Naturalism and Semantic Normativity

Author(s): Jaegwon Kim

Source: Philosophical Issues , 1993, Vol. 4, Naturalism and Normativity (1993), pp. 205-
210
Published by: Ridgeview Publishing Company

Stable URL: https://www.jstor.org/stable/1522841

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1522841?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Issues

D
JSTOR
This content downloaded from
130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
 PHILOSOPHICAL ISSUES, 4
19 Naturalism and Normativity, 1993

Naturalism and Semantic

Normativity

Jaegwon Kim

The recent and continuing debate on the realism/antirealism issue

has largely focused on the nature of semantics. Should the seman-
tics of our language be realist and interpret meaning in terms of
"reference" to mind-independent objects and properties and truth
in terms of "correspondence" to an objective reality? Or should
meaning and truth be understood in epistemic terms such as "war-
ranted assertibility"? The debate has primarily been a debate about
what the correct semantics of our language ought to be. But what if
we look at semantics from a naturalistic point of view? That is, what
if we overlay the constraint of naturalism over the realist/antirealist
debate? Where will we come out on the question of realist vs. an-
tirealist semantics?
These are interesting questions; they are the questions that mo-
tivate Horgan and Timmons's interesting work on what they call
"meta-semantic irrealism" .1 But what is naturalism? Horgan and
Timmons take it to be an ontological thesis to the effect that "ev-
erything -including any particulars, events, facts, properties, etc.-
1 1n "Metaphysical Naturalism, Semantic Normativity, and Meta-Semantic
Irrealism", in this volume.

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
206 J AEGWON KIM

is part of the natural physical world that science investigates".

Whether physicalism ought to be a commitment of naturalism is
a debatable question; however, since this issue doesn't play a role in
Horgan and Timmons' arguments, I will set it aside. In any case, I
take their thesis to mean that everything in the naturalist's ontology
must be capable of being investigated by science, and this includes
not only concrete particulars and their aggregates but also the prop-
erties and relations that characterize them. Naturalism, roughly in
that sense, has been popular and influential in philosophy. Among
the many naturalization projects, "naturalized epistemology" is per-
haps the best known. 2 Now, Horgan and Timmons propose to
bring naturalism to semantics. And this is their "accommodation
project". As naturalized philosophers we must interpret various
types of discourse -e.g., mental discourse, moral discourse, mathe-
matical discourse, etc.- in a way that is consistent with naturalism.
This means that the ontology of any acceptable discourse must be
rendered naturalistically appropriate -that is, it must not posit
entities and properties that cannot in principle be investigated by
science.
Since the accommodation project involves the interpretation of
various types of discourse, there is a need for a general semantic
framework, and this is where Horgan's "contextual semantics" -
what he used to call "psychologistic semantics"- comes in. His
approach is designed to occupy an intermediate position between ref-
erentialism, which views truth as correspondence between discourse
and a discourse-and mind-independent world, and antirealism (Hor-
gan and Timmons call it "pragmatism") which takes truth as "radi-
cally epistemic", identifying it with the so-called "warranted assert-
ibility". On Horgan's view, truth is to be identified with correct
assertibility, rather than warranted assertibility, where what deter-
mines the correct assertibility of a sentence, and hence its truth, is
a combination of two factors: (i) how things are in the world, and
(ii) the norms of assertibility applicable to that stretch of discourse.
What is interesting in Horgan's theory, and gives it power and flex-
ibility, is the fact that it permits these two factors to be combined
in variable proportions, as it were, to suit the particular type of
discourse under consideration. Horgan suggests that mathematical
language is perhaps one where the first plays no role and the sec-
ond takes complete control. At the opposite end, correspondence
semantics holds full sway, Horgan says, although he does not iden-

2 For an excellent review of naturalistic epistemology see Philip Kitcher, "The

Naturalists Return", Philosophical Review, 101 (1992): 53-114.

