WORDS

David Kaplan
Preface. In the Summer of 1984 I was playing with some
techniques for indicating scope, hoping to find a new way into
the vexing problem of substitution on directly referential
expressions in belief contexts. After chasing Woozles for a week
or so I realized that insofar as the problem of substitution in
belief contexts was concerned, I had ended up right where I had
begun, but under a clutter of technology. At that time I wrote in
large letters on a lined yellow pad, "A NEW IDEA IS
NEEDED".
It was syntactical technology that I had been thinking about,
and it occurred to me that perhaps syntax, in some vague sense,
was a key to the puzzles that I had been unable to solve. I
ventured that thought, as it applies to single words, in
'Afterthoughts':
Lately, I have been thinking that it may be a mistake to
follow Frege in trying to account for differencesin cognitive
values strictly in terms of semanticvalues. Can distinctions
in cognitive value be made in termsof the message without
taking account of the medium? Or does the medium play a
central role? On my view, the message-the content-of a
proper name isjust the referent.But the mediumis the name
itself. There are linguisticdifferencesbetween "Hesperus"
and "Phosphorus"even if there are no semanticdifferences.
Note also that the syntactic properties of "Hesperus" and
"Phosphorus", for example their distinctness as words,are
surercomponentsof cognition than any purportedsemantic
values, whether objectual or descriptional.-
Could it be that the elusive cognitive difference between
believing that Hesperus is Hesperus and believing that Hesperus
is Phosperus rests on nothing more than syntax?
In speaking of syntactical,or syntactical-lexical,or what, in

@ David Kaplan 1990

'Pages 598-599 in 'Afterthoughts' in Almog, Perry, and Wettstein (editors), Themes
FromKaplan(Oxford University Press 1989).

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
94 I-DAVID KAPLAN

Demonstratives, differences, I am searching

I called presemantical'2
for a term to capture the elements of form that are independent
of semantics.3 The awkward, syntactical-lexical form, may best
convey my meaning.4
The question of whether two sentences have the same
syntactical-lexical form may seem utterly trivial, but it isn't. It is
part of the purpose of this essay to interest you in that question.
For example, Saul Kripke's intriguing puzzle5about Peter-the
man who first heard Paderewski'sname used in connection with
his musical accomplishments and later heard the name used in
connection with Paderewski's political accomplishments and
then concluded that these were two different 'Paderewski's-
seems to me to involve this sort of syntactical-lexical form issue.
My speculations led me to conclude that I had to go back to
basics and rethink notjust the semanticsofnames, but their very
syntax, the metaphysics of words: How should words be
individuated? What is the nature of a word?
These musings eventuated in my Gilbert Ryle lectures,
Word and Belief delivered at beautiful Trent University in
Peterborough, Canada, and at various other institutions. The
content of the present paper is drawn from the middle lecture.
The whole WordandBeliefprojectis quite speculative. This is the
part in which I have the most confidence.
In the first lecture ("A Puzzle About Direct Reference"), I
take up the problem posed to direct reference theory by
substitution on names within the context of a propositional
attitude report. The problem is that according to direct
reference theory two names that name the same thing have the
same semantic value, and therefore, substituting one for
another, even within a belief-context, should not affect the
semantic value of the sentence. After arguing for years,

'Page 559 in Demonstratives op. cit.

3Linguistsseem to think that "JohnadmiresJohn" and "John admiresJane" have the
same syntacticalformand differonly in what they call "lexicalization" (though they claim
that "John admires himself" differs from both syntactically). Logicians seem to think
that "John walks" and "Jane runs" have the same logicalform (except perhaps when
each is conjoined with "John does not walk").
'When I use the word "syntax" or "syntactical" (sometimes in scare quotes) or
form notion that I have in mind.
"logical syntax" it is usually this syntactical-lexical
'In his 'A Puzzle About Belief' in A. Margalit (editor), Meaningand Use (D. Reidel
1979)

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 95

unconvincingly, that semantic value (properly understood) is

not affected by substitution, I hit upon a brilliant, new, and
completely successful, strategy: argue, instead, that semantic
value is affected by substitution. But I also argue that, contrary
to my own long-standing misapprehension, this result is not
contrary to direct reference theory.
The detailed argument involves an attack on a version of
Frege's principle of compositionality, focussing again on the
syntacticaldifferences,i.e. differencesas syntactical
objects,between
sentences of the form "a = a" and sentences of the form "a = b".6
This raises the issue:What determines an utterance to have the
form "a =a" as opposed to the form "a = b"?And this issue takes
us directly to matters of word individuation.
I am convinced that we can achieve a highly salutary clearing
of the air about the nature of language, and especially about
some critical differencesbetween naturallanguagesand logicians'
idealizations, if we study the ontology and individuation of
words.
The present paper is organised into three main parts. In the
first part, What is a Worda,I contrast two theories of the
individuation (one might say, metaphysics) of words: the con-
ventional token/type theory, which I call the orthographic
conception, and an alternative theory, which I call the common
currencyconception. I will try slowly to entice you into
abandoning the traditional favorite in favor of the new
conception.7
The second part focuses on names, distinguishes such words
from their genera, and combats certain wrong construalsof my
views. I also claim to show in this part that there can be cases of
distinct words which are both homophones and homographs-I
call them phonographs, they sound the same and they are

61 have cometo thinkthattwosentenceswhosesyntax--perhapshereI shouldsay,

whose logicalsyntax-differsas much as "a= a" differsfrom"a= b" shouldneverbe
regardedas having the same semanticvalue (expressingthe same proposition),
regardless of the semanticvaluesof the individuallexicalitems"a" and "b".
' In the
philosophicaltradition,the main idea of the commoncurrencyconception
maybe thoughtto be implicitin theworkofKripke,Donnellan,andperhapsmyselfand
others,but in workthatwasdoneforsemanticalpurposes,not forsyntacticalpurposes.
In my view,theseparticularideasof recenttheoriesof referencehavesimplynot been
sufficientlyexploited.I hopethatthosewhowishtoencouragea viewoftheobjectivityof
languagewill welcomea usefulally in my analysisof words.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
96 I-DAVID KAPLAN

spelled the same-distinct words that are phonographs and

whose semantic values are also exactly the same. The semantics
is the same, the spelling is the same, the pronunciation is the
same, but they are two different words! I will try to show that
this odd result is perfectly natural on a correcttheory of word
individuation.

WHAT IS A WORD?

