Against Class A Treatment of no more:
Experimental Evidence from Czech
FDSL-14, Leipzig
Mojmír Dočekal
03-06-2021
Masaryk University
Basic Contrasts
Modified Numerals
1. bare numerals
(1) This chocolate contains 25 grams of sugar.
2
Modified Numerals
1. comparative modifiers (class A) (more than, less than, over,
?no more than, ...)
(2) This chocolate contains more than 25 grams of sugar.
2. superlative modifiers (class B) (at most, at least,
minimally, maximally, ...)
(3) This chocolate contains at most 25 grams of sugar.
extensive research: Büring 2008; Geurts and Nouwen 2007; Nouwen 2008;
Nouwen 2010; Cummins and Katsos 2010; Kennedy 2015; Alexandropoulou
et al. 2016
3
Modified Numerals and Existential Modals
Comparative modifiers can scope under or over existential
modals.
(4) This bus can carry fewer than 45 people.
a. ♦ > fewer than 45 true - coach bus: 55 people
b. fewer than 45 > ♦ true - city bus: 30 people
4
Modified Numerals and Existential Modals
Superlative modifiers have to outscope existential modals.
(5) This bus can carry at most 45 people.
a. *♦ > at most 45 false - coach bus: 55 people
b. at most 45 > ♦ true - city bus: 30 people
Geurts and Nouwen 2007; Blok 2019
5
Ignorance Implicatures
• sometimes related to the Maxim of Quantity: logically
weaker sentences can signal speaker’s ignorance
• comparative modifiers without ignorance implicature
(6) This chocolate contains more than 25 g of sugar. no II
• superlative modifiers with ignorance implicature
(7) This chocolate contains at most 25 grams of sugar. II
6
Testing Czech no more than
Testing Czech no more than
Nouwen 2008 claims that no more than is a comparative
modifier since:
• both scopes in the existential modal env.
• no ignorance implicature (and with scalar bounding
reading)
(8) Cody’s paper is allowed to have no more than 20 pages.
7
Differences between Czech and English no more
English no can act as an determiner:
(9) a. No man arrived.
b. Every/the man arrived.
Unlike Czech no in ne víc which seems to be a focus particle
(10) #Ne/ žádný
no/any
muž
man
nepřijel.
arrived
8
Differences between Czech and English no more
Slavic focus particles (FP) have to (Jasinskaja 2012 a.o.):
• c-command their associated F-marked expression
• be adjacent to the F-marked constituent
• Czech no behaves like all other FPs, as exemplified in (11)
and (12) with a prototypical FP pouze ‘only’
9
(11) Já
I
se
SE
choval
behaved
[seriózně]F
seriously
*ne/pouze.
no/only.
(12) a. I behave only [seriously]F.
b. I only behave [seriously]F.
c. Já
I
*pouze/*ne
*only/*no
jsem
AUX
se
SE
choval
behaved
[seriózně]F.
seriously
d. Já
I
jsem
AUX
se
SE
choval
behaved
pouze/ne
only/no
[seriózně]F.
seriously
10
But the comparative morphology in Czech no more is present:
víc is a comparative of mnoho, než is used in the comparatives
(13) a. Petr
Petr
měří
measures
ne
no
víc
more
než
than
dva
two
metry.
meters
b. Petr
Petr
je
AUX
starší
older
než
than
Marie.
Marie
Summary of no more vs. ne víc diffs: both are build on
comparative base but no is a determiner while ne focus
particle (constituent negation).
11
Two Theories, Two Predictions
1. Nouwen 2008; Nouwen 2010: based on the morphology, no
more than – comparative modifier
2. Kennedy 2015: the difference between comparative and
superlative modifiers comes from the ordering (semantics)
– strict (comparative) vs. non-strict (superlative)
• comparative fewer than 3: max < 3 strict ord.
• superlative at most 3: max ≤ 3 non-strict ord.
• no more than: can be treated as superlative modifier
12
Predictions
♦ >no more than no more than > ♦
Predictions NMC as CM
NMC as SM *
13
Question Addressed by the Experiment
(14) If no more than is SM, it should sound odd in a context
preferring ♦ > no more than interpretation.
