Philosophy of Language

Silvio Mota Pinto About the author


Scott Soames. Philosophy of Language. Princeton University Press, 2010, p. 199 ISBN 9780691128664

Silvio Mota Pinto

Departamento de Filosofia. UAM - Iztapalapa. MÉXICO.

Attempts to summarize the achievements of the last one and half century of work and the open problems in the field of the anglo-american philosophy of language there have been various. And yet to my knowledge none have so sharply pointed out the main contributions of both the pioneering work in the field and their follow-up as well as succeeded in diagnosing the most relevant topics of controversy as Scott Soames does in his short recent book titled Philosophy of Language. One may even disagree with certain of his views, while at the same time recognizing his penetrating diagnosis of what the most important theses are and where the deep problems lie.

The book is organized with an introduction and seven chapters, of which the first four belong to the longer first part, while the last three constitute the second part. The first part tells us about the solid ground built by the forefathers and their intellectual heirs. The second rather short part examines such polemical subjects as: a) how to conceive of propositions so that they could play an explanatory role within a theory of meaning for a natural language (NL) together with the notion of a possible state of the world (chapter five); or b) how to understand the interaction between epistemic and metaphysical modalities, particularly in Kripke's preferred examples where we are supposed to know a priori of certain contingent propositions that they are true (chapter six); or c) how best to view the distinction between the semantics and pragmatics of NL when it comes to their respective contributions to the propositions expressed or asserted by literal uses of NL sentences (chapter seven). The first chapter examines Frege's and Russell's contributions to the philosophical study of language. Frege is legitimately considered to be the founder of contemporary philosophy of language by first applying the mathematical notion of a function to the semantic analysis of any scientific language. Frege's second insight consisted in his carefully distinguishing between the sense and the reference of every relevant constituent part of a meaningful sentence, while insisting that both the sense and reference of complex expressions must be conceived as obeying their respective principles of compositionality. Given that truth-values are preserved for sentences when we apply Leibniz's principle of substitutivity of co-referential sub-sentential expressions salva veritate, Frege took them to be their referents. The apparent violation of this principle for attributions of propositional attitudes led him to propose that within these oblique contexts expressions refer rather to the sense they normally possess in an extensional context while their new sense would correspond to a mode of presentation or way of determining the thought in question.

The problem with this proposal is well-known: it leads to a potentially infinite class of indirect higher-order senses which makes it difficult, if not impossible, to explain how this semantic hierarchy could be systematically learned. Soames discusses two alternative approaches to Frege's proposal for accounting for the semantics of attributions of propositional attitudes (the first in terms of acquaintance with senses and the second in terms of a non-extensional that-operator on senses of sentences), only to conclude that Frege's is more problematic than its alternatives.

Chapter one's second part is dedicated to Russell's work. Soames begins with the similarities and differences between Russell's and Frege's accounts of propositions and their constituents. Thus, both use the notion of a propositional function to account for the semantic structure of general statements. According to Frege and Russell after The Principles of Mathematics, the existential and universal quantifiers are second-level functions. Soames rightly criticizes this way of explaining the semantics of quantified statements in terms of higher level predication, which will in turn require explanation in terms of quantification, which must be analyzed away in terms of higher level predication, finally generating an explanatory circle. Soames also criticizes Russell for trying to force the semantic analysis of natural language general statements by using only unrestricted universal and existential quantifiers. Treating natural language expressions like 'all philosophers' or 'most students' as corresponding to genuine semantic constituents of their respective propositions is a more straightforward way of analyzing such English general statements.

Russell's contention that definite descriptions are covert quantified expressions is recognized by Soames as a real insight. As already expected, he disagrees with Russell with respect to the latter's analysis of definite descriptions in terms of unrestricted quantifiers. Better would have been to treat them as restricted quantifiers ('the x: x authored Waverley') and therefore to conceive them as corresponding to semantic constituents of the propositions associated with their respective sentences. However, the phenomenon of scope of an operator, applied to definite descriptions when it interacts with other operators, is mentioned as proof that Russell was right in his contention that definite descriptions are quantified expressions in disguise.

Soames levels two further criticisms of Russell's "On denoting" semantic theory. The first is directed against Russell's famous principle of acquaintance, which requires that the constituents of our worldly thoughts and propositions be cognitively transparent to us. When Russell applies his principle to occurrences of ordinary proper names in sentences, he is forced to hold the weird view that they are no genuine proper names, since we have no direct, infallible epistemic access to their referents. Soames' second critique of Russell's semantic theory concerns the latter's argument to the effect that the expression 'exist' cannot play the semantic role of a first level predicate in statements like 'Aristotle doesn't exist', because positive and negative existential statements involving ordinary proper names are really statements involving definite descriptions and these already contain an existential quantifier. But, as Soames correctly argues, there is no problem in conceiving such descriptive expressions as involving restricted quantifiers ('the x: x authored The Nichomachean Ethics') to whose unique satisfier the propositional function x doesn't exist applies, if the above statement is true.