1.1 Expressionsand their Occurrences. The question of how to

individuate linguistic expressions is a delicate one. Here is a
famous sentence from the section 'Use Versus Mention' in
Quine's MathematicalLogic-one of our sacred texts. Quine is
talking about expressions and how we refer to them.
To mention Boston we use "Boston" or a synonym, and to
mention "Boston" we use ""Boston" " or a synonym.
"'"Boston"" contains six letters and just one pair of
quotation marks; "Boston" contains six letters and no
quotation marks; and Boston contains some 900,000
people.
I would have thought that a logician like Quine, who is used
to distinguishing variables from their occurrences, would have
immediately seen that "Boston" contains six letters is false. I
only count five letters in the name "Boston": a "B", an "o", an
"s", a "t" and an "n". There are, of course, two occurrencesof
the letter "o".
When I utter "Help!", no one thinks that my utterance is the
word "help". That would give the language too many words.
When I utter "Help! Help!", I haven't uttered two words, I've
uttered one word twice. Two utterances, one word. Similarly,
when I write "Quiet", my inscription isn't identical with the
word, rather it is an inscription of the word. A single word can,
and typically will, have many utterances and inscriptions.
Uttering and inscribing, or writing, are actions whereby we
produce certain concrete, as opposed to abstract, physical
objects: utterances (sounds), and inscriptions.These objects, the

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 97

utterances and inscriptions,are the physical media by which we

transmit words from one to another.8
Transmission.
1.2 Interpersonal Every word that we know is either
one we invented or one that was transmittedto us in one of these
ways, by way of an inscription or an utterance. (I haven't
resolved how to think about the interpersonal transmission of
words through electronic media and through the use of Morse
code and so on. Let's not worry about that.) The central claim
here is this, that in interpersonal linguistic communication,
interpersonal-I'm very focused here on the distinction between
intrapersonal links, and interpersonal links-in interpersonal
linguistic communication, words must take on a physical
embodiment.
1.3 The Orthographic Conception. Now, what are these words that
we transmit by means of their utterancesand inscriptions?Well,
the prevailing view, especially among those trained in the
traditions of logic (as I am), is that words are the typesof which
utterances and inscriptionsare tokens.This, I now think, is quite
wrong. And misleading, even as a model.
It seems to me in many ways that this is a sort of updated

8I1usedto thinkthat sometimescalledtokens

inscriptions, by logicians,werephysical
of thekinds:pilesof ink,neontubes,piecesofmetal,and
objects,possiblydiscontinuous,
so on. Here is an exampleof sucha discontinuous physicalobject.
NO
Then I realizedthat thereis anotherway in whichwe can forman inscriptionof that
word.You do it like this.
NO
Now here again I makeuse of anotherdiscontinuousphysicalobject(discontinuous
becauseof thepieceremainingfromthemiddleofthe"0") inorderto producea second
inscriptionof the word"NO",but noticethatthe inscriptionis not the discontinuous
physicalobjectwhichconstitutesthe stencil,it's the space.
I firstrealizedthiswhenI thoughtaboutinscribingwordsinstone,andI realizedthat
youarecreatinga physicalobjectwhenyouinscribea wordin stone,butthetokenofthe
wordis not the greatbig heavyphysicalobject,the physicalobjectwhichis the tokenof
thewordis the light-weightspace.Thismustbe the notionofspacein thesenseinwhich
architectsand sculptorscreatespaceby enclosingthemwith largerphysicalobjects.
Note that the enclosuredoesn'thaveto be squareforthe spaceto be square.You can
havea squarespaceinsidea sphere.Andnoticealsothatthisisthekindofspacethatcan
be moved around,from place to place. The space is intimatelyconnectedwith its
enclosurebutnotidenticalwithit.Youcouldknockthecornersofftheenclosurewithout
affectingthe space.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
98 I-DAVID KAPLAN

version of the Platonic notion of abstract forms. The eternal,

unchanging Platonic forms (shapes, perhaps) are the types, and
their physical embodiments, which reflectthese abstract forms,
are the tokens. I think that the token/type model is the wrong
model for the occurrence/word distinction (i.e. the utterance/
word distinction or the inscription/word distinction). The
token/type model best fits what I call the orthographic conception
of a word, the typesetter's conception.
According to this conception, expressions of the language
consist of strings of atoms called 'letters', certain strings form
words. The letters are abstract entities whose tokens, for the
typesetter, are individual pieces of type. (It's strange that on this
conception, type is a token, but that does seem to be the result.)
You surely know this conception very well. We were all taught
it when we studied the syntax of formal languages (not to
mention when we took print shop in high school). It belongs
to the formal subject, formal syntax. And its study is the study of
an algebra.

1.4 The Common CurrencyConception. The TokenlTypeModelversus

the Stage/Continuant Model. I propose a quite different model
according to which utterances and inscriptions are stages of
words, which are the continuants made up of these interpersonal
stages along with some more mysteriousintrapersonalstages. I
want us to give up the token/type model in favor of a
stage/continuant model. This is not, I think, simply another
way of doing the metaphysics of types under the old token/type
conception, but a quite differentconception of the fundamental
elements of language. I think of my conception of a word as a
naturalistic
conception. Becausethe interpersonaltransmissionof
words is so central to my conception, I adopt a phrase of
Kripke's, and I call my notion the Common conception of
Currency
a word.
Here's just one way in whch my conception differsfrom the
token/type conception. On my conception, there is a single
word "color" spelled one way in Canada: "c", "o", "1", "o",
"u", "r" and another way in the United States, "c", "o", "1",
"0", "r". Similarly, there is a single word which is pronounced
shedge-yulein Canada, and skedge-oo-ul in the United States. (I
believe that the English have also adopted the Canadian

