Did all the boys leave or not

1
Did all the boys leave or not?

• Quantifier-negation interaction in English
• Gunnel Tottie
• Anja Neukom-Hermann
• The University of Zurich

Anja Neukom-Hermann delete dash after Anja
2
The interaction of negation and quantifiers is a
problem

• All, every, some, many
• Many arrows hit the target
• (2) Many arrows didnt hit the target
• (3) Many arrows hit the target and many arrows
  didnt hit the target
• (4) It is not the case that many arrows hit the
  target

3
Universal quantifiers are a worse problem

• (5a) All the arrows didnt hit the target
• (5b) Not all the arrows hit the target (some
  missed)
• (5c) All the arrows not-hit (no arrows hit, all
  missed)
• (6a) All the boys didnt leave
• (6b) Not all the boys left (some stayed)
• NEG-Q
• (6c) All the boys not-left (no boys left, all
  stayed)
• NEG-V

4
Logic versus linguistics

• All the boys didnt leave
• Logicians NEG has scope over the verb, so the
  correct reading is NEG-V
• All the boys not-left (no boys left, they all
  stayed)
• Linguists There is ambiguity NEG-Q is possible
  (and more common)

5
Little empirical research

• Carden and other generativists carried out
  elicitation tests to find different dialects
• First corpus study Josef Taglicht

6
Taglichts study 2.5 million words

• The Brown Corpus (American English)
• The Lancaster-Oslo-Bergen Corpus (LOB British
  English)
• The London-Lund Corpus of Spoken English (LLC
  British)

7
There is a third reading

• All the bills dont amount to 50
• Not all the bills amount to 50 (some
  do) NEG-Q
• (10) All the bills not-amount to 50 (not a
  single one does) NEG-V
• (11) All the bills together do not amount to
  50 Collective

8
Humpty Dumpty sat on a wallHumpty Dumpty had a
great fallAll the Kings horses and all the
Kings menCould not put Humpty Dumpty together
again
9
More collective readings

• All the perfumes of Arabia/Will not sweeten this
  little hand (Macbeth)
• All those aspirins cant have done you any good

10
The BNC study100 million words

• How frequent are allnot constructions in natural
  language?
• Are these constructions ambiguous in context?
• Which use is more frequent (NEG-Q, NEG-V or
  COLL)?
• What are the distributions in speech and writing?
• Can any factors influencing the interpretation be
  uncovered?

11
Only allnot
Anja Neukom-Hermann first and third examples are
not ambiguous anyway...

• Not included in our study
• Every and compounds
• Examples where not precedes all
• Not all you say is true

12
No prosody!

• (15) \/ All the men didnt go NEG-Q
• (16) \ All the men didnt go NEG-V

13
Method

• Perl script by Sebastian Hoffmann
• 2,416 sentences found
• Manual check
• 495 relevant cases found (5 pmw)
• 43 were deemed unclear
• 452 examples analyzable

14
Some genuine examples
Anja Neukom-Hermann 19 is indented and others
arent would look nicer if they were all
formatted in the same way

• (17) We live in a fallen world. All is not true,
  so not everything should be believedNEG-Q
• that small degree of compassionwhich all men in
  whatever circumstance or however degraded should
  not be denied. NEG-V
• (19) If all that money we gave to Band Aid didnt
  do the trick, it must be because there are just
  too many of them. COLL

15
Problematic sentences
Anja Neukom-Hermann mention that 20 is speech?

• With adverbials
• (20)the hunting fraternity is not blameless and
  indeed the, all their arguments are not quite
  correct in every sense...
• With more than one negative item
• (21) there was no positive proof that all this
  had not in fact happened, and that it did not
  belong to a mental lapse.