Chapter two discusses Tarski's analysis of truth and its importance for the philosophy of language. According to Soames, Tarski was led to focus on the concept of truth because he was interested in the expressive power of mathemathical theories and in the possibility of characterizing metatheoretical semantical notions in them. Truth was a central one although it was known since the Ancient Greeks that our pre-theoretical conception of the predicate 'true' leads to paradox. Tarski's proposal was to abandon the pre-theoretical concept of truth in favor of an explicitly defined truth-predicate for certain well-behaved languages, which he then showed how to explicate in such a way that would avoid paradox. As to the relevance of the Tarskian notion of truth for casting light on our pre-theoretical concept of meaning, Soames remains definitely pessimistic.

In the rest of the chapter, he criticizes two later attempts to philosophically explain linguistic meaning appealing to Tarski's truth concept: Carnap's and Davidson's. The semantics Carnap proposes in the 40's aims at clarifying notions like meaning, synonymy and analyticity for all the sentences of correctly regimented scientific languages in terms of Tarskian truth and designation as well as the notion of a complete description of a possible state of the world. Among other difficulties, he rightly complains about the very poor notion of proposition that issues from Carnap's semantics, according to which two logically equivalent sentences express the same proposition. As to Davidson's proposal to account for meaning in NL in terms of the semantic conception of truth, Soames' main objection concerns what he calls the problem of justifying the claim that a given Tarski-style theory that yields truth-conditions for all sentences of a certain natural language would qualify as a correct theory of meaning for that language. Since there can always be many truth theories which are both empirically and extensionally equivalent, then the claim that one of them is the correct theory for interpreting the language in question lacks a reasonable justification.

Having shown that truth-conditional semantics of the kind Davidsonians propose as well as intensional semantics of the sort Carnapians suggest are both inadequate as theories of meaning, Soames proceeds in the third chapter to review the prospects of more recent intensional semantics. Concerning the application of Kripkean possible worlds semantics and its deeper theoretical insight into the intuitive distinction between epistemic and metaphysical modalities, Soames maintains that it still would prove inadequate as a theory of meaning for NL. According to him, one illustration of this inadequacy would be the case of two different necessary and a priori sentences which would be true at every epistemically possible situation, and yet whose respective meaning would intuitively differ.

An interesting and successful application of a possible world semantics to give robust truth-conditions of counterfactuals is provided by Stalnaker and Lewis. The idea is that a sentence like "if it were the case that P, then it would be the case that Q" is true at a state of the world w if and only if Q is true at states of the world w* sufficiently similar to w and where P is also true. Here Soames carefully distinguishes between the possible world semantic account of such conditionals and their philosophical analysis in terms of the notion of causation. According to him, a possible circularity in the analysis of counterfactuals in terms of causality and of the latter in terms of counterfactuals is something that doesn't affect the semantic account of these conditionals' truth-conditions in terms of possible states of the world.

Soames concludes the third chapter with a discussion of Montague's proposal of an intensional semantics for NL. Instead of using first-order logic together with more powerful logical systems in order to regiment NL, Montague proposes more direct syntactic and semantic rules for generating complex expressions from their constituents and for interpreting these constituents and their complexes in terms of extensions and intensions. One of the most surprising features of Montagovian semantics is its classifying quantifier phrases and proper names in the same semantic category, that is: both denote sets of sets (for example: 'John' denotes the set of all sets which include its bearer as an element; 'every man' denotes the set of all sets containing every man). Soames' argument against Montague's similar treatment of proper names and quantified phrases maintains that it is more plausible to suppose that ordinary speakers use NL proper names as expressions designating individuals. His most substantial objections to Montagovian semantics, however, are that: a) as an intensional semantics it is incapable of dealing with sentences attributing propositional attitudes to speakers and b) as a sort of truth-conditions semantics it is strictly incapable of playing the role of a theory of meaning for NL.

Chapter four discusses two more specific features of Kripkean intensional semantics, namely: its semantic treatment of ordinary proper names and natural kind terms via the notion of rigid designation and of indexicals as expressions of direct reference. According to Soames, essentialism, rigid designation and the notion of de re necessity come to play a fundamental role in this type of intensional semantics because these are the doctrines and concepts needed in order to apply quantified modal logic for the semantic analysis of a sufficiently rich language like NL. He praises Kripke's modal argument designed to show that ordinary proper names are rigid designators whereas their associated descriptions are non-rigid. He also endorses Kripke's argument against the view that each name has a descriptive content, whose semantic role would be that of fixing its referent. He finally agrees with the author of Naming and Necessity's rough externalist and communitarian account of how the reference of NL names is determined. Overall he takes the view that the position being attacked by the rigid designation semanticist makes the mistake of conceiving the role of reference determination as an aspect of the meaning of names when this is rather an aspect of their use.

