part 1 - 4. Reference

part 1 - 4. Reference, Materiały naukowe, The Hanbook of Pragmatics

[ Pobierz całość w formacie PDF ]
4. Reference : The Handbook of Pragmatics : Blackwell Reference Online
Page 1 of 16
4. Reference
GREGORY CARLSON
Subject
Theoretical Linguistics
Ç
Pragmatics
Key
-Topics
Topics
reference
DOI:
10.1111/b.9780631225485.2005.00006.x
1 The Phenomenon of
1 The Phenomenon of Reference
Reference
In a paper evaluating animal communication systems, Hockett and Altmann (1968: 63Ï4) presented a
list of what they found to be the distinctive characteristics which, collectively, define what it is to be a
human language. Among the characteristics is the phenomenon of Ñaboutness,Ò that is, in using a
human language we talk about things that are external to ourselves. This not only includes things that
we find in our immediate environment, but also things that are displaced in time and space. For
example, at this moment I can just as easily talk about Tahiti or the planet Pluto, neither of which are
in my immediate environment nor ever have been, as I can about this telephone before me or the
computer I am using at this moment. Temporal displacement is similar: it would seem I can as easily
talk about Abraham Lincoln or Julius Caesar, neither a contemporary of mine, as I can of former
president Bill Clinton, or my good friend John, who are contemporaries of mine. This notion of
aboutness is, intuitively, lacking in some contrasting instances. For example, it is easy to think that
animal communication systems lack this characteristic - that the mating call of the male cardinal may
be caused by a certain biological urge, and may serve as a signal that attracts mates, but the call itself
is (putatively) not about either of those things. Or, consider an example from human behavior. I hit
my thumb with a hammer while attempting to drive in a nail. I say, ÑOuch!Ò In so doing I am saying
this because of the pain, and I am communicating to anyone within earshot that I am in pain, but the
word ouch itself is not about the pain I feel. If, on the other hand, I say, with unnatural calmness,
ÑPain is present in my thumb,Ò then I am in this instance talking about pain.
Such intuitions have, for the most part, been extremely compelling, in fact so compelling that the
CORRESPONDENCE
THEORY
OF
MEANING
has, since classical times, in one form or another, been by far the
most persistently pursued notion of how meaning in language is best characterized. Not to put too
fine an edge on it, this is quite simply the idea that the significance or import of natural language
utterances is found in the ways in which they correspond to facts and things in the world around us.
In present times, this finds its clearest articulation in the framework of model-theoretic semantics.
Yet not everyone finds these basic intuitions of aboutness quite so compelling as to base a theory of
natural language meaning upon them. Most notably in the twentieth century, Wittgenstein is generally
interpreted as articulating quite a different view of natural language meaning which, at best, treats
ÑaboutnessÒ as derivative or epiphenomenal (Wittgenstein 1953). Also, Chomsky (1981, 1992, 1995a),
Hornstein (1984), Ludlow (2003) and others have articulated a similarly skeptical view about its
centrality. Since this chapter is about (the notion of) reference, I set aside consideration of such
alternatives and focus exclusively on work which does find this initial intuition most compelling.
The word about(ness) itself, however, is a folk notion that is too general and vague to really get at
something fundamental about natural language. We may ask, quite sensibly, what is Beethoven's
Third Symphony about, what is the relationship of a couple really about, what is a painting by
28.12.2007
4. Reference : The Handbook of Pragmatics : Blackwell Reference Online
Page 2 of 16
Mondrian about, or what was the First World War all about, anyway? Even if we confine ourselves to
linguistic utterances, we find ourselves with a slippery notion that is subject to all sorts of doubt and
uncertainty. In saying to a person on the street ÑMy garden is poor this year,Ò I could very sensibly be
talking about the cool weather, the lack of rain, the presence of pests, or a decision I made some time
ago to plant a certain variety of tomatoes. I could be talking about any of these things, and more.
However, the one thing that is clear that I am talking about in this instance that seems inescapable is
quite simply that I am, in fact, talking about my garden. This is, obviously, because in uttering the
sentence, I use the phrase my garden, whereas in this instance there is no particular mention of rain,
weather, pests, or poor plant selection. To distinguish these two types of aboutness, the term
REFERENCE
is going to be used for those things overtly mentioned in the utterance of a sentence. Thus,
I may be talking about the dry weather, but I am referring to my garden (and the current year as well).
