Mark Israel <misrael@scripps.edu> writes:
>>> "All ... not" can be ambiguous:  "All the people who used the bathtub
>>> did not clean it afterwards."  "Not all" is free of this ambiguity.
>>
>> [...] one would normally say: `None of the people who used the bath 
>> cleaned it afterwards', if that were what one meant.
>
>   I disagree.  I'd say that "All the people who used the bathtub didn't
>clean it afterwards" is a complaint, and that one is more likely to 
>complain about "all the people" than about "none of the people".  What
>do others think?

I think you're dead right.  The solution you propose is exactly
the way most people (imho) would resolve the logical ambiguity
posed by the Quantifier - Negative construction.  Note that it's
not a grammatical solution, but a pragmatic one (in both the
popular [~= 'practical'] sense of 'pragmatic' and the technical
[= 'context-sensitive behavioral evaluation'] sense).

The ambiguity isn't in the logic or the grammar, but is produced
by their interaction.  Logic suggests that the unambiguous formulation
with a quantifier (like 'all') and a negative (like 'not') should depend
on their order, with the earlier being the more general:

   Not all the people who left were angry. ['Not' outranks 'all']
   All the children received nothing.      ['All' outranks 'not']

And that works in some cases.  But the grammar of English has
specific rules about where in the sentence one may place a
negative, especially when negating the verb; it pretty much
*has to* go after the first auxiliary verb:

 (*Not) The (*not) man (*not) has (not) read (*not) the (*not) book (*not).

There are also rules for quantifier placement.  Semantically,
quantifiers (all, some, any, much, each, every, few, a few, lot(s),
1, 2, 3, etc.) must bind nominals of some sort, which makes them
a special sort of adjective in many cases, and thus they may appear
modifying nominals:

   All the people arrived on time.

But they're also logically implicated with the entire proposition, so
some are often said to be able to "float" [not my favorite metaphor,
but we're stuck with it in the technical vocabulary as "Quantifier
Float"] over to an adverbial position:

   The people all arrived on time.

And, I hasten to add, quantifiers are irregular as hell, and one
shouldn't try floating quantifiers at random; this is just the
tip of the iceberg.

The combination of these two rules leads sometimes to situations
in which there are two logically possible interpretations of a
given sentence, like the one above, and then one must find some other
way to distinguish the Q-Neg reading ('for all X (not (P(x)))') from
the Neg-Q reading ('not (for all X (P(X)))').  The most likely
way is something along the lines suggested, because the whole point
of talking is to communicate something, and the listener must consider
purposes in trying to figure out what is being (putatively)
communicated.  Grammar is just traffic rules, to help us get started.

There's a very large literature on quantifier ambiguities.  Guy
Carden did the definitive early studies in the 60's & 70's, and
many others have contributed since then.

---Followup...
   another message on the same topic, with the above as context:

> I'm no linguist, but I trust my ear enough to take on someone who
> obviously is.  Be gentle.
>
> Before we discuss this ambiguity, I deny that it exists.  The
> sentence 'All...did not clean' means that nobody cleaned, period.
> We accept it in speech as hyperbole, just as we understand what
> a child means when he says 'Nobody likes me'.  

I think you should continue to trust your ear.  I find your observations
about usage and meaning generally perceptive and subtle.  So, no problems
there. It's some of the conclusions you appear to draw from them, and
the presuppositions they appear to be based on that I would respectfully
disagree with.

First, about whether the sentence

   All the people who used the bathtub didn't clean it afterwards.

is ambiguous; you deny that the ambiguity exists.  The only thing your
statement can mean is that *you* don't find it ambiguous.  I doubt you
were intending to pontificate about what other people could find
ambiguous, just your own use. Correct me if I interpreted you wrong,
please, and then we can look into telepathy together :-)

A lot of people find it unambiguous; that's not an uncommon phenomenon,
and it's been researched extensively. I mentioned Guy Carden's work in
my previous post; he has looked at "quantifier dialects" and how they
change.  That is, he has studied how people interpret logically
ambiguous sentences containing both a quantifier and a negative, like:

     All the boys wouldn't leave.

(that's not guaranteed to be one of his examples, by the way -- and this
 discussion is not guaranteed to be an adequate summary of Carden's work
 either, nor of the other researchers who have studied this topic. It's
 been about 20 years since I read up on the subject and I haven't made a
 trip to the library for this, so it's just from my fallible memory.
 As you said, be kind. :-) 
If you're persistently curious, however, you can now find a complete bibliography of Guy Carden's publications on his home page at
http://www.arts.ubc.ca/ling/gcard98.htm. Just search on that page for uantifi and you'll get all the relevant titles.
What he found is that ambiguity judgements are unstable, and that,
like most other things humans have to cope with, repeated exposure
to such ambiguity by speakers usually makes them gravitate to
one or another strategy for disambiguation, usually whatever works
in context in their experience.  But they don't all gravitate to
the same pole, which suggests that the choice is a pragmatic one,
not a matter of "correctness" inherent in the construction or the
grammar.