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 99

pronunciation.) We are, of course, familiar with dialectal

variation in pronunciation, but I call to your attention that
there is also dialectal variation in spelling. This dialectal
variation in spelling of the word "color" for example does not,
repeat not, make for different words.
1.5 Dialectand Idiolectin Sayingand Writing.I should point out
that there are also idiolectal variations in pronunciation, as for
example in the case of a speech defect. Some of us also have
idiolectal variations in spelling. Society treats these two kinds of
deviations-in speaking and in writing-in extremely different
ways. We are very liberal nowadays about variations in
pronunciation, at least in the United States. In England,
although there are many and more disparatedialectal variations
in speech, there seemsmore acceptance of the idea that there is a
correct way to pronounce things. In the United States, even the
national networks have anchormen whose pronunciation is
quite different. Dan Rather of CBS speaks with a noticeable
Texas, or Southwestern, accent-a regional dialect. Tom
Brokow of NBC says mi-we-unwhere I would say million-an
uncommon, but not otherwise unheard of, idiolect. And Peter
Jennings of ABC says aboutfor "about"-a Canadian dialect, I
believe. These variants are tolerated.
We do not tolerate, however, idiolectal variations in spelling.
Now it's been suggestedto me that the variationsin pronunciation
we tolerate is regional, or group, variance-i.e. dialectal
variance-and if there were a regional or group variance in
spelling, that would be tolerated also. I don't believe that for a
single minute. We tolerate Brokow's idiolect and all sorts of
other strange and unique speech styles. I'm quite confident that
children from the South, as they start learning to read and write,
have a strong tendency to spell the two words "you all" "y",
'"a"',"1",'"1 (that'ssimply making a natural transcription).But
that this charming spelling dialect is quickly suppressedby the
forces of spelling bigotry and intolerance.
It must be obvious to you that I'm a person who has
suffered-and who continues to suffer-serious discrimination
regardingmy idiolectal spelling. And I wish to point out that in,
say, the seventeenth century, people were much more liberal
about such things than we are today.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
100 I-DAVIDKAPLAN

1.6 How wordschange:PlatonicShadowsversusEvolvingNature. I

make these points to dispel the lurking thought that a word has
some fixed and perfect Platonic form, represented say by the
way it is spelled, which is then embodied in the imperfect,
variable, and changing ways in which people pronounce it.9
Historically of course the matter was quite the opposite. Speech
is prior and writing is a transcription of speech. So there is no
metaphysically fixed form in either speech or spelling, no matter
what social programs are mounted for standardization. There
are spelling variations, there are pronunciation variations, there
are all kinds of variations that take place over time.
On the other hand, on my conception, there are two
phonographic words "base" (meaning 'low'and bottom),not, as
the orthographic conception would have it, a single word with
two meanings. (And, as we will see, I do not simply define wordas
an orthographic word combined with a meaning.)
1.7 CreationistandDevelopmental Linguistics.I will want, eventually
(but not in the present paper), connect these speculations with
to
the questions of whether the Babylonians believed that Hesperus
is Phosphorus and whether Peter believes that Paderewskiis not
Paderewski. So let us consider the name "Hesperus". I have a
story about how the word "Hesperus" came to us. The story is
surely incorrect (after all, the Babylonians didn't speak Greek),
but it gives the flavor of my views. I imagine that at some point
some Babylonian looked up in the sky one evening and said (in
Babylonian) "Oh, there's a beauty. Let's call it-", and then he
introduced the name. What he did was to create a word. He
created a word as a name, a tag, for this beautiful heavenly
body. He then passed that word on to other people through
inscriptions and utterances. Those people passed it on to others
and so on (I'm going to talk more about this passage). As it went
through different communities, the way this word was pro-

recently read a book about Darwin in which it was said that an enormousamount of
'I91
evidence for evolutionary ideas was already available when Darwin was working. One of
the things that stopped people from putting it all together was this Platonic idea of the
fixed, eternal, and unchanging form setting the limits of variation for the shadowy
objects of the sensible world. They were carried away from the thought that there could
be natural kinds which went through the dramatic transmutations claimed by
evolutionary theory. As I said earlier, I think of my conception as being naturalistic, as
owing more to the theory of evolution than to algebra.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 101

nounced and written changed in very dramatic ways, through

whatever processes account for dialectal variation.
The presupposition of these processes of change are the
principles of continuity in accordance with which a changing
word retains its identity. As wepassthroughvariouscommunities
at various stages in our lives, we also change dramatically. (I
think I am probably more different now from the way I was
when I was, say, eight yours old, than the word "Hesperus" is
now from the way it was when it was, say, five minutes old.) But
we still have the notion that we are a single entity. And so it is for
the word. Changes in pronunciation and spelling need not
suggest the notion of replacement of one word by another, which
then takes up the task in the manner of a relay race. Rather, we
can use the notion of a single entity undergoing change.

1.8 Intrapersonal Now, I've talked about the processby

Continuity.
which the word was transmitted interpersonally.What about the
process by which the word is transmitted intrapersonally?This is
the most difficult stage to understand, at least for me to
understand, because it is so deeply implicated in cognitive
psychology. Suppose a particular word is transmitted from you
to me. Now at some point I make a transmissionto someone else.
Question: Am I transmitting that very word?
Some word was transmitted to me by way of utterance or
inscription. I transmit some word by way of utterance or
inscription. We can phrase the question in this way: Take the
utterance or inscriptionreceived and the utteranceor inscription
transmitted. What makes it that the transmissionis an utterance
or inscription of the same word as that received? We can thus
turn the question into a question about the relationship between
input and output utterances or inscriptions.
I don't mind if you want to continue to call utterances and
inscriptions "tokens", although I'd prefer "utterance" or
"inscription", so long as we do not get caught up in the
metaphysics of the token/type model. Because it is beyond
doubt that the utterance or inscription transmitted couldbe an
utterance or inscription of the same word as that received,
although the diff•rence in phonographology, the difference in
sound or shape or spelling, can be just about as great as you
would like it to be.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
102 KAPLAN
I-DAVID

"Just about as great as you would like it to be", that's a strong

statement. I will try to defend it.
Let me restate the problem using the technical apparatusand
terminology of cognitive science.

Black box

output - Input

Something goes on between the reception of an utterance as

an input and the transmission of a distinct utterance as an
output. What happens in the black box during this intrapersonal
processing, what is it that connects particular input and output
circuits? What is it that makes a particular output, the
transmission of the same word as that carried by a particular
earlier input? I can't provide a detailed answer to this question,
but I can give you some examples to show you how great the
differences in sound, shape, or spelling can be in cases in which,
unless we are completely dominated by the token/type model,
we would agree that the word being transmitted is the same
word as that received.
Consider this thought experiment: I say the name of an
individual, possibly a name known to the person to whom I am
speaking. The subject is to wait for a count of five, and then
repeat the name. I say a name, then the subject says the name. I
say the next name, then the subject says the next name. So, ifI
say "Rudolf', the person says "Rudolf"; "Alonzo"--"Alonzo";
"Bertrand"--"Bertrand", and so on. Because we have to worry
that the subject might be, in Kripke's term, reticent, if he
succeedsin repeatingthe name, we rewardhim with a dollar,or, if
he has tenure already, a thousand dollars, enough at any rate to
motivate him. I think that if we set up the story in this way-the
subject is highly motivated, he is sincere, he is not reticent, he is

