared with the uses and customs
> (constitution, law, customs, etc). Further on are art and religion, which are
> manifestations of mind ["esprit"], where personal initiative has an important
> part to play, and where there is no assumption of any exchange between
> individuals..._Le langage_, property of the community, like the usages,
> responds in the individual to a special organ prepared by nature...And as the
> goal of _le langage_, which is to make onself understood, is of necessity
> absolute in every human society, in the state in which we know them, it
> results that the existence of a _langage_ is proper to any society .....To be
 >
> developed: Necessary existence of _le langage_ in every human community.

I apologise for having had to quote at such length. Perhaps, by way of
compensation, I could offer a small puzzle: for French specialists, please
translate the words of the following sentence which are in upper case

 The speakers of any LANGUAGE can accomplish a great many communicative tasks
 with the sentences of their LANGUAGE.

If LANGUAGE is translated by the same word each time, please justify this. If
there is a difference between the words used in any French translation, please
say what the difference is.
 -----------------------------------------------------------------------------
I would like to return now to the more pertinent
question:

viz_The reality of linguistic rules_.
1. have we agreed on what KIND of rule a linguistic rule can be?
2. has anyone offered a definition of REALITY?
3. can linguistic rules be more REAL than a map?
4. are maps "real"?