Consequences:
• theoretical: support for one type of (modified) numerals
theory;
• distinguishing two types of differentials:
1. regular: slightly less
2. morphologically comparative but semantically superlative
(Czech no more than)
14
Experiment
Design
• joint work with Hana Krajíčková
• two experiments and two research questions:
(15) a. Does Czech no more behave more like a comparative
or superlative modifier (in the modal environment)?
b. Does Czech no more behave like other differential
quantifiers?
Further: exp 2 – it included all the conditions of exp 1
15
Design
• Czech native speakers
• Likert scale 1-5
• the appropriateness of one of the conditions in a context
• truth-value judgment task where a context described a
situation strongly preferring the wide scope of the
existential modal over the degree quantifiers
16
Design
• 16 items and 16 fillers,
• 98 subjects participated in the experiment (implemented
on L-Rex), all of them passed fillers (uncontroversial TVJT)
• four conditions
17
Design
4 conditions:
1. standard comparative modifier (méně než ‘fewer than’):
fewer
2. standard superlative modifier (nanejvýš ‘at most’): at-most
3. no more modifier (ne víc než ‘no more than’): no-more
4. standard differential comparative modifier (trochu méně
než ‘slightly less than’): slightly-less
18
Design
• fewer and at-most tested the acceptability of modified
numerals without differential
• slightly-less, no-more tested the presence of a
differential (vague and zero degree differential)
19
Design
The design was 2x2 factorial:
• comparative vs. superlative modifier (classA,classB) x
• absence/presence of a differential (DiffYes,DiffNo)
• the main conditions:
1. fewer: [+classA,-Diff]
2. at-most [-classA,-Diff]
3. no-more [-classA,+Diff] contra Nouwen 2008
4. slighty-less [+classA,+Diff]
20
Example item
Context: Alex is reading the following sentence on a chocolate
bar packaging:
(16) Toto
this
balení
packaging
může
can
obsahovat
contain
a. fewer
méně
fewer
než
than
b. at-most
nanejvýš
at-most
21
Example item
Context: Alex is reading the following sentence on a chocolate
bar packaging:
(17) Toto
this
balení
packaging
může
can
obsahovat
contain
a. no more
ne
no
víc
more
než
than
b. slighty less
trochu
slightly
méně
less
než
than
60
60
gramů
grams
cukru
of-sugar
Alex says: ’So, in this chocolate bar there can be sometimes
even 65 grams of sugar.’ 22
Results
• mixed-effects linear model with subject and item
intercept+slope random effects (R package lmerTest)
• dependent variable was the subject’s response
• several models, and the one that describes data the best
(the less fitting models included models with main effects
only and models where no more was treated as a CM):
• the model with independent variable conditions classA/B
vs. DiffYes/No and their interaction
23
Results
• negative main effect of classB (SM) (t-value: -11.004,
p < 0.001)
• positive effect of the absence of a differential (t-value:
3.946 p < 0.001)
• a negative interaction of classB (SM) by DiffNo (t-value:
-3.129, p = 0.002)
• at-most was less acceptable than fewer considering that
both of them are without differentials
24
Results
• Tukey’s pairwise comparison of the conditions: only
at-most and no-more were statistically indistinguishable
(t-value: -0.478, p = 0.964)
• all other pairs: differed significantly
• the boxplot representing means and SEs below
25
Boxplot
Figure 1: Boxplot of responses 26
Results
The experiment confirms:
• the scope behavior of Czech no more construction follows
the pattern of superlative, not comparative modifiers
• ↔ subjects rejected no-more to the same extent as
at-least
• the significant difference between no-more and
slightly-less
• ↔ no more is a superlative differential quantifier and
slightly less as a comparative diff quant.
27
Results
Surprising result:
• low acceptability of all conditions: even the most default
comparative modifier without a differential (cond fewer)
had µ=2.51 (SD: 1.61, SE: 0.04)
• possibly priming effect of the most frequent everyday
contexts like (18), which strongly prefer the maxd > ♦
reading, just the opposite against the contexts described
in our exp.
(18) Tato
this
elektrokoloběžka
electric-scooter
může
can
jet
run
méně
fewer
než
than
25
25
km/h.
km/h.
