Judgment Ascriptions
Kjell Johan Saeboe
University of Oslo
2nd International Conference on Meaning and Quotation (ICQM2)
Zentrum für Allgemeine Sprachwissenschaft, October 16 – 18, 2008
1 Subjective Attitudes
Across languages, a family of propositional attitude verbs seem to require that
the embedded clause contain some “subjective predicate” –
– a predicate of “personal taste” (PT), as in the Swedish (1),
– a non-PT relative adjective in the positive, as in the German (2), or
– a modal used in a deontic or bouletic sense, as in the French (3).
(1)
Några tycker att kärnkraftverk är vackrare
än vindmöllor.
some think that nuclearworks are beautifuller than windmills
‘Some find nuclear power plants more beautiful than windmills.’
(2)
Die Preise sind zwar
normal für Eau de Parfums, aber ich
the prices are tobesure normal for eau de perfumes but I
persönlich finde trotzdem, dass sie hoch sind.
personally find nonetheless that they high are
‘The prices may be normal for perfumes, but I still find them high.’
(3)
90% trouvent que chacun
devrait comprendre au moins une
90% find
that everybody should understand at least one
langue étrangère en plus
de sa langue maternelle.
language foreign in addition of his language maternal
‘90% feel that anybody should know at least one foreign language.’
1
Unless the complement contains a suitable predicate, the sentence is infelicitous,
as are the Norwegian (4) and the French (5).
(4) #Mange forskere
synes at dinosaurene ble utryddet
av et
many researchers seem that dinosaurs were extinguished by a
voldsomt kometnedslag for 65 millioner år
siden.
violent cometimpact for 65 million years since
(5) #Les Catholiques romains trouvent que les âmes de ceux qui
the Catholics Roman find
that the souls of those who
meurent sans
avoir entièrement satisfait à la justice divine, vont
die
without have entirely
satisfied to the justice divine go
après la mort dans le purgatoire.
after the death into the purgatory
It is not obvious what makes these sentences semantically deviant. To be sure,
there is an intuition that it does not make sense to entertain a subjective attitude
to something which is either a fact or not a fact, irrespectively of the subject;
but this is still far from an explicit and explanatory analysis.
Henceforth, the relevant verbs will be referred to as subjective attitude verbs.
2 Theories of Judgment
According to Lasersohn (2005), PT predicates depend on (a time, a world, and)
a judge, a semantic index on par with (time as treated in PTQ and) the world
(→ Relativism). The intension of a word like beautiful is a nonconstant function
from judges to more traditional intensions.
Stephenson (2007) ascribes to PT predicates and epistemic modals an argument
occupied by PROJ , denoting the actual judge, or pro, denoting a contextually
salient entity. The extension of a word like beautiful is a nonconstant function
from judges to more traditional extensions.
Glanzberg (2007) argues that the parameter of judgment should be a contextual
parameter (→ Contextualism), an experiencer class E to be set by the context.
Stojanovic (2008) argues that it does not matter for semantics whether it is a
semantical or a contextual parameter, since semantically, contextualism can do
everything that relativism can. She presents a contextualist semantics without
a judge index but with a distinguished variable xT occupying the implicit judge
argument associated with PTPs.
One can sort theories along two axes: whether PTPs have a judge argument,
and whether there is a judge index; cf. Figure 1. I will argue that the bottom
right case is best for accounting for the facts about subjective attitudes.
2
+ judge index
÷ judge index
÷ judge
argument
Relativism
(Lasersohn)
Contextualism
(Glanzberg)
+ judge
argument
Corelativism
(Stephenson)
Cotextualism
(Stojanovic)
Table 1: theories of judgment
3 Shifting Judges
According to Lasersohn (2005), a preposition like for can serve to fix the judge
to the individual denoted by the complement (in a simplified notation):
Prepositional Judge Shift according to Lasersohn (2005)
[[ α for β ]] j,t,v = [[ α ]] b,t,v where b = [[ β ]] j,t,v
Because according to Stephenson (2007), a PT predicate denotes a function
from judges, the preposition can be simply the identity function on individuals:
Prepositional Judge Shift according to Stephenson (2007)
[[ for ]] v,t,j = λye y
When a PTP occurs with a for phrase, or, in Slavic languages and to a certain
extent in German, a dative phrase, it does not license a subjective attitude verb:
(6)
Jeg synes spillet
er morsomt (# for alle). (Norwegian)
I seem game-the is fun
(# for all)
This suggests the following generalisation:
(7)
Generalisation
A subjective attitude verb is only felicitous with a complement clause
whose intension or extension is a nonconstant function from judges.
Some scholars have suggested that verbs like consider , find , or think can fix the
judge of their object to their subject. In the formulation of Nouwen (2007: 5),
“such an approach would have to say that”, i.a.,
x thinks 1 y is tasty is true in world w with respect to judge j iff,
in x’s think-worlds, y is tasty is true with respect to judge x
Stephenson’s (2007) proposal is (in a simplified notation):
3
Attitudinal Judge Shift according to Stephenson (2007)
[[ think ]] v,t,j = λφ(s(i(et))) λze ∀hw, t0 , xi ∈ Dox v,t,z : φ w,t0 ,x = 1
This “meaning for think . . . has the effect that a sentence of the form “x thinks
that S” is equivalent to “x thinks that S is true as judged by x”. In many cases
this will be equivalent to “S is true as judged by x” . . . ” (Stephenson 2007).