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 103

reflective-whatever that means, then we are very strongly

inclined to say that when this person speaks, he is repeating the
very name that he heard. I'm not saying what's happening
inside the black box, I'm not saying how he does it, I'm just
saying that from the description of the case it's clear that we
would agree to describe his output as a repetition of thatname.
This notion of repetition is central to my conception.
There is a physical transmissionof my output utterance,which
is the subject's input utterance, to the subject's black box, then
there is a psycho-physical (or better, a physico-psychological)
transition where those sound waves hit the ear and something
goes up and is put into what we call memory, and then, after five
seconds, it (that something) is called out of memory, goes
through a psycho-physical transitionin the vicinity of the mouth
and throat and the output utterance appears. (I am purposefully
vague as to whether the psychopart takes place in the mind or in
the nervous system, though I lean to the answer "both".)
The two psycho-physical transitions, at the ear and at
the mouth, significantly affect the subject'sability to match the
physical propertiesof the sound waves that go out (his output) to
the physical properties of the sound waves that come in (his
input). (I am assuming here that sound waves with 'matched'
physical propertieswill 'sound' the same to each single auditor.)
For example, on the output side, I may say the names in my deep
and profoundvoice, and the subject may not have the apparatus
to produce a deep and profound voice. He has a high squeaky
voice and that's the best that he can do. If you happen to know
someone, as I do, who has a certain kind of congenital hearing
defect, you know that although his or her productivemechanism
may be perfectly normal, the problem at the input side produces
noticeable and characteristic kinds of changes from the sound
pattern that come in to the sound pattern that go,.s out. (At
least, as I hear those sound patterns.)
So individual differences in the physiological processing at
these psycho-physical transition points may make it that what
comes out is not going to resemble what went in. However, and
this is what is important for this thought experiment, the exact
functioning of the psycho-physical transition mechanisms is
irrelevant to our characterization, our intentional character-
ization, of the case as one of 'repeatingthe word he heard'. He

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
104 I-DAVID KAPLAN

may not do it well, from the external point ofview. But it is what
he is doing. We don't question that the connection has been
made in the black box. No matter how poor the subject's
imitativeability (his ability to make his output resemble his
input), we can imagine circumstances in which we would say,
"Yes, he is repeating that name; he is saying it in the best way
that he can."
Contrast this with a wealthy mischievous subject who has
decided that he's going to play a trick on us and instead of
repeating the names as he hears them, ignores the input and just
utters names at random (or, he may have prepared his own list
ahead of time which he recites in order). Even if, by
happenstance, the sounds that come out in these two cases
equally resemble the sounds that went in, the firstcase is a case of
repetition and the second is not.
The identification of a word uttered or inscribed with one
heard or read is not a matter of resemblance between the two
physical embodiments (the two utterances, the two inscriptions,
or the one utterance and one inscription"').Rather it is a matter
of intrapersonal continuity, a matter of intention: Was it
repetition?We depend heavily on resemblancebetween utterances
and inscriptions [using resemblance here not to mean matching of
physical characteristics but of their appearance as we look and
listen] in order to divine these critical intentions. If it sounds like
it
"duck", probably is "duck". But we also take account of
accent and idiolect and all the usual clues to intention. It is the
latter that decides the matter.
In fact, as Philip Brickerhas pointed out, when we repeatwhat
someone has said, we don't aim to imitate the pronunciation, we
aim to standardize it (by our own standards). Imagine asking a
third party to repeat what a speaker with an unintelligible
accent has said. Would he imitatethe speaker?
There is a story about a missionary who mispronounced
certain verbs of the local language, with the result that his
sermon sounded like an exhortation to highly questionable
behavior. All this, much to the amusement of the natives. But
what was the missionary saying?Is what we say really so subject
10One of the eventsthatsetme thinkingalongtheselineswasa talkby ArthurDanto
in whichhe asked,as I recallit: In whatway do an utteranceand an inscriptionof a
singlewordresembleone another?

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 105

to the vicissitudes of ear and tongue, or to the dialect of local

custom? I would claim that a mispronunciationof a word is an
utterance (a pronunciation, if you will) of thatword. It seems to me
that cases in which one utters a word other than the one
intended must be exotic indeed."
There are, of course, familiarways in which one can fail to say
the word one intends to say, for example by intending to say the
magic word that opens the cave door, but forgetting exactly
which word that is. Or intending to say the name of an old
acquaintance at a party, but drawing a blank. (This last is a
looming fact of life for many of us.) Here the word is given under
a concept, as Frege would say. This is quite a different thing
from mispronunciation, even startling, inexplicable mispro-
nunciation.
In view of the fact that individual differencesin physiological
processing at the psycho-physical transition points may affect
the resemblance of output to input, we can imagine somehow
getting into the transition mechanisms of the subject and
putting filters of various kinds on them so that we get--and this
was the claim that I made earlier-we get differences in sound
just aboutas greatas wewouldlikebetween what comes in and what
goes out. If, however, we were convinced that the source of the
change was as described-due to the filters at the psycho-
physical transition points-and that the cognitive link that has
to take place inside the black box was in order, we would still say
it's the same word. It'sjust that the subject can't come anywhere
close to giving the word a standard pronunciation.'2

" The exotic cases I have in mind are those in which to theastonishment the
of speakerthe
wrong word came out. I have been witness to, and subject to, this experience on several
occasions in connections with proper names. "Wait a minute", the speakersays, "Did I
just say 'Eleanor'? I meant 'Harriett'." Some dark force has reached into the speaker's
psyche and misdirected the hand of intention.
1 We are familiar with
everyday occurrences of this in connection with people who
have different native languages and who have a differentrepertoireofsounds. One of my
Japanese friends, who spoke unaccented Californian, was trying to explain to me how to
say two of my favourite words, one of which is "Netsuke" and the other is "Hokusai".
There is a "u", as we write it in English, in both of those words which doesn't exactly
disappear, and isn't exactly sounded. He kept saying "You are saying this [and hewould
imitate my pronunciation];you should be saying this [and he would pronounce the word
'correctly']". I couldn't hear the difference between his imitation and the 'correct'
pronunciation. Conversely, as we know, some of our Japanese friends have great
difficulty with the R-L distinction, a distinction that we easily make.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
106 I-DAVIDKAPLAN

