SciELO - Scientific Electronic Library Online

vol.29 número2Introduction: Referential descriptions: for and againstA plea for (purely) singular propositions: The cases of belief correction and de re attitude reports índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados



  • No hay articulos citadosCitado por SciELO

Links relacionados

  • En proceso de indezaciónCitado por Google
  • No hay articulos similaresSimilares en SciELO
  • En proceso de indezaciónSimilares en Google


Análisis filosófico

versión On-line ISSN 1851-9636

Anal. filos. vol.29 no.2 Ciudad Autónoma de Buenos Aires nov. 2009



Definite descriptions, misdescriptions and semantic content: different ways to solve a tricky puzzle*

Justina Díaz Legaspe

Universidad Nacional de La Plata - CONICET - GAF


Michael Devitt (2004, 2007) claims that the predicative material that constitutes complex referential expressions makes a semantic contribution to the proposition expressed. He thus deviates from direct referentialism, according to which every referential expression -either simple or complex- contributes just with an object to the proposition expressed, leaving the predicative material out of the semantic content. However, when dealing with misdescriptions, Devitt has suggested a pragmatic way out: the audience can understand what the speaker is referring to even if the object does not fall under the corresponding description. From my perspective, this proposal questions the semantic validity of the predicative material, together with Devitt's original claim. In this paper, I propose a way to solve the problem posed by misdescriptions that appeals to the idea of epistemically relativized properties, according to which the properties ascribed to the object -by means of the predicative material- correspond to the way the speaker thinks of it and not to the way the object really is.

KEY WORDS: Definite description; Reference; Misdescription.


Michael Devitt (2004, 2007) sostiene que el material predicativo que constituye las expresiones referenciales complejas hace un aporte semántico a la proposición expresada, alejándose así del referencialismo directo, para el cual toda expresión referencial -sea ésta simple o compleja- contribuye sólo con un objeto singular a la proposición expresada. Sin embargo, al enfrentarse al problema de las descripciones fallidas, Devitt ofrece una salida pragmática: el oyente comprende a qué se refiere el hablante aun cuando el objeto referido no caiga bajo la descripción utilizada. Esto pone en cuestión la validez semántica del material predicativo, desestimando la postura original de Devitt. En el trabajo propongo una solución a este problema, apelando a la idea de propiedades epistémicamente relativizadas, de acuerdo con la cual lo que se predica del objeto por medio del material predicativo corresponde a las creencias del hablante, y no a lo que el objeto realmente es.

PALABRAS CLAVE: Descripción definida; Referencia; Descripción fallida.

1. Michael Devitt (2004, 2007) poses two requirements on complex referential expressions -that is, those constituted by a determiner (i.e. 'the', 'that') and a property term constituting the predicative material:

(i) there must be a causal-perceptual link to an existing object grounding the singular thought they help to express;
(ii) the predicative material must make a semantic contribution to the proposition expressed.

Hence, a sentence containing any of those kinds of expressions will be true only if the object causally linked to the subject -the intended object- falls at the same time under the property term; that is, sentences like 'That F is G' or 'The F is G' will be true only if the objects referred to by their respective subjects are actually Fs. In this regard, Devitt's position stands not only against Russellian and neo-Russellian views on sentences containing those expressions, but also against referentialist perspectives that take definite descriptions to be directly referential terms, contributing just an individual to the proposition expressed.1 In what follows I will make use of the classical problem raised by misdescriptions in order to suggest how a moderate referentialism as the one Devitt is proposing could be articulated by modifying the second requirement above-mentioned so as to offer a non-pragmatic answer to the problem at stake.