Lasersohn’s (2007) suggestions concerning consider are similar.
The common denominator of these analyses is this: The subjective attitude verb
is reduced to the corresponding objective attitude verb modulo the judge shift,
transforming φ to that constant function from judges that assigns to any j φ(x).
(8a) (on one reading) is predicted to mean more or less the same as (8b):
(8)
a.
b.
She thinks the game is fun.
She thinks the game is fun for her.
Stephenson does not treat unambiguous subjective attitude verbs like find , but
if one were to treat them in the same way as think , there would not be a strong
basis for predicting the infelicity of (4) or (5), etc.; (9a) would be predicted to
mean more or less the same as (9b):
(9)
a. #Homer finds Bart gay.
b. Homer believes Bart is gay.
4 Shifting Judges (cont.)
Consider this option: The sole contribution of the attitude, verb and subject, is
to fix the judge, cf. the semantic definition of the Norwegian verb synes in (10);
then what is wrong with (4) and (5) is that the whole attitude is redundant: It
maps a judge-invariant function onto itself.
(10)
synes 0v,t,j = λφ(s(i(et))) λx φv,t,x
Alternatively, the judge could be treated as an argument of the taste predicate,
dispensing with the judge index:
(11)
synes 0v,t = λφ(s(i(et))) λx φv,t (x)
Actually, the time index can be dispensed with as well:
(12)
synes 0v = λφ(s(e(it))) λx φv (x)
A taste predicate like beautiful could be defined as follows:
(13)
beautiful 0v = λxe λye λs Bv (x)(y)(s)
And after (13) has combined with mary and the result has combined with (12)
and the result again has combined with anne, we get:
4
(14)
λs Bv (m)(a)(s)
( anne find mary beautiful )
The ye argument is usually implicit. Unless filled by something explicit through
attitude verbs, prepositions or cases, it must be filled by a contextual parameter,
pro (Stephenson) or a distinguished variable, xT (Stojanovic 2007: 701): “there
are restrictions on the range of individuals plausibly assignable to xT .”
- I, you, the cat, people in general . . .
- see Moltmann (2007) for a discussion of these questions of perspective.
Note that this analysis and its relativist counterpart, encoded in (10), give two
different answers to why subjective attitudes need judge-sensitive complements:
assuming (10), (4) and (5) are interpretable, but the attitudes are superfluous;
assuming (12), the composition terminates because of a type conflict.
Still, as yet there is no obvious reason to prefer contextualism over relativism.
Stojanovic writes that while the choice cannot be decided on semantic grounds,
considerations about pragmatics, syntax, or their interfaces with semantics may
well push us the one way or the other. Some such evidence follows.
5 Locality
Not only do subjective attitudes require the embedded clause to contain some
subjective predicate, this predicate must be the main predication in the clause.
As it appears, it must be narrowly in focus, and it must not be embedded. It
cannot even be conjoined with a judge-insensitive predicate:
(15)
a. You must be handsome and below 45. (web example)
b. #She finds him handsome and below 45.
c. She finds him handsome and pleasant to be with. (web example)
This tells in favour of the cotextualist analysis – (co)relativism offers no way to
predict the b case, because the attitude will have free access to the judge index
unless it is bound by something in between. On the other hand, on an analysis
with a judge argument, this will have to be filled by xT to avoid a type conflict.
The argument must be saturated for the intersective conjunction to succeed.
Several classes of facts can be adduced to strengthen this argument.
(16)
a.
Hun
she
‘She
b. ??Hun
she
synes alle røykere er usympatiske.
seems all smokers are unpleasant
finds all smokers unpleasant.’
synes alle som er sympatiske, er ikkerøykere.
seems all that are pleasant
are nonsmokers
In (16a), the judge argument of usympatiske is accessible to the attitude if the
subject undergoes QR. (16b) should mean more or less the same, but even if
the subject is not QRed, the judge argument must be saturated in that DP.
5
(17)
a.
Jeg synes du er gift
med en vakker mann.
I seem you are married with a beautiful man
‘I find the man you are married to beautiful.’
b. ??Jeg synes du kjenner en vakker mann.
I seem you know a beautiful man
(17a) is felicitous because everything in the complement but the taste predicate
vakker ‘beautiful’ can be read as presupposed, as indicated by the paraphrase;
such a reading is much more difficult to get with (17b).
6 Time
The sketched analysis is reductionist and faces a problem with tense and time:
while there are two finite verbs and thus two tenses in the attitude construction,
the attitude verb translation (12) only has room for one tense. Consider first:
(18)
Music is always connected to the personal imagination of beauty. I am
growing, I am changing and my imagination of beauty is also changing.
What was beautiful for me today is not necessarily beautiful tomorrow.
Yesterday I did not find this piece of music beautiful; today I do.