1.9 The Criterionof WordIdentityis not Resemblance. My chief

point is this. There is a word taken in; there is a word sent out.
There is a mystery about what has to go on inside the black box
in order for it to be the same word, a mystery about how to
analyze in detail what must go on in order for it to be the same
word. (I'm imagining an analysis in the language of psychology,
not brain physiology; an analysis in programming, software
terms, not in electronic circuitry, hardware terms.) But the first
thing that we should get out of ourheads is the idea that we can
tell whether the input and the output are utterancesof the same
word by looking at (or listening to) the physical object that
comes out, and looking at (or listening to) the physical object
that goes in, and trying to make a phonographic comparison of
the two to see whether they are similar enough in some specified
way.
1.10 ContinuitythroughMemoryis Hardto Trace.When the word
goes through the black box, when the word is received fromone
person and stored for passage on to the next person, it isn't, of
course, put into the pocket in the way in which a coin can be
stored in its passage from person to person. The coin is put into
the pocket and there it is located. There is a definite answer,
whether we know what it is or not, as to whether the lucky coin
that your coach gave you is really the very one that his coach
gave him, or whether it is a different coin that looks pretty
much like the lucky coin (in fact, is indiscerniblefrom it, in the
way of modern artifacts).
In the case of the word, we feel that the comparable question
doesn't have the same very straightforwardanswer, because it
isn't put into the pocket, it is put into memory. Remember, in
our experiment we said to the subject "I'm going to say a name.
Then waitfor a countoffive and repeat the name". This form of
storage, in the mind (rather than in the pocket), makes the
continuity much harder to trace.
1.11 ShortCircuitsandGaps.On my theory of words there are two
special problems having to do with the way in which the words
are stored in memory that make for particular difficulties.Each
problem corresponds to a kind of error regarding word
individuation to which even a competent language user is
susceptible. The first kind of error is this. A transmission of a

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 107

word may come into the black box and then another
transmissionof the same word may come into the black box, and
inside the black box this very same word might get stored in two
different locations, stored,so to speak, as the transmissionsof two
different words. If I say "tomato" and you say "tomahto", a
naive subject may take us to be speaking of differentvegetables.
It really isjust one word. There have been two receptions of the
same word, but it is not storedas a single word. It is stored as two
different words.
Somehow, in the black box, the differentbranchesof the same
word (i.e. the different input utterances of the same word)
weren't all properly linked together. And that means that when
the black box emits, it thinks it has two words, and it will make a
choice as to which'word' it is going to emit.
The converse phenomenon also occurs. Consider a speaker
with three friends, "Mary", "Merry", and "Marry" as he calls
them (the last might be a nickname for Zsa Zsa Gabor). Now in
some American dialects these three words are pronounced
indistinguishably;they are homophones. From a psychological
point of view, from the point of view, as it were, of the black box
of the speaker, three quite differentwords are being uttered. He
may even think he is pronouncing them quite differently,he has
differentspelling for them, he puts his tongue in a differentplace
when he speaks. But from the point of view of the listener, what
comes out is at bestthree homophones.
To see how easily this can occur, consider a case in which no
error is made. Let's take an example of someone who has two
friends, two distinct friends, named "John" and "Jon". Then
the person can choose to say "Look, it's tautological thatJohn is
no taller thanJohn. And also thatJon is no taller thanJon. But it
is not the case that Jon is no taller than John." What has he
done? He has these two names, they're homophones, and he
makes two choices of the first name and produces a sentence of
the form "a R a". He then makes two choices of the second
one and produces a sentence of the form, shall we say, "b R b".
And then he makes one choice of the one and one choice of the
other, and he producesa sentence of the form"a R b", in terms of
the very words that were used. (Suppose the relation R were
identity.)It is important to recognize that having achieved this
insight, we can kick away the spelling ladder.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
108 I-DAVID KAPLAN

This situation canarise. If we do kick away the spelling ladder

and spell both names the same way, we have two, common
currency, phonographic names for two different people. One of
these names was created (i.e. introduced as thatname, a name of
that person), say, by the first friend's mother, and the other one
was created, say, by the second friend's grandfather. The two
names have had rather differentlife histories,but they met when
both were stored in the black box of the common friend.Within
the black box they were correctly stored as being two different
words, and then when the black box emits, it chooses which
word it is going to emit, which 'circuit' it is going to continue."•
But note that the very same phenomenon could easily occur
when there was only one name and the person made an error in
thinking that there were two different names, in thinking that
there were two differentcommon currency names. Imagine now
that there was only one common currency word, but that it was
stored in this way as if it were two. An error in word
individuation is being made by the person whom this black box
represents, an error, as I would call it, in syntactical-lexical
form. He is, of course, a perfectly competent speaker of the
language, a native speaker in fact. This errorthat he is making is
not really to be held against him, because it could happen to any
of us. There are so many people to be named and so few generic
names to go around. (Generic names are the genera, or species,
of our individual common currency names. More about this
later.) Errors of the first kind are taking one name to be two.
Peter is making this kind of error in Saul Kripke's
"Paderewski"case, mentioned in the preface. It is my belief that
the analysis in terms of word individuation is valuable, and
perhaps critical, in understanding that fascinating case and in
distinguishing it from the more familiar Hesperus-Phosphorus
cases in which there is no problem of word individuation. I don't
claim that all mistaken identity cases,. not even all that pose

13Again, an actual case:My mother'sprimarycare physicianis Dr. Shapiro.He

referredherto a specialist,another'Dr.Shapiro'as it happened.Mymotherreportedher
gratitudeto Dr. Shapirofor sendingher to Dr. Shapiroand comparedDr. Shapiro's
virtuesto thoseof Dr. Shapiroin a blithepiece of discourse,clearlyobliviousto the
homonomy.I was racingto keepup (whichI wasstrangelyable to do). Butfromher
pointof view,she wasquiteproperlyusingtwo differentwordsto referto twodifferent
people.Why shouldtherebe a problem?