This is helpful, in that it localizes and objectifies a certain type of aboutness in a reasonably clear and
intuitive way. Yet, even here there is all manner of cause for question and uncertainty. For example, in
the utterance above, might I also be referring to myself (by using my), gardens in general (by using
garden), the quality of being poor, and so forth? Intuitively, these questions have sensible answers
both yes and no. But it does remain a very solid intuition that I am referring to my garden, where an
intuitively based denial would seem far less convincing. For this reason, the focus of a theory of
reference has been on those elements of a sentence or utterance which most clearly display the
intuitive phenomenon of reference, leaving aside the subsequent questions for resolution within a
more precisely articulated theory. The types of words and phrases that canonically display reference
(see Strawson 1950) include demonstrative and indexical words and phrases (e.g. this table, that cat,
I, this), proper names (Aristotle, Paris, Fred Smith), and singular definite terms (the woman standing
by the table, my garden, the author of ÑThe RepublicÒ). Phrases and words of these types, not only in
English but where they appear in any other natural language, unequivocally Ñpick outÒ some particular,
definite individual or object. The point is, if these things don't exhibit the phenomenon of Ñreference,Ò
then we should all close up shop on this particular topic and find something else to work on.
2 Semantic Reference
Semantic Reference
2.1
2.1 Frege
2.1
Frege
Reference, then, is a kind of verbal Ñpointing toÒ or Ñpicking outÒ of a certain object or individual that
one wishes to say something about. But what, then, is the connection between the meanings of the
particular words of the language we use in order to accomplish this, and what is picked out as a
consequence? In order to frame this question, let us consider what has been typically called the
NAéVE
THEORY
OF
REFERENCE
. This was by no means first articulated by Frege (1879) (one immediate precursor
was Mill 1843), but Frege seems to have taken the idea and pursued it the furthest within a new
conception of how to do things - using the tools of formal logic - that appears to have been a
genuinely novel development on the intellectual scene. The basic notion is that the meaning of an
entire (declarative) sentence of a natural language is intimately connected to its truth value, and the
contributions of the words and phrases within a sentence to the meaning of the whole are determined
by the contribution they make to the truth value of the whole. Or, as McGinn (1981) puts it:
ÑReference is what relates words to the world of objects on whose condition truth hinges.Ò
If one then turns specifically to intuitively referential phrases and words and calculates the
contribution they make to the truth value of the whole, one encounters an initially surprising result:
that the truth value of sentences containing referential phrases is (in part) determined by what the
phrases themselves refer to, and not by any other or further characteristics of the phrases themselves.
From an intuitive point of view, if I say (falsely) that Ringo Starr wrote the novel War and Peace then
the truth value of this sentence has not to do with any particular beliefs or conceptions I or anyone
else might have about the world, but rather what Ringo himself, that guy out there, has and perhaps
has not accomplished. Let
be the person Ringo Starr. It is as if I am saying something to the effect
that:
wrote War and Peace.
Slightly more technically, and the success of this is easy to overlook, any phrase that has the
reference will be automatically guaranteed to yield a sentence of the same truth value if placed in
the same syntactic location in the sentence as the phrase Ringo Starr. Thus, supposing that the
phrases the most famous drummer for the Beatles, Jimmy Smits's boyhood hero, and that man over
28.12.2007
4. Reference : The Handbook of Pragmatics : Blackwell Reference Online
Page 3 of 16
there have, on an occasion of use, the reference of Ringo Starr, their contribution to the meaning of
any sentence will be K and nothing more. This will mean that all the following sentences are likewise
guaranteed to be of the same truth value as ÑRingo Starr wrote War and PeaceÒ; as will, in fact, any
other way whatsoever of referring to the particular man
, for this is the contribution any such phrase
will make to the truth value of the whole:
(1)a. The most famous drummer for the Beatles wrote War and Peace.
b. Jimmy Smits's boyhood hero wrote War and Peace.
c. That man over there wrote War and Peace.
Thus we have an actual diagnostic for what is intended by the term reference, namely, preservation of
truth value. Consider the following to see how this might go. I wonder if the word someone is a
referring term, and on an occasion of use can be used to refer to Ringo. (This would seem intuitively
plausible under certain circumstances. Suppose, for instance, I host a birthday party for Jimmy Smits
and have invited Ringo as a surprise guest, and when Jimmy complains how the party is dragging I
might say presciently, ÑYes, but someone has yet to arrive!Ò) Now consider the contribution to the
truth value of ÑsomeoneÒ in the following:
(2) Someone wrote War and Peace.
The judgments here are not wholly secure, but most people who think about these things agree that
what has been said here is, in fact, true, whereas if it were referential and had the value , it would
have to be false. Assuming these intuitions hold up, then someone is not a referential phrase (though
see Fodor and Sag 1982 for a different point of view). It makes some other contribution to the
meaning of the whole.
