Page 339 - Encyclopedia of Philosophy of Language

importantquestionistowhatextentthesyntacticand semantic viewpoints can be harmonized.
1. Syntactic and Semantic Contrasts
Here are two pairs of examples (1-2), one modal and the other epistemic, which illustrate the distinction (Quine 1976):
eachn-placeatomicrelation-symbolRofy,which n-tuples of members of X satisfy R at w and which do not,andforeachrigiddesignatorof&,whichobject it designates at w. If weW, then a 'qualitative dupli- cate' w' of w can be constructed by choosing a set Y of the same size as X such that Y contains every xeX denoted in Jt by a rigid designator of 3?, and a particular 1-1 correspondence f between X and Y, and then by replacing xeX with f(x)eY throughout the description of w (the denotations of the rigid des- ignatorsstaythesame).ThenifJlisapossibleworlds model, one defines a Fine-weakening of M to be a model Jt' with the same actual world as Jt but in which each other world has been replaced by a quali- tative duplicate (allowing different choices of Y or f for different weW). Next, one can say that a sentence is essentially de re if it is not equivalent to a de dicto sentence: '(Vx)D(Fx=3Fx)' is de re but not essentially so, since it is equivalent to 'D(Vx)(Fx =>Fx).' The main result, due to Fine, is that a sentence a is essentially dereiftherearemodelsJt andJt',Jt' aFine- weakening of Jt, such that a holds in one model but not the other (Fine 1978). Reflection on these definitions indicates that what is disrupted in the move fromJttoJt'isthepatternofrecurrenceofspecific objects from world to world. This precisely pins down the semantic feature of de re sentences underlying the syntactic criterion.
3. The Epistemic Case
There is little prospect of anything comparably rig- orous and comparably significant for epistemic contexts. On one view, the very legitimacy of such forms as (2b) and (4) can be disputed, on the grounds that the position of the term in (4) is not open to substitution by a coreferential term (even if you are John Smith, Ralph might deny that John Smith is a spy) and hence not open to quantificational binding (Quine 1976). In reply, one could challenge either the failure of substitution claim, for there is no failure of substitution of rigid designators in the modal case, or else the idea that quantification is acceptable only if substitution does not fail (Salmon 1986; Kaplan 1986). To make further progress, one would then have to work out the semantic intuition underlying the de re label in the epistemic examples, the intuition that in (2b) and in (4) Ralph's belief is in some sense about an object, while in (5), assuming he deduced 'the shor- test spy is a spy' from 'there are finitely many spies and no two have the same height,' his belief in that same sense is not about an object. The nature of this contrast is still controversial (Burge 1977; Forbes 1987).
See also: Modal Logic; Necessity; Reference: Philo- sophical Issues.
D(3x)(x is greater than 7) (3x)Q(x is greater than 7)
Ralph believes that (3x)(x is a spy) (3x) Ralph believes that x is a spy.
(la) d b ) (2a) (2b)
(la) and (2a) are de dieto, for each has a complete proposition in the scope of its governing context. But (lb) and (2b) are de re, for each has a variable which is free within the modal or epistemic context. This syntactic difference corresponds to an obvious sem- antic one: (lb) and (2b) say that there is an object such that it is necessarily greater than 7 or that it is believedbyRalphtobeaspy,whichexplainsthelabel 'de re.' Contrast (2a): it does not say that there is any particular object which Ralph believes to be a spy.
However, there is a problem about classifying sen- tenceswithexpressionswhichpickoutspecificindi- viduals, for example, in (3-5):
D (Aristotle is a member of the species homo sapiens) (3) Ralph believes that you are a spy (4) Ralph believes that the shortest spy is a spy. (5)
The problem is whether each should be classified as de dicto, since each contains a complete proposition within the scope of the governing context, or as de re, since each seems to concern a specific object.
2. The Modal Case
The syntactic and semantic classifications for modal contexts can be harmonized by making use of the notion of a 'rigid designator' (Kripke 1980). A rigid designator is an expression such that, whenever it picks out x at one world and y at another, then x = y. So 'Aristotle' and 'you' (relative to a fixed context of utterance) are rigid while 'the shortest spy' is nonrigid. One can then define a 'de dicto sentence' of a modal language to be one which contains no occurrence of a rigid designator within the scope of a modal operator, and no occurrence of a variable within the scope of an operator whose binding quantifier is not within the scope of that operator. Example (3) is therefore de re (for (4) and (5), see below).
One can now explain exactly what the de dicto/de re distinction amounts to in the modal case. A possible worlds' model Jt for a language & consists in a col- lection of worlds W (with one singled out as the actual world) and a collection of possible objects X together with, for each weW, a description which says, for
De Dicto/De Re
317