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 109

problems for direct reference theory, resolve to mistakes about

the identity of words. But this one does.l"'
There is a second kind of errorto which we are all susceptible.
Errors of the second kind occur when there actually are two
persons with phonographic, common currency names (but still
distinct names, one created by the mother and the other by the
grandfather), and the two common currency names are linked
and stored in a single location (and it is from that single location
that all future utterances of"John" will arise). This is an errorof
the second kind: taking two names to be one.
The second kind of error is a short circuit-two different
circuits got wired together-zap, the whole thing goes up in
smoke. I am inclined to think that when two different common
currency words are wired together in this way in a given black
box, which then pulls from that common source and transmits,
nothing whatsoever is being said. Is it transmitting the first
word? Is it transmitting the second word? I think there isjust no
answer to that question. The two words have been co-mingled
in such a way that there is just no answer. Harking back to
earlier remarks about the difference between confusing the
identity of persons and confusing the identity of words, I note
that I am not claiming that in errors of the second kind we
cannot tell what the speaker means, or to whom the speaker is
referring.It is rather that, even if we could identify the referent,
we could not thereby infer the identity of the word.16
"4HowarewetothinkaboutKripke's famousLondon-Londre case?Shouldweregard
it asacaseofdialectal ittothePaderewski
thusassimilating
variance, case?Orshould
we
regard "London" and"Londre" words,andtreatitasa Hesperus-Phosphorus
asdistinct
case (or perhaps a Germany-Deutschland case)?
"sA confused identity case that does notseem to resolve to a mistakeabout the identity
of words is that described by Russell in 'On Denoting' to interpret the resultsof giving
"the author of Waverly" primaryscope in "George IV wished to know whetherScott was
the author of Waverly" (a case of quantification in). "This would be true" Russell says
"if George IV had seen Scott at a distance, and had asked, 'Is that Scott' ". George IV's
innocent confusion, expressed in "Is that Scott?" seems to me to involve no syntactical-
lexicalconfusions, as I would put it. At the moment, the case also seems to me to pose no
serious threat to direct reference theory. (Oddly enough, it does pose a serious threat to
Russell's own theory, since the primary scope interpretation allows us to prove the
sentence attributing an interest in the law of identity to the first gentleman of Europe.
Try it, using "Scott" as a genuine name.)
'"Keith Donnellan in section ix of 'Proper Names and Identifying Descriptions'
(Synthese21(1970): 335-58, reprinted in D. Davidson and G. Harman (editors)Semantics
of .NaturalLanguage(Humanities Press, 1972)) gives us the mistaken identity case of
J. L. Aston-Martin. A party-goer mistakenly takes another man at the party to be the

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
110 I-DAVIDKAPLAN

Errorsof this kind are quite differentfrom Paderewskicases of

the first kind in which there is a single word that is being
transmitted but the speaker makes the mistake of thinking it to
be two words.
II
WHAT ARE NAMES?
2.1 Namesas Words.I have spoken of words,though my examples
have often involved names.And truth to tell, it is namesat which I
aim. It is names that have been thought to challenge direct
reference theory.
Names are a special kind of word, so special that some have
thought them not to be a part of a language at all. I disagreewith
this and will emphasize ways in which names are like other
words, but I do not disagree that names are special in several
ways. Even if one were to conclude that names are so unlike
other words as not to be regarded as a part of any particular
language, this should not count against applying the earlier
principles of individuation to names.
All that I have said of the interpersonal transmission and
intrapersonal processing of common currency words seems to
me to carry over directly to names. Remember Paderewski,Jon
and John, and my mother's two Drs. Shapiro.
Furthermore, it seems clear to me that my processing of my
friends' names is correct.They are different common currency
words. Different people created their names and did it on
different occasions, so it seems fitting to say that they have
different names, that their names are different common
currency words.
Once, when I gave a talk on these matters, Paul Benacerraf
said to me, in connection with David Israel who was in the
audience at the time, "I thought that you and David had the
same name, now you tell me you have different names". John

famous philosopher J. L. Aston-Martin. In the party-goer's subsequent discourse,

Donnellan identifies the referentsof successiveuses of the name as variously:the man at
the party or the famous philosopher. His technique is to use the indices of intention to
identify the topic of discourse.No sorting of words occurs, as is obvious from the fact that
in the case, as described, only a single name is involved. It is interesting to note that
Donnellan's techniqueswould seeminglyworkjust as well in identifyingdistinct referents
even if one of the two referring expressions were anaptoric on the other.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 111

Perry put it somewhat more strongly afterwardssaying, "Your

views are insane because there's just this one name 'David',
which has been around since biblical time".
It might be thought that this observation counts against my
conception of words. Not at all. These ancient names to which
Perry alludes are naturalisticobjects.They live in the world, not
in Plato's heaven. They are cultural artifacts, created by us,
transmitted by us, stored by us. Surely, there is no argument
here for the token/type model.
At most what we have here is a disagreement about subtleties
of individuation. And perhaps not so serious a disagreement as
to be invulnerableto a conceptualdistinction.Let us acknowledge
this other conception of a name, which I call agenericname,just as
we acknowledged the orthographic conception of a name. The
notion of a generic name is useful for clearing up the apparent
disagreement over when two utterances are utterances of the
samename, and it may be useful for some practical purposes-
for spelling checkers and typographer, things like that. But for
serious semantics, I think that it is my common currency
conception that would be important.
There is the generic name "David", and then there is my
name "David", there is David Lewis' name "David", and so on.
These three are all distinct words. The latter two have-and
here I speak carefully-a semantic function: They name
someone. The first, the generic name doesn't name anyone
(doesn't name anyone,perhaps it names or is an unnatural kind).
Furthermore, it doesn't pretend to name anyone (as certain
empty common currency names do).
Generic names may be the closest thing in my theory to the
Platonic forms that word types were said to be by the token/type
theorists. (I personallypreferthe species/specimen analogy, and
don't forget the evolution of species.) But even if you wish to
think of generic names as types and my common currency
names as their 'tokens' (ugh!), two utterances of my name
"David" are utterances of the same'token' of the generic name,
whereas an utterance of my name and an utterance of David
Israel's name are utterancesofdifferent'tokens'.To put the point
another way, no matter how you slice it, the individuation of
common currency names, as I have described it, must be taken
into account.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
112 I-DAVIDKAPLAN

Thus, I encourage the idea that in addition to the common

currency names-these distinct, fairly recent names, David
Israel's and mine for example, which were created at different
points, and which have had distinct life histories-in addition to
the common currency names, there is also another kindof word
which we call a 'generic name'.
2.2 CausalChainsand GenericNames.The recognition of generic
names poses a delicate problem. When people think about the
picture of the transmissionof words that I have tried to outline
here, they call it the causalchaintheory, or something like that,
and they think that the question of intrapersonal continuity is
just a question of whether the utterances of the name that was
emitted was caused by the utterance of a name that was
received. But you see, there are very delicate issues as to what
goes on in the black box and the nature of the causation. Indeed,
when I was named "David", my parents had David Hume in
mind, and they so admired him that they thought, "Let's name
our son with the same name". So I was named in honorofDavid
Hume. My parents didn't just make up the name "David", as if
they were sitting there trying to think of a name for the baby, and
they suddenly said, "Duh, Duh, Day, Day, Dave, Dave, David,
That's it! David! ". That is not what happened. There was a pre-
existing generic name "David". My parents were aware of it
and of many of its associations, including the fact that the
common currency name of the great philosopher David Hume
was to be drawn from it. That is how they thought about the
generic name. Having David Hume's common currency name
in mind, and in honorof its referent,they decided, as is our custom,
to name their child with a common currency name drawn from
the same generic name. That is the sense in which there is a
causal connection between my common currency name and
that of David Hume. But such a connection is clearly not of the
right sort for the names to be the same. The earlier discussion
of the notion of repetition was an effort to show that the so-called
causal link fell in the realm of the intentional, and to
discriminate it from certain other intentional connections like
being named in honorof another.
2.3 AreNamesPartsof a Language?When I first started thinking
about generic names I was aware of a lot of literature about