One might, thus far, look upon this discussion as a rearticulation of
LEIBNIZ
'
S
LAW
of the
intersubstitutability of indiscernibles salva veritate. But there are some objections to this that have
been the source of continued inquiry to the present time, which Frege also tried to deal with, chiefly
in Frege (1892). One objection, that I will mention and put to the side, is that one must not use
examples where the use of a term is metalinguistic. Words and phrases function as names of
themselves occasionally in language. When so construed, they do not have reference to the ÑusualÒ
objects and individuals, but to different objects, e.g. the linguistic objects themselves. Thus (3a) is
not to be intersubstituted for (3b) (preserving truth):
(11)a. Ringo Starr is a stage-name.
b. The most famous drummer of the Beatles is a stage-name.
c. Richard Starkey's more famous alias is a stage-name.
However, any other phrase with the reference the name Ringo Starr will preserve truth value. Thus,
(3c), unlike (3b), will have the same truth value as (3a), having the same reference.
While it is not always a straightforward matter to determine metalinguistic usage (e.g. consider the
discussion of
METALINGUISTIC
NEGATION
, Horn 1989), this particular objection has had primarily
nuisance value in the development of a theory of reference. More telling is one type of intuitive
objection and another based on failures of intersubstitutability. The intuitive objection can be simply
illustrated thus. If the meaning of a word or phrase is its reference, and ÑRingo StarrÒ and Ñthe BeatlesÔ
most famous drummerÒ have the same reference, then they have the same meaning. This just plain is
not so: these phrases have obviously different meanings. This objection has clear force. The other
objection gets to the heart of the nave theory of reference: that phrases with the same reference are
not always intersubstitutable preserving truth value. This phenomenon has received a huge amount of
attention in the literature. One facet of this objection comes from the behavior of propositional
attitudes. The following pairs of sentence can easily diverge in truth value:
(4)a. James believes that Ringo Starr is a solo singer.
28.12.2007
4. Reference : The Handbook of Pragmatics : Blackwell Reference Online
Page 4 of 16
b. James believes that the BeatlesÔ most famous drummer is a solo singer.
Having followed sporadically the later stages of Ringo's career, and having no idea whatsoever of any
connection he might have had to the Beatles, James could well be described as having the first belief,
but not the second (he assumes any such drummer is a drummer and not a solo singer). The problem
is, if reference, under the nave theory, is all that contributes to truth value, then both sentences
could (crudely again) have the following contents:
(5)a. James believes that
is a solo singer.
b. James believes that
is a solo singer.
and so, being identical in content, have the same truth values. To object that there is, in fact, a
reading of these sentences which does have this consequence - the de re reading where, intuitively,
the speaker is the one taking responsibility for the contents of the referring phrases - does not
adequately address this point. There is a reading (perhaps the more natural one) where identity of
truth value is not the consequence, and on the purely nave theory of reference discussed here this
simply should not happen. This is traditionally called the de dicto reading, and if the theory thus far is
correct there should be no such phenomenon.
The other major type of consideration is that of the contents of identity sentences, which are
generally assumed to be successfully analyzed by the Ñ=Ò relation. Such sentences do not appear to
introduce operators giving rise to opaque or de dicto contexts, but nevertheless are a similar source
of puzzlement. If the contribution to the meaning is the reference of the noun phrases in the
following sentences, then both ought to have the same Ñcognitive valueÒ (a phrase that will be
somewhat clarified below).
(6)a. The BeatlesÔ most famous drummer is Ringo Starr.
b. Ringo Starr is Ringo Starr.
That is, both have the value:
(7)
=
But while the second is very obvious and can be known to be true a priori (assuming both instances of
ÑRingo StarrÒ are the same, see below for comments on this), the first seems to convey contingent
information that may actually come as news to some people. This is a genuinely different kind of
objection, because in fact Ñ=Ò preserves truth value given identical referents. Whichever way one finds
of referring to Ringo Starr, intersubstitution will in fact yield identical truth value.
Frege's proposed solution to these problems is well known and often written about, but is itself
problematic. The proposal is that words and phrases, besides having a reference, also have something
which, in English, is called a
SENSE
. This ÑsenseÒ of a word or phrase is what distinguishes otherwise
coreferential expressions. Ringo Starr and the BeatlesÔ most famous drummer may have identical
referents, but are distinguished by their senses. The sense contains the Ñmode of presentationÒ of a
referent; it is an objective, and not a subjective, thing but it is what we psychologically ÑgraspÒ in
understanding a word or phrase, and in so grasping enables us to find out the reference of the word
or phrase. However, in Frege's view, it is not the psychological grasping itself that actually determines
the reference, but rather the objective sense itself that is responsible for determining the reference.
