Philosophy of Language
Philosophy of Language
Philosophy of Language
Zegarac, V. (1998) 'What is phatic communi how the meaning of complexes depends upon
cation?', in V. Rouchota and A. Jucker (eds) such part-meanings, together with the contribu
Current Issues in Relevance 1heory, Amsterdam: tions of syntax). It is no accident that these two,
John Benjamins. syntax and semantics, also constitute the core of
artificial logical languages: System Theorists self
avowedly take such languages as their models,
Philosophy of Language not least because one of their aims is to capture
Introduction: preliminaries and game plan logical relations among natural language sen
tences.
Philosophy of language is an extraordinarily rich This may sound very much like the project of
field. It has a history stretching back, in the Chomskyan generative grammar. And, indeed,
Western tradition, to the pre-Socratics. And, in Chomsky's earliest work finds its roots in the
the last century or so, it has been of central System Perspective. But there remains a funda
concern in both the Anglo-American and Con mental difference. For the philosopher of lan
tinental traditions. Obviously, a brief survey guage who inclines towards the System
cannot hope to cover such intellectual abun Perspective, and in sharp contrast to the
dance. What's more, as this encyclopedia itself approach of generative grammarians, the rules
attests to, pragntatics is an equally rich aca presented are not designed to capture how we
demic endeavour. Any mere overview of their humans mentally process language, any more
intersection must, then, narrow its focus. As a than formal logic seeks to characterize the psy
result, my specific topic will be: What has Anglo chological processes involved in human rea
American philosophy of language contributed to soning. To offer a standard comparison: for the
the study of utterance meaning in context? System Theorist, to describe a language is akin
The game plan is as follows. I present two to laying out the rules of chess itself, as opposed
traditional perspectives in philosophy of lan to offering a description of the thinking processes
guage, and describe some illustrative contribu within a given chess player. (For discussion, see
tions of each to pragmatics. I end by explaining George 1989; Iten et al. 2007; Katz 1981, 1985;
how these two philosophical perspectives have Lewis 1975b; Soames 1984, 1985).
recently been combined, thereby affording a still The parallel between a language as construed
richer and deeper contribution to pragmatics. by the System Theorist and an algebra leads to
the next key feature of the System Perspective:
Two traditiona.1 perspectives in philosophy of its understanding of what a 'linguistic item' is. A
language linguistic item, on this view, whether a word,
phrase or sentence, is something that exists in
It is a simplification, not to say artificial, to abstraction from use and users. Linguistic items
summarize twentieth-century Anglo-American are types, as opposed to tokens/instances/utter
philosophy of language in terms of only two ances of those types. (For those unfamiliar with
dominant perspectives. Nevertheless, given the the type/token contrast, think of the difference
expository purposes of this article, that's mostly between the novel War and Peace itself and the
what I will do. tens of thousands of copies of this novel. Each
One tradition, call it the System Perspective, copy has a certain weight and extension, a cer
thinks of a language as a collection of formal tain location, etc. This is like the tokens of lin
rules: rules which are so simple as to require no guistic symbols: they are spatiotemporally
insight to apply them. A language, on this view, located specific instances, with all the associated
is like an algebra, with its axioms and rules of physical features of physical objects. But the
proof. System Theorists have mainly focused on novel itself is not located in anyone place, has
two kinds of rules: those of syntax (which no specific weight, etc. This is like a linguistic
describe how minimal linguistic elements are put type.) Linguistic items, on this view, are thus
together into complex wholes) and those of outside us. Indeed, they are outside the physical
setnantics (which describe what each minimal world as a whole: like nurnbers, sentences,
linguistic element means in the language, and phrases and words - the types, that is - are not
Philosophy of Language 349
manage to be about our world: they manage this utterance content, philosophers such as Kaplan
because they are tokens of types which, as the (1989a) and Stalnaker (1970, 1978) introduced
System Perspective highlights, are themselves the notion of a character. This is a function (again,
about our world. in the mathematical sense of 'function') from
One might reasonably complain: 'We don't certain aspects of a context to a truth-relevant
need all the formal complexity and abstractness entity. For instance, the standing meaning of 'I
of the System Perspective to tell us about the once lived here' would be such a character: spe
standing meaning of linguistic items. We merely cifically, a function from triples of <person, time,
need to describe the use to which the various location> to a proposition about the person who
linguistic tools are put.' But, I would insist, this is the speaker in the context, to the effect that he
overlooks a more specific contribution of the or she lived at the location of the context at
System Perspective. Only given the resources of some point prior to the time of the context.
