LINGUIST List 3.387

Wed 06 May 1992

Disc: Rules, Tone Grammar

Editor for this issue: <>

Directory

1. Eric Schiller, Re: 3.371 Rules
2. Martti Arnold Nyman, On rules
3. Richard Sproat, Tone Grammar Summary
4. , Tone Grammar

Message 1: Re: 3.371 Rules

Let me see if I got this right:

1. We want to develop a theory of grammar which applies to all (human?)
languages.

2. Ideally, our model will conform to "some notion of economy", whatever
that is (C/c/k/Kase theory, Binding Theory and Theta Theory are only
economical in the sense that it is hard to put a price on something which
changes on a weekly basis, so we might as well call it inexpensive).

3. To do so, we presumably need some information on what the properties
of the languages under discussion actually are. To this end, any
explicit form of grammatical description should be useful.

4. Ergo, producing grammatical tools and explicit grammars can help
to produce the required input to the explanatory device.

I was raised to believe that the goal of grammatical theory was,
in the end, an explanation of why language behaves as it does. But
I do recall that descriptive adequacy was a step on the path. The
trend is clearly to tweak one's theory whenever required. Having
read Chomsky (1992, Jan 13), we can now see that all the mental
gymnastics involved in getting analyses involving the now-defunct
projection principle to work are only of marginal impact.

On the other hand, the classic and well-written descriptive grammars
in now-defunct theories are still useful to those of us who are
striving to achieve the elusive goal of explanatory adequacy.

I picture much of contemporary syntax as a neurosurgeon slashing
away with high-tech lasers, without having first taken a bit of
anatomy. Sometimes the results are right, even highly impressive,
but not nearly so good as if the surgeon had as much knowledge as
skill.

What is lamentable about the present situation is that we hold
explanatory adequacy to have such a higher staus than descriptive
adequacy, that much of language desciption is considered sub dig
for graduate students who aspire to a career.

This could lead to a dangerous imbalance in our profession, if
allowed to go unchecked. We should aim to produce linguists who
are capable and knowledgable, who have enough contact with human
language to be able to contribute to the development of theories
of language, at least insofar as it is possible to prevent some
wrong turns.

Example: Would Mark Baker have invested so much work in getting
serial verbs to comply with the Projection Principle if he was
aware of the many exceptions to the "universal" configurational
restrictions he proposed? No only was there clear evidence in
a dialect (footnote as abnormal) in the language under discussion
(Sranan), but plenty of evidence from other, unrelated languages
such as Khmer. Now there was no way he could have known about the
Khmer data, since few grammatical descriptions of the language have
dealt with serial verbs and compounding. I do not criticise Baker
for not knowing about the counterexamples (prior to his 1990 CLS
visit, anyway), but rather that unless we, as a community, create
rewards for high-quality descriptive work, the theoretical community
is driving blind.

Eric Schiller
University of Chicago

Re Guido Vanden Wyngaerd (3-357):

I must break my longish silence caused by a few urgent jobs:

> since Nyman doesn't seem to begin to understand
> what the problem is here, let me expand on my question once more: we find
> that speakers *observe* (not violate) the wh-island constraint; we find
> that they form yes-no questions by having recourse to hierarchical
> structure rather than linear precedence, i.e. they unfailingly produce (1a)
> rather than (1b):
>
> (1) a Is [the man who is tall] __ in the room?
> b Is [the man who __ tall] is in the room?

I must admit my dull-wittedness. Sorry about that! I also agree that
a real issue is involved here.

> If speakers proceeded on the basis of inductive generalisation or analogy
> or some such principle, one would expect a more or less random distribution
> over (1a) and (1b) in the acquisition stage, quod non.

Why would a more or less random distribution over (1a) and (1b) be
expectable in the acquisition stage, if speakers unfailingly -- and so,
as the only pattern for children to base inductive generalizations on --
produce (1a)?

> Hence speakers
> possess a certain knowledge or follow certian rules, and one would like to
> know where they get this knowledge from.

Are you suggesting Chomskyan "tacit" knowledge? If so, then you might
think that rules grow in people's minds like biological organs.

> The question now is: how can one tell if rules speakers
> follow, such as the wh-island rule or the yes-no question formation rule,
> are Grules or Lrules?

Where does the conviction come from that speakers "follow" the
wh-island rule? This rule, as formulated and so named by a grammarian,
is clearly a theoretical generalization (Grule). Notice that the
phenomenon here captured in insular terms is in principle amenable to
a different theoretical formulation: e.g., in Dick Hudson's Word Grammar
(_English_Word_Grammar_. Oxford 1990), which is a dependency-based
approach, (1b) would be ruled out by the fact that it infringes the
Adjacency Principle (not to be confused with GB's Adjacency).
 While it is grammatically correct to say
 (2a) Is the man who is tall in the room?,
it would be grammatically incorrect (and nonsensical) to say
 (2b) *Is the man who tall is in the room?
In my terminology, (2a-b) are technically rule-sentences which exemplify
(and witness for the existence) of the corresponding norm of language
(Lrule). It is typical of norms that they may be violated by mistake or
by joke (cf. Alexis Manaster Ramer's [3-371] posting on _man_the_ as
a [putatively] possible NP); sometimes also by force, in order to
bring home a theoretical point [e.g. (1b) above].
 Lrules (or norms-of-language qua institutional or cultural facts)
are typically learned or acquired by experience. But some norms are
so deeply rooted in human nature that their violation is more or less
unnatural and requires an extra effort. (This is one of the basic tenets
of Stampean natural phonology, unless I'm mistaken.) For example,
anyone standing on two hands (instead of two legs) in a cocktail party
would certainly violate a norm of socially correct behaviour. In this
case, the 'two-leg' constraint is almost vacuously a norm, because it
would be hard to violate it. In the same vein, the Lrule of which (2a) is
an exemplification, is almost vacuously a norm, because only a linguistic
professional could violate it by producing (2b). But it is a norm,
nonetheless.

> And what merit is there to making such a distinction
> at all?

This distinction is (meta)descriptive of generative grammarians' actual
practice.

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

> Finally, Lansing points to English:
>
> > A _silver knife_ (low high) is a knife made out of silver, and a _silver
> > knife_ (high low) is a knife for cutting silver.

This has nothing to do with tone, but rather accent placement. Both
_silver_ and _knife_ are accented in the former case, but in the
latter case _knife_ is deaccented. The association between accent and
tone is a great deal more complicated than the claim suggests.

Richard Sproat
Linguistics Research Department
AT&T Bell Laboratories			tel (908) 582-5296
600 Mountain Avenue, Room 2d-451	fax (908) 582-7308
Murray Hill, NJ 07974			rwsresearch.att.com

None of the examples mentioned in John Cowan's recent posting
summarizing responses on this topic appears to involve a language
that uses language SOLELY to mark grammatical as opposed to lexical
distinctions. Which makes one wonder about stress. Does anybody
know of an example of a language in which stress is distinctive
but only used to mark grammatical categories. I have some possible
examples in mind but these need some further reflection, so if anybody
has clear examples, I for one would be interested (the examples I
am thinking of, which I am not too sure about, would involve
Tongan and perhaps Persian).