In the second half of chapter four, Soames discusses Kaplan's direct reference semantics for indexicals. The semantics of pure indexicals is taken up first. According to Kaplan, the meaning rules associated with indexicals of this sort relate contexts of use with their respective semantic content, which together with the semantic contents of the other expressions occurring in a sentence have truth-conditions with respect to a possible circumstance of evaluation. Various other features of direct reference semantics are clarified like, for example, the distinction between rigid designators and expressions of direct reference. Further complications come up when it is a matter of providing a semantic treatment of demonstratives, since the meaning rules attributed to them relate contexts of use plus subjective elements like demonstrations or speaker's referential intentions with their respective content. Soames overall assessment of Kaplan's logic of indexicals is mixed: although it brilliantly explains the intuitive a priori character of contingent sentences like "I am here now" and contains also invaluable insights about the meaning of pure indexicals, the logic of indexicals fails, according to him, to provide a plausible semantics for NL demonstratives.

In chapter five, Soames insists on his argument for the semantic indispensability of the notions of proposition and possible world. According to him, an independent theory of propositions is needed, although it is not to be found in Russell or Frege, who took propositions to be intrinsically representational independently of us. Rather, such a theory should be part of a naturalistic account of the representationality of propositions in terms of the intrinsic representational properties of our cognitive states. By conceiving propositions as types of mental cognitive events by means of which agents most basically and atomically predicate properties and relations of n-uples of objects and besides by conceiving these events as objects of first-person acquaintance, Soames hopes to solve the problems related to the old Platonistic account of propositions and in particular the problem of the unity of the proposition, i.e. the problem of predication.

Essentially linked to the concept of proposition is that of a possible state of the world. This is the notion required for explaining the semantic evaluation of propositions with respect to truth. According to Soames, it makes no sense to speak of the truth of a proposition unless it is relative to a possible state of the world w, which is in turn characterized as a maximal consistent set containing either structured true atomic and normally non-modal propositions or their true negations. Of course, this notion of world-state must be enriched in order to account for the semantics of modal and belief propositions; for example, a proposition like possibly there are Higgs bosoms is true at w if and only if it is true at some world-state(s) metaphysically possible from w. It must also be enriched in the sense that it must include singular propositions about objects that do not exist in the actual state of the world. Soames closes the chapter with the rather puzzling remark according to which possible world semantics in the rich sense mentioned above should not be taken as theories of the meaning for NL modal and nonmodal sentences like belief attribution sentences.

The penultimate chapter deals with the epistemic modalities and particularly with Kripke's controversial examples of a priori contingent propositions. According to him, paradigmatic examples of such propositions are those for which the referent of a rigid designator is fixed by the conceptual complex associated with a non-rigid definite description (for example: the proposition that one meter is the length of the platinum stick kept in Paris' Institute for weights and measurements). Soames disagrees, by arguing that knowledge of the singular proposition associated with these cases is normally a posteriori, that is: based on his own perceptual experience, the reference fixer knows of this length [one meter] that it is the length of the famous Parisian stick. Better examples of a priori contingent, according to Soames, are propositions of the form p if and only if actually p (for instance: Princeton University has a philosophy department if and only if actually Princeton University has a philosophy department), where the actuality operator applies to propositions and predicates of them the property of being true at the actual state of the world.

Philosophy of Language's last chapter deals with the controversial question of how to draw the boundaries between the respective provinces of NL semantics and its pragmatics. Soames discusses, more specifically, the relations between the semantic (meaning) and pragmatic (contextually determined presuppositions, conversational implicatures, etc.) contributions to the propositions literally asserted by the utterances of NL sentences (S) or expressed by uses of S in thought. Two conceptions of such relations are discussed. According to the traditional conception, the semantic content associated with concrete literal uses of S is always a complete proposition, which might be enriched as a result of pragmatic factors operating in the context of communication, whereas Soames conceives such semantic content or meaning as a set of constraints on literal uses of S, which in the case of sentences with demonstrative indexicals or incomplete descriptions doesn't suffice to determine a proposition but requires pragmatic information shared by language users to do so.

If Soames is right, then the most adequate way to draw the line between NL semantics and pragmatics would proceed by identifying the invariant minimal content common to all literal uses of NL sentences-its semantic contribution to the proposition asserted or expressed by such uses-and the difference between asserted or expressed content and invariant semantic content belongs to the province of pragmatics.

In my opinion, the greatest merit of Soames's book is that of finding a perfect balance between the lucid and penetrating exposition of the most relevant problems in contemporary philosophy of language with carefully thought-out solutions to them. All those who are really interested in the topic must read it.

Publication Dates

  • Publication in this collection
    24 June 2013
  • Date of issue
    June 2013
UNICAMP - Universidade Estadual de Campinas, Centro de Lógica, Epistemologia e História da Ciência Rua Sérgio Buarque de Holanda, 251, 13083-859 Campinas-SP, Tel: (55 19) 3521 6523, Fax: (55 19) 3289 3269 - Campinas - SP - Brazil