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
 19. NATURALISM AND SEMANTIC NORMATIVITY 207

tify in this paper the kinds of discourse for which he considers such
realist semantics is fully appropriate. The question that Horgan and
Timmons want to answer in their paper is this: Just where on this
spectrum of possible semantics, ranging from fully realist correspon-
dence semantics to radical antirealist pragmatic-epistemic seman-
tics, should we locate semantic discourse itself -that is, discourse
involving such properties and relations as truth, meaning, sameness
of meaning, reference, extension, implication, and the like?
Horgan and Timmons choose not to place semantic discourse at
the realist extreme, or anywhere close to it. For they tell us that
although truth is correct assertibility and there are true sentences,
the property of correct assertibility does not exist. And since there is
no such property, there are no facts of the form "Sentence S is cor-
rectly assertible". That is, the nonexistence of semantic properties
entails the nonexistence of semantic facts. Indeed, there can be no
"fact of the matter", on this view, about whether anything is cor-
rectly assertible or isn't correctly assertible. Given the central claim
of Horgan's contextual semantics that truth is correct assertibility,
we seem driven to the conclusion that there are no true, or false,
sentences. And this may look like a reductio of contextual semantics
itself, or at rate of its raison d'etre. No wonder that Horgan and
Timmons refer to this as "the threat of ultimate incoherence" for
their theory.
I will no go into the details of their attempt to dissipate this appar-
ent threat. Rather, I want to focus on the question why they want to
locate semantic discourse toward the antirealist end of the spectrum,
if not at the extreme end point. We can begin with a short answer:
it's because they hold irrealism and nonreductionism about seman-
tic properties. Irrealism says that semantic properties and facts do
not exist, and nonreductionism says that they are not reducible to
properties and facts that are part of a naturalistically acceptable
ontology. I have been persuaded by Horgan and Timmons that if
nonreductionism and irrealism are accepted for semantic properties,
contextual semantics has the potential to yield a more satisfying
"accommodation" for them than fully referential semantics.
But why do they accept irrealism and nonreductionism about se-
mantic properties. Because they think that semantic properties are
normative, and that normative properties in general are irreducible
and are not part of the naturalistic world. I will try to avoid this
general issue about normative properties, the central point at issue
in the moral realism debate. But why do Horgan and Timmons
think that semantic properties are normative properties, to begin
with?

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
208 J AEGWON KIM

To Horgan and Timmons, normativity arises out of "stance-tak-

ing". What's stance-taking? Taking a stance is something like ac-
cepting a set of rules and conventions to regulate one's behavior vis-
a-vis other members of a given group, for the purpose of behavior
coordination within the group. Stance-taking is normative, or gener-
ates normativity, because it involves commitment to a set of norms.
Linguistic normativity, on this account, arises out of "normative-lin-
guistic stance-taking", commitment to a set of conventional norms
that regulate the uses of noises, marks, and gestures for communica-
tive purposes. And "linguistic behavior, speaking or writing 'in a
language', is behavior from within a stance, behavior qua speaker
of a given language. To speak a language is to act from within the
relevant linguistic-normative stance" (p. 192).
This is not an implausible, or totally unfamiliar, picture of lan-
guage as a social instrument. It's picture that is of a piece with
the "language game" model of language. But Horgan and Timmons
go on to repeat exactly the same thing for semantic discourse it-
self, namely discourse about language -or about language and the
world. They say: "To employ semantic discourse is to speak (and
judge) from within a specific linguistic-normative stance", adding
"it is not to predicate a PROPERTY or state a FACT" (p. 194).
To speak English is, let us agree, to behave "from within" a certain
normative-linguistic stance -or from within "the English-language
stance", as we might say. But suppose you want to engage in a
bit of semantic discourse about English. According to Horgan and
Timmons, this, too, involves stance-taking; that is, you could do this
only by taking a certain normative-linguistic stance. That seems like
a reasonable thing to say: semantic discourse after all is discourse,
and it must therefore involve a linguistic stance. But what particular
stance would you be taking? The English-language stance?
That is possible, but perhaps not necessary. For consider Karl,
the noted German semanticist specializing in English. Karl asserts,
in German, statements like the following (please imagine these as
German sentences):
(1) "Husband" is synonymous with "male spouse" in English
(2) "Munich" refers to Muenchen
(3) "Snow is white" is true in English iff snow is white
In writing them, or using them to communicate his findings, Karl is
obviously speaking from "the German-language stance" (remember
that these are supposed to be German sentences), the stance that