2. Let's suppose that my friend and I are at a bar and I say, with a particular individual in mind:

(1) The man in the corner drinking a martini looks like my uncle

There is, indeed, a man in the corner looking exactly like my uncle, but he is not drinking martini but plain innocent water poured in a martini glass -who knows why. In any case, my friend understands what I have said and nods in agreement: we both have the same individual in mind. Among other things, there are two important questions to answer regarding the sentence at stake. In the first place, we need to know whether it is intuitively true or false. This is not an innocent question, since, as we shall see, depending on the proposed answer, one perspective or the other will award itself the honor of being the theory that best respects our semantic intuitions. In the second place, we have to ask for the content of the sentence, the proposition it expresses, which should agree as much as possible with those intuitions.
Is the sentence true or false? Contrary to what happens with obviously false sentences and obviously true ones, this question receives only stuttering answers from non-philosophers: some claim that it is true, since the person I want to speak of does look like my uncle (the referentialist intuition); some others claim that it is false (the generalist intuition) since no one in the bar is drinking a martini -and looking at the same time like my uncle (Donnellan 1966). If common sense is to be preserved, both sides must be captured by the theory.
What kind of proposition does the sentence express, then? And, more particularly, what role does the predicative material play in the sentence? Here is where waters divide. As it is widely known, Russell (Russell and Whitehead 1910) would have formulated (1)'s content in terms of:2

(R) (the x : Fx) Gx

Thus, the sentence would be true were there one and only one individual both drinking a martini (F) and looking like my uncle (G).3 On this view, the predicative material involves a property that enters the content of the proposition expressed. Being 'the F' a misdescription of the individual intended by the speaker, the sentence comes out false, since no one in the bar satisfies both F and G. This does not respect the intuitions of those who claim that the sentence is true, on the basis of the fact that the individual referred to by the speaker by means of the misdescription is indeed a G. Respecting those intuitions would imply considering the definite description involved as a singular term and the proposition expressed as a singular one. And (R) is not, alas, a singular proposition but a general one. On the other side, any theory yielding (1) as false is forced to explain the referential success of the speech act, namely, the fact that in uttering (1) I have successfully communicated with my friend. In order to do just this, neo-Russellians have appealed to pragmatic derivations from the content expressed to the content meant.
In contrast, a direct referentialist who considers the expression 'the F' as a genuine singular referential device can extend Kaplan's insights on the meaning and content of the sentences containing demonstratives to the treatment of sentences containing definite descriptions. According to Kaplan, the character of a demonstrative has the role of determining the object of reference but it does not enter into the corresponding content; instead, it is just the object referred to the one entering the content, and hence the proposition expressed by the sentence (Kaplan 1978 and 1990). Extending Kaplan's ideas, the direct referentialist takes the predicative material to be similar in role to a demonstrative's character: a mere device for determining the intended referent, some help for the audience. Thus, were a the logical name corresponding to the intended object in (1), the expressed proposition would be something like:

(K) Ga

In this case the proposition expressed is clearly singular. As should be appreciated, direct referentialists can easily respond to the problem raised by misdescriptions: since the predicative material is just a tool for helping the audience determine the referent, it does not need to provide us with a property really possessed by the object -if my friend knows that the man is drinking water but is clever enough to know that I may think, because of the shape of the glass, that he is drinking a martini, he will know where to look at in order to assess the man's likeness to my uncle.4 The sentence turns out true in the situation presented.
Now, as far as Devitt's position is concerned, he wants to have it both ways. He wants (1) to express a singular proposition and not a general one, pace the Russellians; he also wants the predicative material constituting the description to play a semantic role, pace the Kaplanian direct referentialists. Although he does not say how he conceives of the logical form of sentences like the above one, we can imagine the following options:

(D1) Fa Ga


(D2) (the x : Fx x = a) Gx 5 - 6

On both options, (1) comes out false since the first conjunct is false in (D1) and because a does not bear the F property in (D2). Therefore, the view has to face the problem of the referential success of (1)'s utterance. Following the neo-Russellians on this point, Devitt seems to have opted for a pragmatic move: the audience is capable of going from the expressed false proposition to the true one that includes the intended object a by means of a Gricean derivation of a singular proposition about a. Now, when defending RD, Devitt had opposed the neo-Russellian endorsement of a pragmatic move.7 Looking for an alternative that avoids the move in question, in what follows, I will propose to introduce a slight difference in his order of ideas, considering the possibility of a purely semantic answer.