the System Perspective can we explain the Hence, should this sentence be spoken by, say,
unbounded productivity of human speech: the Noam Chomsky, in Boston, on January 1st,
System Perspective's rules are recursive (i.e. the 2009, the character of the sentence would, by
output of a rule can serve, once again, as an means of a language-internal rule, deliver as
input) and the meaning of each sentence is output the content NOAM CHOMSKY ONCE
determined compositionally (i.e. the meaning of LIVED IN BOSTON AT SOME POINT
the type is exhausted by what its minimal lin PRIOR TO JANUARY 1st, 2009. The key
guistic elements mean, and how those parts are point is that the System Perspective not only
put together by the syntactic rules). These two helps us understand utterance meaning as
features yield a potential infinity of meaningful deriving directly from context-insensitive type
expression types, including ones that have never meaning, it also helps us understand how utter
before been tokened. The need for this kind of ance meaning is fixed by variable context.
complex machinery can be illustrated with a Let's turn now to some illustrative contribu
relatively simple example. The study of utter tions from the Use Perspective. These come in
ances in context needs to say how a particular two flavours: literal and non-literal utterance
use of, for example, 'Last night I dreamed that I content. Use Theorists pointed out early on that
spoke with a tiny pink elephant in the University there at least two kinds of content that are lit
of Western Ontario student centre' manages to eral - indeed, contents which derive wholly from
mean what it does. This requires saying what the meaning of the type itself - but which the
this sentence type means, which in turn requires System Perspective tended to ignore. These are
the kind of complex and abstract rules afforded contents that seemingly do not alter the truth
by the System Perspective. conditions of the utterance, but nevertheless
Second, the System Perspective has afforded contribute to its literal meaning. One sub-vari
invaluable insights into an important way in ety, which Grice labeled 'conventional iDlpli
which utterance meaning, in a specific context, catures', shows something about the speaker's
typically outstrips the standing meaning of the attitude towards the truth conditions of the
linguistic expression used. Specifically, it has utterance, e.g. that he or she finds them surpris
provided insights into certain features of utter ing, or in tension with each other. Instances
ance meaning that, though they do not derive include 'surprisingly', 'but', and 'therefore'.
entirely from standing meaning alone, are Thus, putting things roughly, to say 'Surpris
nevertheless highly constrained by it. Examples of ingly, John won' is truth conditionally equivalent
this kind of linguistically constrained contribu to John won.' However, using the former sen
tion by context include deDlonstratives ('this', tence is a way of linguistically indicating sur
'that'), pure indexicals ('here', 'now', '1'), and pnse.
tense markers. Syntactic mood affords another example of
The details are not especially important for non-truth-conditional content that attaches to
present purposes. Nonetheless, here is a brief the expression type. Mood is an indicator of
example to spell out the idea. To account for illocutionary force potential. Thus, the sentences
how (certain) context-sensitive words help to fix 'Chomsky left' and 'Did Chomsky leave?' per
Philosophy of Language 351
tain to the exact same topic, that is, their truth such contributions in this regard include Don
conditional contents are identical. But they are nellan (1966), Grice (1975) and Searle (1975b,
by no means synonymous sentences. Instead, the 1979a).