16
Unclear cases
Anja Neukom-Hermann font size of 22 and 23 is
smaller than previous examples!

• It seemed all of her fathers old acquaintances
  had not been too impressed by his choice of wife.
  NEG-Q or NEG-V
• (23) all that weve done over the last 40 years
  is not as important as what we do next week.
  NEG-V or COLL

17
Figure 1. The distribution of readings of 452
allnot constructions in the BNC
18
Figure 2. Readings of allnot constructions in
BNC (n 452) and Taglichts study (n 23)
Anja Neukom-Hermann there are spaces before and
after the first , but not the second
Anja Neukom-Hermann see also comment in the
field below
19
all not constructions in speech and
writing in the BNC

• The entire sample (unclear included)

20
Fig. 3. Proportions of allnot readings in S and
W in the BNC ()
Anja Neukom-Hermann check all titles of tables
and figures either bold or not bold here you
have a dot after the title - do you really want
that?? doesnt look good alone on a new line
21
Figure 4. Differences between text domains (pmw)
22
Breaking newstwo important factors influence
readings

• 1. Occurrence in formulaic expressions
• 2. The use of all as a pronoun (NP head )or a
  predeterminer

23
Some frozen expressionsare always NEG-Q
Anja Neukom-Hermann I would group lost with
doom since they are semantically similar, as
are well, perfect and good
Anja Neukom-Hermann maybe delete neg-q and coll
in examples here and on next slide, since title
says it already
Anja Neukom-Hermann lost and well should be bold
in examples as on next slide

• all is not lost
• all is not gloom (and doom)
• all is not well
• all is not perfect
• all is not good
• A third aide insisted all was not lost between
  Charles and Diana.
• Sock Shop admitted earlier this year that all
  was not well with its American outlets.

24
Some frozen expressions are always COLL
Anja Neukom-Hermann title on previous slide is
neither bold nor in italics change title on this
slide
Anja Neukom-Hermann enumeration here with dot
and on previous slide with squares

• all NP in the worldnot
• all the marketing and computers in the world
  wont help you.
• the coroner had pointed out how all the
  instruments in the world could not have detected
  it.
• (28)as if all this were not enough, schools have
  started managing their own financial affairs.

25
Completely frozen!
Anja Neukom-Hermann all was not well is not an
example of a modified case here choose rather
all may not be lost (cf. liz paper)

• not a single instance of not all is/was lost, but
  32 instances of all is/was not lost
• no instances of not all is/was well, but 44
  instances of all is/was not well
• More if we count modified cases like all is not
  yet lost, all may not be lost
• No frozen examples of NEG-V!

26
Fig. 5. Free and frozen allnot constructions
Anja Neukom-Hermann again a dot after the title
- delete?
27
Table 2. FREE and FROZEN expressions in all...not
constructions in the BNC
Anja Neukom-Hermann are you going to show this?
not too complicated?
Anja Neukom-Hermann Text durch Klicken
hinzufügen delete

• More frozen examples in W!

28
The other major factor

• The function of all as a pronoun (NP Head) or a
  Predeterminer
• Pronoun All is not lost
• Predeterminer All the boys did not leave
• All of the boys did not leave

29
Fig. 6. Readings all as pronoun or predeterminer
30
Fig. 7. All as pronoun or predeterminer in the
BNC frozen and free use
31
NP complexity plays a part
Anja Neukom-Hermann insert a space after
complex 1

• Pronoun (bare all) All is not lost
• All N All the boys
• Complex 1 All 1 premod or 1 postmod
• All the pretty horses,
• All the money in the world
• Complex 2 All premod and postmod
• All the pretty horses in the field

32
Fig 8 . Complexity and readings of
all-constructions. Free and frozen uses n452
Anja Neukom-Hermann readings in the plural?
33
Fig. 9. Complexity and readings of
all-constructions. Free uses only n 284
Anja Neukom-Hermann readings in the plural?
34
What does all this mean?

• Structural, semantic and pragmatic factors
  contribute to the meaning of allnot sentences
• Truth-functional semantics will not account for
  them
• We need a cognitive framework like frame
  semantics

35
Many problems remain

• Why do speakers use allnot constructions at all?
  
• Why not notall if that is what they mean?
• Are things different when every and its
  compounds serve as the universal quantifier?