Thus, reference is determined indirectly from expressions of a language (this includes mathematical
notation): a bit of language expresses a sense, which in turn determines a reference. This holds in the
case of proper names as well - the names Richard Starkey and Ringo Starr (or Hesperus and
Phosphorus, or Cicero and Tully, to revert to more traditional examples) have different senses
associated with them despite common reference.
Frege's solution to the problem of de dicto meanings appears, initially at least, to work but strikes
many people as unduly complex and counterintuitive (see, especially, Barwise and Perry 1983). In
28.12.2007
4. Reference : The Handbook of Pragmatics : Blackwell Reference Online
Page 5 of 16
certain syntactically definable contexts such as embedded clauses, a referring expression does not
have as its reference its ÑusualÒ one, but rather its sense. Thus, in the propositional attitude examples
such as those above, the reference of Ringo Starr and the BeatlesÔ most famous drummer is not the
ÑusualÒ , but rather the ÑcustomaryÒ senses of each (we will call them S
1
and S
2)
, which differ from
one another, and it is these senses which now contribute to the meaning of the whole. And since the
senses are different, one now has different propositions that can diverge in truth value:
(8)a. James believes that S
1
is a solo singer.
b. James believes that S
2
is a solo singer.
There are two somewhat odd consequences of this solution. If the sense of an expression is, in these
instances, its reference, and reference is determined by its sense, and since the reference determined
by the senses S
1
and S
2
is in contexts such as those in (6), then there has to be another sense (S
3
)
that will determine S
1
and still another (S
4
) that will determine S
2
as their references in examples such
as those in (8). But, unlike the customary senses, we have no clear intuitive grasp of what these might
be. Further, the claim is that referring phrases in de dicto contexts have as their meanings different
things from what they have in de re contexts. Complicating things still further is that if you have a de
dicto context embedded within another de dicto context (e.g. ÑJohn was surprised that the Queen of
England believed the BeatlesÔ drummer was Ringo StarrÒ), then in the most deeply embedded context,
the reference of a referring expression is no longer its customary sense, but rather the sense that
determines its customary sense, introducing a third-order sense that must determine that as its
reference. This works recursively, so that if there are n embedded contexts in a single sentence, in the
nth context there would have to be an nth+ 1 sense to determine the nth sense as its reference. (This
is not an incoherent proposal. Within the formal framework of Montague 1973, for example, what
correspond to such higher-level senses are recursively definable, though any sense beyond the
customary one is a constant function.)
The oddness is compounded somewhat by Frege's view of sentence meanings. In ordinary contexts,
such things do have a reference, which he takes to be a truth value, and a sense, which he takes to be
a proposition of a Ñthought.Ò In de dicto contexts, however, the same recursive piling up of senses
occurs as with referential phrases, so that an embedded sentence ends up meaning something
different from its unembedded counterpart, a doubly embedded sentence has still another meaning,
and so on. This strikes many people, again, as a bit strange.
When we return to the issue of identity sentences, which do not involve de dicto contexts, it is a little
hard to see how Frege's suggestions lead to a definite solution. For the phrases used have their
customary reference, so all true identity sentences express a proposition of the form a = a (where a is
some arbitrary referent), though within a belief context, for example, different senses will emerge to
distinguish the (higher-level) propositions created.
Making use of the discussions to be found in McDowell (1977) and Dummett (1975), this is what may
have been intended. From the point of view of one understanding an utterance, ÑgraspingÒ the sense,
which determines the reference, does not enable one to automatically grasp the reference itself. If this
were so, and we happened on Smith foully murdered, all we would need to do is to hear someone
utter the phrase (in a de re context) Smith's murderer and the identity of the murderer would be
automatically known to us; but, obviously, it is not. Likewise, we would (in at least one uncharitable
interpretation of Frege's framework) only have to understand a sentence in order to know its truth
value. To check on how many copies of an article we need to submit to a journal for publication, we'd
only need to hear someone go through a list ÑThe Journal of Modern Fregean Studies requires one
copy È two copies È,Ò etc., until we hit on the reference Ñtrue.Ò
But there has to be some kind of psychological connection between grasping a sense and determining
a reference (let's call this relation ÑfindingÒ a reference, incorporating Russell's notion of
ÑacquaintanceÒ as a Ñdirect cognitive relationÒ (Russell 1910Ï11). Consider, for instance, your
understanding of the phrase ÑMy sister's oldest daughter.Ò If you can read this paper then you clearly
understand what this means - you ÑgraspÒ its sense - but it is very doubtful you are antecedently
familiar with that particular person. That is, the reference is unknown to you. However, if you wanted
28.12.2007
[ Pobierz całość w formacie PDF ]

zanotowane.pl

doc.pisz.pl

pdf.pisz.pl

anio102.xlx.pl