declarative mood of the former sentence To recapitulate, I have summarized recent
encodes, as part of its content in the language, Anglo-American philosophy of language by pre
that the illocutionary force is assertoric, whereas senting two (idealized) traditions: the System
the interrogative mood of the latter sentence Perspective and the Use Perspective. I have also
encodes that the illocutionary force is inter illustrated some of the contributions of each tra
rogatival. Crudely, the import of these illocu dition to pragmatics, here understood as the
tionary forces is that the first sentence has a 'use study of utterance meaning in context. Many
theoretic' content USED TO STATE, while the more examples could be presented, but the
second sentence has a 'use-theoretic' content above provide sufficient background to move
USED TO ASK. Other force indicators are forward. The final section will describe a revo
more specific and, rather than attaching to syn lutionary means of combining the two perspec
tactic mood, they are carried by an explicitly tives.
performative verb, such as 'promise', 'swear' and
'pronounce'. Thus, the expressions 'I promise
COrTlbining the two perspectives
to - ' and 'I swear to - ' wear their use-theoretic
contents on their sleeves: the first has as part of Our problem amounts to this. Given that the
its content USED TO PROMISE while the two perspectives are complementary in numer
second has as part of its content USED TO ous ways, we ought to combine them. Yet, they
SWEAR. appear to be in deep conflict about many fun
In addition to linguistically constrained con damental matters. So, unifying them seenlS dif
tributions to literal content that involve prag ficult if not impossible.
matics, the Use Perspective has contributed Let's first revisit the obstacles to unification.
enormously to our understanding of non-literal The two perspectives disagree about what a
usages. Such features of utterance content, that language is (a collection of formal rules versus a
are shaped more by speaker's intentions than socio-cultural practice), what a linguistic item is
by standing meaning, include conversational (an abstract type versus a concrete performance)
implicatures, speaker's reference, metaphorical and about what meanings are (abstract truth
speech, and indirect speech acts. Each of these is relevant entities versus actions). In short, one
described elsewhere in this volume. It suffices, tradition focuses on users and usage, while the
then, to provide a quick example of each. Saying other abstracts away from them. Another prob
'I am French' to convey that one is a good cook lem for combining the two perspectives is what I
illustrates conversational implicature: the will call the 'ontological gap'. Linguistic types
speaker says one thing but implies another. An are abstract entities. Existing outside space and
exanlple of speaker's reference would be using time, they cannot themselves cause utterances.
jake's mother' to speak of Jake's much older And, in turn, no collection of utterances, no
sister. This phrase does not itself designate the matter how large, in and of itself constitutes a
sister. Yet, in the right circumstances, a speaker type. On a related note, Use Theorists are wont
may manage nonetheless to refer to Jake's sister to insist on the many diverse actions we perform
with these words. Metaphorical speech is famil with language, and on the contextually bound,
iar to all. For instance, Dylan Thomas famously creative jumble that is actual talk, whereas
wrote about his father, 'Do not go gentle into System Theorists higWight the pristine elegance
that good night', meaning that the latter should and compositional-recursive power of languages
fight to remain alive. Finally, indirect speech acts themselves. Even setting aside their disagree
include using an interrogative, the assigned use ments about what languages, linguistic items and
of which is asking a question, to make a meanings are, then, how can the two perspec
request. 'Do you have any cold beer in the tives be combined, if there is such an enormous
fridge?' may be used, in context, not to enquire metaphysical gulf between their respective
but to request politely. Use Theorists who made objects of study?