Here the complement is nonfinite, so no problem: the set of states given by (19)
will (through a covert aspect be transformed to a set of times and) be filtered
through the past tense or the present tense and finally be anchored to the day
before the time of utterance or the day including the time of utterance.
(19)
λs Bv (m)(i)(s)
But what if the complement is finite so there is another, maybe different, tense
there? Then in some way, that second tense must be encapsuled , as we might
encapsule the embedded tense in the two synonymous Norwegian sentences (20a)
and (20b) along the lines of the paraphrase (20c):
(20)
a.
b.
c.
Jeg synes det er bedre i år enn i fjor.
I seem it is better in year than in fore
‘I think it is better this year than last year.’
Jeg synes det var bedre i fjor enn i år.
I seem it was better in fore than in year
‘I think it was better last year than this year.’
I find the way it is this year better than the way it was last year
I find the way it was last year better than the way it is this year
This way, the better state can reach the present tense of the attitude verb.
7 Double Indexing?
One proposition can depend on two different judges, cf. (21).
6
(21)
The mother snipe thinks that the ugliest baby birds are beautiful.
This could be handled in a relativist analysis by double indexing, insulating the
ugliest baby birds from judge abstraction with an actuality operator on judges.
But it is as easy to handle it in a cotextualist analysis: the ugliest predicate can
have its judge argument saturated by the designated variable (here presumably
to be anchored to something generic) while the beautiful predicate can have its
judge argument shifted to the mother snipe through the attitude verb.
8 Standards and Ordering Sources
Subjective attitude verbs emerge as diagnostics of judge-variant morphemes,
and in addition to taste predicates,
1 the positive morpheme (Kennedy 2007) and
2 modals with normative ordering sources (Kratzer 1981)
prove sensitive to j or project a judge argument, cf. (2) and (3).
As for 1, the judge argument should be attached to the contextual parameter
brought along by this formative, according to Kennedy (2007), the standard s:
The Semantics of pos according to Kennedy (2007: 17):
[[ pos ]] = λgλx g(x) ≥ s(g)
Here s is a context-sensitive function from measure functions to degrees that
returns a standard of comparison based on properties of the adjective g and on
features of the context of utterance, “in such a way as to ensure that the objects
that the positive form is true of ‘stand out’ in the context of utterance, relative
to the kind of measurement that the adjective encodes” (Kennedy 2007: 17).
To make the dependence of s on a judge visible, the definition of the positive
could be augmented to include a judge (and a state) argument:
(22)
pos 0v = λgλxλyλs [ gv (x) ≥ s(y)(g)](s)
As for 2, the judge argument should be attached to the contextual parameter
encoding preferences, according to Kratzer (1981), the ordering source g: (23)
is true in v in view of modal base f and ordering source g iff the proposition
that everyone understands a foreign language is a necessity wrt. f (v) and g(v),
where g(v) can well be the speaker’s preferences in v.
(23)
Chacun devrait comprendre au moins une langue étrangère.
A sensitivity to judges could be built into this definition in the following way:
(24)
f,g(x)
ought 0v f,g = λφλx O v
(φ)
7
Mostly, though, s and g will not genuinely depend on the judge but be constant
wrt. this argument, so that it is idle.
9 Conclusions
(1)
If a – semantical or contextual – parameter of judgment is assumed,
subjective attitudes shift it, and otherwise mysterious facts about
subjective attitudes constitute evidence in favour of assuming one.
(2)
If subjective attitudes are indeed assumed to shift the judge parameter,
a contextualist theory in terms of a judge argument is better equipped
to account for the facts than a relativist or a contextualist account
in terms of a shiftable index or parameter.
References
Cappelen, Herman (2008) “Content Relativism and Semantic Blindness”, in Kolbel and Garcı́a-Carpintero (eds.) Relative Truth, Oxford: OUP, 265–286.
Glanzberg, Michael (2007) “Context, Content, and Relativism”, in Philosophical
Studies 136, 1–29.
Kennedy, Chris (2007) “Vagueness and Grammar: The Semantics of Relative and
Absolute Gradable Adjectives”, in Linguistics and Philosophy 30, 1–45.
Kratzer, Angelika (1981) “The Notional Category of Modality”, in Eikmeyer and
Rieser (eds.) Words, Worlds, and Contexts, Berlin: de Gruyter, 38–74.
Lasersohn, Peter (2005) “Context Dependence, Disagreement, and Predicates of
Personal Taste”, in Linguistics and Philosophy 28, 643–686.
Lasersohn, Peter (2007) “Relative Truth, Speaker Commitment, and Control of
Implicit Arguments”, to appear in Synthese 94.
Moltmann, Friederike (2007) “Relative Truth and the First Person”, semanticsarchive. net.
Nouwen, Rick (2007) “The (absolute) truth about taste”, handout for presentation
at Szklarska Poreba 8.
Stephenson, Tamina (2007) “Judge Dependence, Epistemic Modals, and Predicates of Personal Taste”, in Linguistics and Philosophy 30, 487–525.
Stojanovic, Isidora (2008) “Talking about Taste: Disagreement, Implicit Arguments, and Relative Truth”, in Linguistics and Philosophy 30, 691–706.
8