3. We want, then, several things: we want (1) to express a singular proposition. We want the property expressed by the predicative material to get into the proposition expressed. We want that proposition to respect and explain the oscillations of intuitions regarding the sentence's truthvalue, and now, on top of it, we want to avoid appealing to any kind of pragmatic derivation. Can we have it all?
The predicative material of complex referential expressions does help the audience to pick out the object intended by the speaker; thus, the audience will look for the object falling under F within the perceptual or epistemic context of speech. However, as Kaplanian referentialists would hold, that function does not require that the speaker should select a predicate truly applying to the intended object; in order to draw the audience's attention to it, it is enough that the speaker believes that the object bears a certain property, or that the speaker believes that the hearer believes that it does. Thus, in a situation where both speaker and audience believe that an object is F, although it is not, the sentence 'The F is G' will successfully communicate a thought about that particular object. Now, Devitt's second requirement for singular reference could be slightly modified so as to reflect that trait: accordingly, there would be no need for the intended object to fall under F, but instead it would be mandatory that the speaker or the audience believes that it does. To this aim, we can introduce the idea of epistemically relativized properties signaled with a star, as in 'F*a' to refer not to the property F, but to the extension the speaker attributes to the property F in her frame of beliefs. If we stipulate that those epistemically relativized properties are the ones that give rise to the predicative material, the content expressed by (1) should be something along the lines of:

(D1*) F*a Ga


(D2*) (the x : F*x x = a) Gx

In order to achieve this relativization, three things should be done: first, definite descriptions should be treated as referential expressions; second, the context of utterance must be widened to contain, among other things, the speaker's beliefs as a parameter; if not all, at least those related to (i) her representation of the world surrounding her and (ii) her representation of her audience's beliefs. Third, the definite description's referential assignment must be linked somehow to that special contextual parameter, so that it can get its reference not from the actual world (from the fact that a is actually an F) but from that epistemic parameter (from the fact that the speaker thinks that a is an F).
In order to clarify what may seem to be nebulous, we could try and define a * property like F* in the following way:

(the x : F*x x = a) Gx is uttered with truth by a speaker S if and only if S believes that Fa and it is the case that Ga

If this is so, the consequence will be that the predicative material associated with complex referential expressions should be always located within the scope of an hyperintensional operator such as BELIEVES.8 This makes sense, since part of the information transmitted by (D*1) or (D*2) is that the speaker believes or believes that another person believes) that a can be described as an F; contrary to what a direct referentialist would claim, (D*1) and (D*2) do not convey that a certain object has a certain property, but that the speaker believes that a certain object can be depicted in a certain way.
Then, an adequate report of (D*1) and (D*2) as uttered by S would be:

(2) (BELIEVES S) (the x : Fx x = a) Gx

Adding the additional information that it is indeed the case that Ga would put matters in this other way:

(3) (BELIEVES S) ((the x : Fx x = a) Gx) ∧ Ga

which corresponds to the definition of the sentence (1) with the starred-property F* included in the predicative material.
This, in turn, would explain why we have oscillating intuitions regarding (1)'s truth-value: on one hand, it is true that Ga; on the other hand, although a is not an F, S believes that it is or believes that the audience believes that it is. So, although (1) is logically false, (3) is true, which explains why a sentence containing a misdescription in subject position can be referentially successful.