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 113

names that says that names are not part of any language because
one can just create new names at will, whereas the lexicon of
words other than names forms a relatively stable body
characteristic of the language. If you think in a careful way
about common currency names and generic names, you will
recognize that this is not correct. Common currency names can
be created at will. But so far as I know, natural languages have,
at a given time, a fairly fixed stock in its lexicon of generic
names." When you expect a child you go to the bookstoreand
you buy a book called something like "What to Name the Baby;
3,000 Generic Names of English".Just as the dictionary listsand
'defines' 50,000 wordsof English, the generic names book will list
generic names and provide some lore about each name, such as
information about several famous people whose common
currency names have been drawn from it. When we name
people, we usually draw from this relatively small, finite lexicon
of generic names.
The idea that you can freely name someone anything you
choose is very misleading. "What shall we name the baby,
dear?" "How about Tkbtkbtkbt?" No way! It wouldn't be
English. How would you pronounce it? "Tkbtkbtkbt"just isn't
a generic name ofEnglish. You can'tjust decide you are going to
name a baby with that. There are a lot of names, you have a lot
of choices, but you can't name it that.
My point here is that names, like other words, must subscribe
to certain regularities. Certain of these regularities have to do
with admissible sound and spelling patterns, and others are
simple matters of social control. The free creation of names is
possible only to the degree that the linguistic community will
tolerate it. And the degree of tolerance in a linguistic
community for linguistic deviance in naming practice is an
empiricalquestion, not one to be solved by philosophical analysis.
PARIS-Marc Borneckexpected some interestingreac-
tions when he named his daughter Prune last fall. He
hasn't been disappointed.
To begin with, his mother-in-law hated the name

"71am thinkinghereof genericnamesforpersons.Therearealso'rules'constraining

the genericnamesforracinghorses,showdogs,etc.JohnM. Carroll'sWhat's inaName
(W.H. Freeman,1985)containsmuchinterestinginformation aboutnamingpractices.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
114 I-DAVID KAPLAN

(which in French means plum, not prune). But little Prune

turned out to be so winning that Grandmother soon was
won over to the name, too.
The neighbors were taken aback. But Mr. Borneck is a
beekeeper by profession, and they figure--rightly--that
this wasjust another of his effortsto get his children to love
nature.
But the people who were really outraged were in the
French government. And that was what mattered the
most.
The local prosecuting attorney informed the Bornecks
that the name was not just absurd but also illegal. He sent
policemen to the Borneck home to confiscate their family
book, a quasi-officialregisterof important family events, in
which local registrarshad inscribed the name Prune. And
he ordered that the girl henceforthbe known by her middle
names, Mae Kim.
He didn't like those much, either, and says he could have
voided them, too--substituting, if necessary, names of his
own choosing. But he says he wanted to be a nice guy about
it.
The Bornecks,who still call their daughter Prune, have
taken the matter to court, and so far they have lost. They
have discovered, like other unconventional parents before
them, that first names in France are strictly regulated by
law-a law drafted in 1803 under Napoleon, who disliked
offbeat names although his own didn't seem to hold him
back much.
"Personally, I think 'Prune' is kind of cute," says
Michele Signoret, A Justice Ministry official. "But is it a
name? That's up to the courts."'s
2.4 PromisesKept by the MischevousBabylonian.Following the
discussion of generic names, I am now in a position to fulfill my
promise to show how there can be distinct names which are
phonographsand which also haviethe same semantic value. Let
'8 The Wall StreetJournal,Western Edition, June 10, 1987. I am informed that in
Mussolini'sItaly a law was passed forbidding the naming of children with any (generic)
name that had already been used. This contrastswith the Netherlands in which you are
not permitted to use any name that has neverbeen used. (Perhaps the Italian law only
forbade naming children after the parents.)

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 115

me tell you about the case of the mischievous Babylonian. One

evening, the mischievous Babylonian looked up and saw Venus,
and he thought to himself "This one is just as beautiful as
Phosphorus, so let's call it 'Phosphorus' too". Now maybe he
actually knew he was naming the same thing; maybe he was
a time traveller playing a little joke on his Babylonian friends.
But more likely, he didn't know. He just decided that he would
name this rather attractive 'star' in honorof the great and
beautiful heavenly body seen in the morning in a rather
different location during a different season. So he names, or
perhaps we should say renames, Venus "Phosphorus". We may
suppose that this name passes muster with the Babylonian
Justice Ministry and comes into common use. Now it seems clear
that we have two common currency names "Phosphorus",one
somewhat older than the other, and that they start out, at least,
as phonographs. Who knows, after a little while they may drift
apart in terms of pronunciation because as astronomers talk
more and more about the sky, they might feel that it is confusing
to have the same generic name "Phosphorus"for two 'different'
heavenly bodies. So they might add suffixes and start calling
them "Phosphorus I" and "Phosphorus II". But for a little
while, they were distinct phonographic common currency
names with the same semantic value.'9
2.5 Are CommonCurrency NamesJust GenericNames + Referents?
I want explicitly to disavow one possiblemisapprehensionabout
my notion of a common currency word. It is not simply the
notion of an orthographic word combined with a meaning, as
some have thought because of the way I distinguish common
currency names from generic names. They have thought that a
generic name was just an orthographic word, and a common
currency name was just an orthographic combined with a
meaning (or a referent, insofar as they differ). No!
As noted above, generic names are also natural objects. They
undergo the same changes in spelling and pronunciation that
the individual common currency names do, more so because
" I have
recently become aware of a similar claim in footnote 28 of Saul Kripke's
NamingandNecessityfirst published in G. Harman and D. Davidson (eds) Semanticsoj
Natural Language(Dordrecht: Reidel, 1972); revised edition published as a separate
monograph, Naming andNecessity(Oxford: Basil Blackwell, 1980). Reference is to the
revised edition.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
116 I-DAVID KAPLAN