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
 19. NATURALISM AND SEMANTIC NORMATIVITY 209

speakers must take if they are to count as speakers of German. But

obviously this has nothing to do with the fact that (1 )-(3) are seman-
tic sentences; it only has to do with the fact that they are German
sentences. In this respect, Karl seems to be doing no more, or no
less, normative stance-taking than his German colleagues in physics
or paleontology.
If Karl's stance-taking should go beyond that of his fellow-Ger-
man speakers, it must be because of his involvement with another
language, and here the English-language stance is the only other
stance on the scene. Must Karl take the English-language stance
in uttering German sentences such as (1)-(3)? It isn't clear why he
must. It seems to me that Karl's perspective on the semantics of
English need not be anything more than one of an observer, not of a
participant. He doesn't have to be a player of the English language-
game; he is merely a spectator observing the game, for he wants to
find out the rules according to which the game is played. If this is
right, there is unlikely to be semantic normativity, or normativity
attaching specifically to semantic discourse or semantic properties,
over and beyond the normativity attaching to general language use.
The conclusion then seems plausible: no specifically semantic norma-
tivity needs to be acknowledged over and beyond general linguistic
normativity.
And in uttering the likes of (1)-(3), Karl seems to be merely re-
porting facts, pure and simple. These sentences report facts about
language use among the speakers of English -about the semantic
rules in force among them, or the consequences of those rules. In
short, they report, in German, facts about "the English-language
stance", facts about the normative behavior of a group of people.
But this doesn't mean that they themselves are statements of norms,
or normative statements (in the sense in which an article of the
United States Constitution is such a statement, and not a descrip-
tion of one). What Karl says has something to do with norms and
normative behavior; but that doesn't mean that in saying it he is
himself engaged in some special kind of normative behavior (over
and beyond the normative behavior involved in speaking German).
If I am studying Sanskrit semantics, is my activity governed by the
Sanskrit-linguistic stance?
Compare all this with the rules of the road (something Horgan
and Timmons actually mention). In driving a motor vehicle in a
given country (or perhaps in applying for a driver's license), you have
taken, let us agree, a certain normative stance: you have agreed to be
governed by a set of rules, such as driving on the right-hand side of
the road. So all licensed drivers of Florida drive "from within a nor-

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms
210 JAEGWON KIM

mative driving stance", as we might say. However, if you are a traffic

engineer, or a sociologist, observing and reporting on the driving be-
havior of the inhabitants of Florida, you don't have to be doing that
from any special normative stance -certainly not from the Florida-
driving stance. After all, you don't need a Florida driver's license to
do it!
In conclusion, let us briefly consider correct assertibility, which
Horgan and Timmons propose to identify with truth. As was noted,
correct assertibility for Horgan and Timmons is determined by two
factors, the operative linguistic norms governing the relevant stretch
of discourse and the way things are in the world. Now, it seems to
me that what norms are operative, say, for the English language,
are fixed by the facts about the stance-taking behavior of English
speakers. If this is right, the two factors that are supposed to jointly
determine correct assertibility seem to collapse to one, namely the
way things are in the world. Why doesn't this show that correct
assertibility itself is entirely fixed by the facts of the world? Why
doesn't this make correct assertibility a real property, a property
that truly exists as a properly naturalistic property?
Look at this another way. Remember that on Horgan and Tim-
mons' view, something is naturalistically acceptable just in case it
can be investigated by science. Under their account of linguistic nor-
mativity, I don't see a clear reason why correct assertibility could not
be a possible object of scientific investigation. Aren't they in fact
tacitly assuming this in giving a sketch of an evolutionary account
of linguistic normativity in terms of social behavior and "stance-
taking"? Why isn't this a "scientific investigation" of linguistic nor-
mativity?
Horgan's contextual semantics is interesting and thought-provok-
ing; I believe it is an important contribution to the metaphysics of
semantics. Here I have raised some questions about his and Tim-
mons' application of this approach to semantic discourse itself, m
the hope of eliciting further clarification and illumination.

This content downloaded from

130.225.157.199 on Tue, 03 Oct 2023 11:21:24 +00:00
All use subject to https://about.jstor.org/terms