28
Analysis
Analysis
The scope behaviour of Czech NMC is of an superlative
modifier profile
• generally: our exp confirms Kennedy 2015
• implementation: there is no positive difference in degree
between the arguments of the comparative more
1. following Nouwen 2008: analyze German/Dutch nicht
mehr/niet meer as a negative differential expressing
2. nicht mehr α = λP.¬∃d [maxd(P(d)) = α + d ]
29
Analysis
• the negative differential analysis is equivalent to the
superlative at-issue semantics of at most
• in Kennedy’s style of class A/class B analysis, we can
classify Czech no more as a superlative modifier
(19) a. λP.¬∃d [maxd(P(d)) = α + d ]
b. ≈ λP.maxd(P(d)) ≤ α (after Kennedy 2015)
30
Analysis applied
The analysis correctly derives:
1. the wide scope of the class A modifiers no more and
at-most:
maxd(♦contain(ChocBar, d)) ≤ 65g
2. incompatible with Alex’s continuation and predicts low
acceptability of no-more and at-most
31
Analysis applied
The weak surface scope
♦[maxd(contain(ChocBar, d)) ≤ 65g]
allowed only for comparative modifiers
• explains the higher acceptability of fewer and
slightly-less (whatever the reasons for obligatory wide
scope of SM over existential modals are, see Blok 2019)
32
Consequences
1. morphology isn’t always the right clue: Czech no more
behaves as class B, despite its comparative morphology
2. the experiment brings support for the CM vs. SM theory
presented by Kennedy 2015: the distinction between class
A/B = the type of ordering relation (strict vs. non-strict) –
semantics
• Czech no more can be interpreted as ¬ (strict) → ordering
entailments of non-strict ordering
• regular differential quantifiers (slightly-less) remain
strictly ordered, thus class A
33
Cross-linguistic speculations
So far: three types of NMC-languages:
1. no more as class A, English type of NMC (bounding
inferences and both scopes w.r.t. existential modals)
2. no more as class B, Czech type of NMC (only maxd > ♦,
lack of bounding inferences: Dočekal 2017)
3. languages where NMC depending on its realization
behaves as CM or as SM (Hungarian according to Balázs
Surányi (p.c.))
34
Cross-linguistic speculations
The variation seems to be related to the morpho-syntactic
status:
1. a focus particle/constituent negation in NMC (Czech)
behaves as a superlative modifier
2. a negative quantifier (English) in NMC leads to the
comparative modifier behaviour
35
Thank You for Your Attention!
Bibliography
Bibliography i
[Ale+16] Stavroula Alexandropoulou et al. “Pragmatic
inferences with numeral modifiers: Novel
experimental data”. In: Semantics and Linguistic
Theory. Vol. 25. 2016, pp. 533–549.
[Blo19] Dominique Blok. “Scope Oddity: On the semantic and
pragmatic interactions of modified numerals,
negative indefinites, focus operators, and modals”.
PhD thesis. LOT, 2019.
[Bür08] Daniel Büring. “The least at least can do”. In:
Proceedings of WCCFL. Vol. 26. Cascadilla Press
Somerville, MA. 2008, pp. 114–120.
37
Bibliography ii
[CK10] Chris Cummins and Napoleon Katsos. “Comparative
and superlative quantifiers: Pragmatic effects of
comparison type”. In: Journal of Semantics 27.3
(2010), pp. 271–305.
[Doč17] Mojmír Dočekal. “Upper bounded and un-bounded
‘no more’”. In: Acta Linguistica Academica. An
International Journal of Linguistics (Until 2016 Acta
Linguistica Hungarica) 64.2 (2017), pp. 213–231.
[GN07] Bart Geurts and Rick Nouwen. “’At least’et al.: the
semantics of scalar modifiers”. In: Language (2007),
pp. 533–559.
38
Bibliography iii
[Jas12] Katja Jasinskaja. “Information Structure in Slavic”. In:
Handbook of Information Structure (2012).
[Ken15] Christopher Kennedy. “A” de-Fregean” semantics
(and neo-Gricean pragmatics) for modified and
unmodified numerals”. In: Semantics and Pragmatics
8 (2015), pp. 10–1.
[Nou08] Rick Nouwen. “Upper-bounded no more: the
exhaustive interpretation of non-strict comparison”.
In: Natural Language Semantics 16.4 (2008),
pp. 271–295.
39
Bibliography iv
[Nou10] Rick Nouwen. “Two kinds of modified numerals”. In:
Semantics and Pragmatics 3 (2010), pp. 3–1.
40