4. Now, such a formulation of the content of (1) assures us of several things. First of all, it locates the property involved by the predicative material within the content of the sentence, as Russellians do. Second, since it contains an individual constant (or a functionally equivalent expression in the case of (D*2)), the proposition involved is indeed singular. Third, it explains how it is that any competent speaker oscillates when attributing a truth-value to (1): as it can be seen in the clearer (3), it is true that Ga, and it is true that the speaker thinks that a is an F, but it is not clearly true that Fa, and therefore, the evaluation of the sentence becomes unclear. Finally, even when it includes the property involved by the predicative material within the content expressed, it manages to explain the referential success in almost the same way in which the direct referentialists do: since the predicative material is under the epistemic mode, my friend will understands that what I am intending to refer to is not actually an object falling under F, but an object I take (or I take him to be taking) to fall under F, which is clearly not the same thing (specially not in cases of a perceptual or a conceptual mistake).
In sum, if the properties constituting the predicative material in complex referential terms could be epistemically relativized in this or in some other way, it would be easier to satisfy Devitt's requirement of including them into the proposition expressed by the sentence without preventing it from being singular. Moreover, the relativization helps us avoid the pragmatic jump from the proposition expressed to a proposition meant. Finally, it provides us with a nice answer to the misdescription problem, since it accounts for the oscillation between truth-values perceived by a competent speaker in front of sentences containing misdescriptions, and, at the same time, it also explains the referential success associated with such sentences.


* I want to thank Michael Devitt, Eleonora Orlando, Laura Skerk, Ramiro Caso and Juan Manuel Berrós for their comments to a previous version of this paper.

1 A typical Russellian position is presented in Neale (1990). Such perspective combines Russell's theory of descriptions with Gricean pragmatic derivations, and is inspired by Kripke's (1977) paper.

2 Here we present a more modern formulation of their idea. See Neale (1990).

3 " _ looks like my uncle" is, of course, a relational property which should then include a place for the individual constant referring to my uncle. However, we will take it as if it were a monadic predicate. Nothing in this paper depends on this point.

4 That is, if I know that he is drinking water but I take it that my friend is unaware of this fact.

5 I take it that, although this is technically a quantified sentence, the functional equivalence between a singular term or individual constant a and a expression like (the x : x = a) can work to express, at the same time, the existence and uniqueness implied by the determiner, and the fact that the sentence refers to a particular individual. On the other hand, Soames chooses this formula to express the content of partially descriptive names. See his (2002) paper.

6 Regarding this point Devitt remains neutral on the possible ways in which the logical form of these sentences could be formulated. That includes my (D1) and (D2).

7 That is, in this particular regard. It is clear that a Gricean pragmatic story is the best way to explain a number of other cases in which what is conveyed is not literally what is said.

8 BELIEVES is hereby being regarded as a relational operator linking a believer (S) and the object of her belief.


1. Devitt, M. (2004), "The Case for Referential Descriptions", in Reimer, M. and Bezuidenhout, A. (eds.), Descriptions and Beyond, Oxford, Clarendon Press, pp. 280-305.        [ Links ]

2. Devitt, M. (2007), "Referential Descriptions and Conversational Implicatures", European Journal of Analytic Philosophy, 3 (2), pp. 7-32.        [ Links ]

3. Donnellan, K. (1966), "Reference and Definite Descriptions", Philosophical Review, 77, pp. 281-304.        [ Links ]

4. Kaplan, D. (1978), "Dthat", in Syntax and Semantics, vol. 9; reprinted in Yourgrau, P. (ed.), Demonstratives, Oxford, Oxford University Press (1990), pp. 11-33.        [ Links ]

5. Kaplan, D. (1977), "An Essay on the Semantics, Logic, Metaphysics and Epistemology of Demonstratives and Other Indexicals", in Almog, J., Perry, J. and Wettstein, H. (eds.), Themes from Kaplan, Oxford, Oxford University Press, pp. 481-564.        [ Links ]

6. Kripke, S. (1977), "Speaker's Reference and Semantic Reference", in French P. A., Uehling, T. E. and Wettstein, H. K. (eds.), Contemporary Perspectives in the Philosophy of Language, Minneapolis, University of Minnesota Press, pp. 6-27.        [ Links ]

7. Neale, S. (1990), Descriptions, Cambridge, Mass., The MIT Press.        [ Links ]

8. Russell, B. and Whitehead, A. N. (1910), Principia Mathematica, 1, paragraph 14, Cambridge, Cambridge University Press.        [ Links ]

9. Soames, S. (2002), Beyond Rigidity, Oxford, Oxford University Press.         [ Links ]