they are older and range more widely. Generic names are not
orthographic words. In fact, they almost are the common
currency names but for a differencein individuation. I say almost
because there are inscriptions of generic names that are not
inscriptions of any common currency name (as in WhattoName
theBaby).These matters are subtle (and perhapsare slipping out
of my grasp).20
This suggests another incorrect hypothesis: that common
currency names are generic names combined with a meaning
(or a referent). This is more plausible because generic names
have the appropriate wordly character but are entirely without
meaning (they are empty vessels). However, this analysis would
not give us the actualcommon currency names of a natural
language. On this conception there is a possiblename which
combines me, as referent,with the generic name "Phosphorus".
But I have never been named that.
So suppose we try, as a last ditch attempt: A common
currency name is a generic name combined with a referent that
has actually been given that name. (Don't come down too
heavily on 'given name', that's another topic.) But even this is
wrong. The case of the mischievous Babylonian makes it clear
that distinctcommon currency names can share the samegeneric
name (i.e. be of the same genus) and have the samereferent.
All these proposalsto understandcommon currencynames as
'combinations' miss an essential feature of my conception.
Common currency names (and other common currency words)
are not abstract constructions, they are natural objects. Not
physical objects, though most will have physical embodiment at
many places and times. And not mental objects, though most
will have mental embodiment (an oxymoron?) at many places
and times.

'0Suppose the common currency name used to dub the baby is not drawn from any
pre-existing generic name. Then a new generic name is introduced in the act of dubbing
the baby with its common currency name. Now suppose that the common currency
name flourishes,but that the generic name is never used (i.e. never uttered or inscribed)
except through uses of the common currency name. What exactly does the difference
between the two names consist in?They must be distinct since the generic name doesn't
name anyone and also has a capacity (even if unexercised, ex hypothes)--to 'generate'
other common currency names-that the given common currency name lacks. This
problem seems analogous to asking about the ontological status of a new species of
animal of which only one specimen ever exists.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 117

One might think of them as trees. Stemming out from their

creation, with physical and mental segments; the mental
segments able to produce many physical branches and able to
merge many physical branches, and the physical segments each
stemming from a unique mental segment and able to produce
many mental branches. (Perhaps I need a diagram). At any
rate, they are objects of the created realm, created by language
makers. The world is not brimming with unspoken words.
Words never actually created are not.
2.6 CanCommon Currency WordsChangeinMeaning?Or, to be more
specific and more directly relevant to the purposes I have in
mind, can a common currency nameundergo a change in
referent?There is noprimafaciereason against it. I re-emphasize:
The identity of a common currency word lies in its continuity,
both interpersonaland intrapersonal,as has been discussed.It is
a matter for furtheranalysis to say whether such an entity could
change meaning (or reference). It is certainly no part of my
conception that it cannot.
The matter does, however, call for carefulthought. One might
consider two kinds of polar cases:In one case you intend to use (to
repeat)a given common currencyname with whatever referentit
may have. ("What is Hesperus?"you ask,overhearinga conversa-
tion in which the name is used.) In the antipodal case, you intend
to dub a particularthing usingan apt genericname. In the former
case there is continuity, in the latter, creativity, a new name is
created. But there are those troubling cases (first thrust upon
our consciousness by Keith Donnellan, and then Gricefully
reconceptualized by Saul Kripke) that seem to lie in between:
the man with the Martini, the false introduction, and their ilk.
It would be gratifying to be able to show that the process
whereby a common currency name appears to change its
referent involves these middling and conflicting intentions in
such a way that when we are prepared to say that utterancesof a
name now have a new referent, there will have been sufficient
weight given to the creative side of the intentions for us to claim
that a new name has replaced the old. It might thus turn out that
names don't in fact change in referent.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
118 I-DAVID KAPLAN

So one strategywouldbe this.Try to showthatsomethinglike

Donnellan's useis involvedincasesofapparent
referential change
of referentof a givencommoncurrencyname.Then tryto show
that this sortof referentialuse is sufficientlylikecreatinga new
commoncurrencynamefromthe genusof a givennameso that
by the time a new 'semanticreferent'appears,a newnamedoes
also. I don't know that such a strategywould be successful.
2.7 EssentialProperties of Names.One consequence of my view of
names as wordly creatures may be to renew interest in some
matters that had seemed obvious or irrelevant.
In 'Uber Sinn und Bedeutung' Frege argues, as against his
earlier 'Begriffschrift' view, that "Hesperus = Phosphorus"
could not mean that the name "Hesperus" is co-referentialwith
the name "Phosphorus" because the connection between a
name and its referentis arbitrary. "You cannot forbid the use of
an arbitrarilyproduced process or object as a sign for something
else." Frege concludes that no "genuine" knowledge would
then be expressedby "Hesperus= Phosphorus",only knowledge
of what one might call an accident of human culture.
But is it a mere accident that our name "Hesperus"names the
planet Venus? To my way of thinking that is like askingwhether
Da Vinci's painting of the Mona Lisa might have been a picture
of a horse. Certainly Da Vinci might have painted a horse
instead of a woman. As Frege might say, "You cannot forbid a
painter from arbitrarily deciding what to paint." But would
the resulting painting have been the Mona Lisa? I won't try to
answer this question. But I note that though Venus might never
have been named "Hesperus" (it might never have been named
at all), and though Alpha Centauri might have been named
"Hesperus" (it might even have been named "Hesperus"due to
a last minute change of mind by the Da Vinci-like Babylonian
who in fact named Venus "Hesperus"),it does not follow that our
name "Hesperus" might have named Alpha Centauri, at least
not on my view of names.2'
The question, "Is it possible that a name which in fact names
a given individual, might have named a different individual?"

2 One refinement: there may be reason to think that our name "Hesperus" could
have cometo name Alpha Centauri. My question is: At its creation, could it have named
Alpha Centauri?

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions
 WORDS 119

is, for me, a substantial metaphysical question about the essence

of a common currency name. By calling the question a
substantial,metaphysicalone, I do not intend to puff up its
importance nor to make it seem mysterious or occult. Perhaps,
in the end, the question calls only for a decision, or perhaps, in
the end, the question will seem unimportant. This may be the tao
of substantial metaphysical questions. But there is not, I believe,
an obviouslycorrect answer.22

22The last minute assistanceofJoseph Almog and Keith Donnellan has done much to
make this readable.

This content downloaded from 193.140.201.95 on Tue, 02 Jun 2015 14:09:02 UTC
All use subject to JSTOR Terms and Conditions