My guess would be that you have achieved this adaptation and now have a
pragmatic strategy (based, it appears, on interpreting these utterances
as hyperbolic) that serves you well in perceiving and interpreting them.
Congratulations.  But, as with all personal adaptations, it's usually a
mistake to expect others to have identical experience, or to draw
identical conclusions from it.  And it can occasionally cause trouble to
depend on that happening.

> Eventually the other meaning may become idiom, but till then it's
> a simple blunder in written English.

As I've mentioned before on a.u.e, I would hesitate to call any blunder
'simple' without evidence.  This one strikes me as rather complex, in
fact.  But there is no doubt that it is a blunder in written English,
precisely *because* it strikes some people as ambiguous. It *is*
ambiguous in use - summing over the entire set of English speakers -
because even those like you who find it unambiguous, do not do so in
the same way, so there is no way to predict whether speakers and
listeners will be using the same convention. Unintended ambiguity is
almost always a serious blunder in writing.

>> There are also rules for quantifier placement.  Semantically,
>> quantifiers (all, some, any, much, each, every, few, a few, lot(s),
>> 1, 2, 3, etc.) must bind nominals of some sort, which makes them
>> a special sort of adjective in many cases, and thus they may appear
>> modifying nominals:
>>
>>    All the people arrived on time.
>>
>> But they're also logically implicated with the entire proposition, so
>> some are often said to be able to "float" [not my favorite metaphor,
>> but we're stuck with it in the technical vocabulary as "Quantifier
>> Float"] over to an adverbial position:
>>
>>    The people all arrived on time.
>
> I don't see how 'all' can modify anything except 'people' in this
> sentence.  It is in a position usually occupied by an adverb, true,
> but that is immaterial - it is there for a completely different
> reason.  The 'all' is placed after the noun simply for emphasis.

Well, "modification" isn't really at issue here.  There are at least
three parameters involved in Q-Float constructions like this:

 logical:     which Noun Phrase is bound by the quantifier "all"?
 grammatical: what is the syntactic category status of "all"?
 pragmatic:   what distinction if any does Q-Float signal in this case?

The logical parameter requires that there be some NP that functions as
a variable for the quantifier.  I think that's what you mean by
"modification", but I'm not sure.  In any event, the binding is
the same in the Floated and unFloated versions, so I'm not denying
that modification is going on here.

But "modify" doesn't just mean "bind"; it also has a grammatical
component.  A word that modifies another word must be placed in a
particular syntactic relationship with it.  The grammatical parameter
requires that "all", like any word, be put in an acceptable place in the
sentence.  Unlike Latin, English doesn't have an inflectional system, so
we use word order to do the same things Latin did with its inflectional
system, and we use it very thoroughly. This thoroughly, in fact:

   All the sweaty people must have been running this morning.
   The sweaty people all must have been running this morning.
   The sweaty people must all have been running this morning.
   The sweaty people must have all been running this morning.
  ?The sweaty people must have been all running this morning.

The issue here isn't whether these "mean the same thing" or not, just
whether they mean at all.  I find the last one questionable. Try this
again with "each" and a similar though not identical pattern emerges.
The places where "all" can go are adverb positions, between auxiliary
verbs, as we both agree.

So why *isn't* it an adverb?  Because "it's there for a completely
different reason"?  Sorry, that's no explanation.  Anybody can make up
their own rule and use it, but that doesn't necessarily apply to others,
and it *certainly* doesn't apply to the sum of all possible uses by all
speakers.  Which is the responsibility of grammar, if it can be said to
have responsibility.

Finally, before everyone's eyes glaze over, let me say again that your
perceptions about the potential senses of these sentences strike me as
right on.  What you might consider, however, is that they are highly
context-sensitive, and that in a sufficiently different context you
might well not come to the same conclusion.

And that if you are tempted to formalize your (admittedly correct and
useful) intuitions about the interpretation of these sentences into
universal grammatical rules, it's best to be very clear from the outset
about describing what's going on.

This is probably much more discussion than you had in mind, and I'm
really not trying to drown you.  Sorry once again, but the issues (as I
keep saying) really *are* complicated. Language is alive, just as we
are, and it's just as complex as any other living thing.

  --------------------------------------------
 -John Lawler                     More grammar
  Linguistics Program   University of Michigan
 "..and, who knows? Maybe the horse will sing."