Lecture Four: Detailing the Theory of Meaning: Reference

Philosophy 504

September 8, 2005

I.          One Aspect of Meaningful Language: Reference

A.                 As we have construed it, meaning is representation; in particular, linguistic meaning is to be cashed out in terms of the representational properties of pieces of language (i.e., tokened words, sentences, etc.).

B.                 There is no reason to believe that these representational properties form a homogeneous variety; in fact, we should expect that different parts of language will have different representational functions corresponding roughly to syntactic differences, such as different parts of speech.

C.                 One type of representational property is reference.  Reference is a relational property: significant chunks of language have it when they stand in relations with things outside of language. 

1.                  Pieces of language that have reference as their primary job are called referring expressions.

2.                  Paradigm examples of referring expressions include names and indexicals.

3.                  This relation is a central part of our understanding of linguistic meaning, which is rooted in the ability of language to take us places.  That is, language is important because it serves as a medium that can be used to represent things in the world.  The main mechanism responsible for this ability is reference.

II.        Is Linguistic Meaning Just Reference?

A.                 Reference is important because it is what conveys us from language to the world.  As representation is the capacity of indicating how things stand with (typically) something else, it is clear that reference is a crucial aspect of the representational capacity of language.  But perhaps it is not just a crucial aspect—perhaps it is the only one?

1.                  Language is valuable because of what it can do for us—it helps us survive and thrive.

2.                  Language does this by functioning as a type of window on the world—it is, after all, a medium, which is to say that it stands between language users and the things language users use language to talk about.  The ability of language to function in this role depends on its ability to take us from itself (i.e., the scratches and sounds) to something outside itself.  This ability is, arguably, reference.

3.                  Perhaps, then, meaning just is reference?

B.                 It seems unlikely, for (at least) three reasons:

1.                  Frege’s Problem:  Consider “Hesperus is Phosphorus”.  If reference was all there was to meaning, then this would be synonymous with “Hesperus is Hesperus.”  Right?  But it isn’t—we can do more with the former than we can with the latter; further, we learn more with the former than we do with the latter.  Thus, reference cannot be all there is to meaning.  (This is the main deliverable of Frege’s “On Sense and Reference”.)

2.                  The “Professors of Laputa” Problem: If all parts of a sentence refer, how do we get the unification necessary to account for the fact that propositions aren’t just sets?  Keep in mind that whatever else we point to is a part of the meaning, so on the reference=meaning approach, it can only be accommodated as another referent.

3.                  Syncategorimatic Terms:  Consider ‘and’.  It is meaningful, right?  (It differs in meaning from ‘or’ and ‘but’, although it is closer in meaning to the latter than to the former.)  But to what does it refer?  It makes no sense to think that it derives its meaning from a referent.  Rather, it derives its meaning from the function it serves in sentences of which it is a part—these terms derive their meanings from the sentences in which they occur, based on their functional contribution.

III.       What Refers?

A.                 The second and third arguments above seem to establish that there are meaningful linguistic items that do not refer.  But how extensive is reference?  What linguistic items stand in this relation?

B.                 Varieties of Reference

1.                  Singular Reference:

a.                   Names

b.                  Indexicals & demonstratives

c.                   Definite descriptions—this category has spawned a rich literature devoted to determining just whether or not these are referring terms or, rather, something else.  Russell gets credit for initiating it, with both Strawson and Donnellan contributing important articles.  What is the deal, exactly?

2.                  General Reference:

a.                   Is there such a thing?

b.                  What are we to make of terms like common nouns (e.g., ‘tiger’), adjectival nominals (e.g., ‘red’), adverbs (e.g., ‘quickly’), verbs (e.g., ‘speak’), etc.?  Do these refer?

c.                   Examine this.  There is a projective relationship that obtains between these and the world, but should we take it to be reference?  The Laputa Problem above also demonstrates that not everything can be referential.

d.                  What are the limits that we should place on this?

C.        Other Meaning Categories: quantifiers, predicates, functions, etc.

IV.       Modeling Reference

A.                   Reference is a relation, one that we can symbolize R(n₁, n₂, … , n_n), where ‘R’ stands for the relation, and the “n_i”s stand for the arguments.  That is, R relates all the arguments together.  Our challenge as semanticists is to examine this relation and determine what the arguments are and how they are related to one another.

B.                   Argument Determination

1.                  The first step is to determine how many argument positions there are:  is it a two-place relation, three-place, etc.? 

a.                   We can say that words refer to objects, or that people refer with words to objects, or that people refer with words in specific utterances to objects, etc. 

b.                  How do we distinguish between these?  What arguments can we adduce in support of one over the other?

c.                   My Take: the four-place is where we should start, holding constant certain argument positions for purposes of abstraction.

2.                  What fills the argument positions?  In the four-place relation, it will be speakers, words, utterances, and objects. Adequate modeling will require sophisticated metaphysics here.  (Consider Barker’s Renewing Meaning.)

C.                   Relation Determination

1.                  We need to distinguish this from other relations, e.g., “left of”, “

2.                  We assume that we are after the intension that explains the associated extension.

a.                   Our work above is on point here—do we distinguish singular from general reference?

b.                  Should this be seen as static or dynamic?  Should it be understood in terms of speech acts, or as an abstract rule between types? 

V.        Where Do We Go From Here?

A.                 We need to model the other semantic categories, making sure that we bring them together with reference.

B.                 All of this should be done in a way that respects compositionality, a property of language that corresponds to the “bringing together” of the various categories. 

C.                 We can look at this from the perspective of linguistic content, i.e., the rules associated with the various linguistic items on which we focus.

1.                  Sentences rule here, with propositional contents being the basic unit. 

2.                  Word and phrase content is understood in terms of what these things contribute to the meanings of sentences.  (See the syncategorimatic terms above.)

3.                  A propositional content is a rule associated with a sentence that can be modeled as a function.  Consider Perry’s analysis: “Given such and such, Φ is <true> iff so and so.”  This is his content analyzer.  The contents are the Φs here. 

D.                 Traditionally, sentences have been associated with one content, its truth condition, but this seems mistaken for a host of reasons.  One alternative is to associate them with a variety of contents, and the content analyzer does this. 

E.                  These contents turn out to be the right sort of thing metaphysically, and they can be cashed out in information-theoretic terms.