352 Philosophy of Language
Yet, as noted, there are many reasons to hope content of utterances, our knowledge of lan
for a unification of the two perspectives. In par guage includes (a), (b) and (c):
ticular, each covers the other's omissions. In
terms of content, both truth-theoretic meaning (a) Knowledge of the truth-relevant content
and use-theoretic meaning (e.g. conventional of context insensitive minin1al elements
implicatures and illocutionary force) are neces (b) Knowledge of the character of context
sary for a complete account of natural language. sensitive minimal elements
Similarly, non-literal content is important, but so (c) Knowledge of the non-truth-conditional
is literal content: it's essential to keep in mind content of minimal elements
the enormous part that standing meaning, i.e.
the meaning of the type, plays in fixing utterance (In addition, to account for the productivity of
content. And in terms of philosophical orienta speech, note that we also know the recursive
tion' it seems that language is productive and syntactic rules for putting the minimal elements
rich in both senses canvassed in the last para together, and the semantic rules for computing
graph: context-invariant generative capacity and the meaning of a complex expression on the
creative usage in context. basis of its syntax and the truth-theoretic and
In short, while it seems hard to unify the two use-theoretic content of its parts.) With respect
perspectives, there are strong reasons for doing to non-literal content, we also know many non
so. The solution I would like to draw attention to linguistic facts which allow us to interpret speech
is inspired by the work of the linguist-philoso in ways that knowledge of language alone would
phers Chomsky (1986), Fodor (1983), and espe not permit. We know general facts about the
cially Sperber and Wilson (1995). They reject a world and about people, and we know specific
common commitment of both traditional per facts about the speech situation. Finally, we are
spectives, namely, that psychology is irrelevant able to combine such knowledge with the infor
to the philosophy oflanguage. Yes, language is a mation that language proper affords. It is this
system of symbols but, crucially, it is a system diversity of mental capacities that allows us to go
which we humans know. And it is precisely from the literal content of the utterance in con
because knowledge of linguistic rules is stored in text, as afforded by highly constrained linguistic
the mind/brain that it can give rise to use. rules, to non-literal contents.
The first advantage of this approach is that it I would summarize the proposed unification
bridges the ontological gap. Abstract entities with a slogan: 'Language is by equal measures a
may not cause utterances, but mental states can. system of symbols which we know and use'. That
And, while no collection of utterances can give is, it is fundamentally and essentially all three.
rise to types, mental processes can extract type This merges the two traditional philosophical
meaning from a collection of tokens. So, knowl perspectives. It thereby allows their independent
edge of the system bridges the gulf between lin contributions to come together - yielding a still
guistic types and tokens. In a similar vein, we deeper and richer contribution of recent Anglo
can admit both kinds of creativity (i.e. within the American philosophy of language to the study of
language itself and in usage) by distinguishing, as utterance content in context.
Chomsky does, between cODlpetence and per
formance. The former consists in the rules R.S.
known. Importantly, however, linguistic compe
tence is only one of the causes of performance: See also: Assertion; Austin,].L.; Bar-Hillel, Y.;
our performance (i.e. actual speech) is an inter competence, linguistic; competence, pragmatic;
action effect of such knowledge with much else context; cooperative principle; demonstratives;
besides. That's why we get gloriously near-anar explicit/implicit distinction; formal pragmatics;
chic speech from such highly structured linguis Grice, H.P.; implicature; indexicals; intention;
tic rules. logical form; maxims of conversation; metaphor;
Recognizing that we know the system also modular pragmatics; modularity of mind thesis;
makes it easy to accommodate all the varieties of neo-Gricean pragmatics; ordinary language phi
content discussed above. With respect to literal
Philosophy of Mind 353
question; radical pragmatics; reasoning; refer In the wake of Chomsky's (1959) attack on
ence; relevance theory; scalar implicature; behaviourism, two major materialist theories
Searle, J.; semantic minimalism; semantics about the nature of mental states have emerged:
pragmatics interface; speech act theory; utter the (psychophysical) identity theory and func
Carston, R. (2002) 17zoughts and Utterances: 17ze H 2 0' and 'lightning = electrical discharge', it is
Pragmatics of Explicit Communication, Oxford: hoped that scientists could generate identities
Lycan, W.G. (2000) Philosophy of Language: A category, as in the oft-used example of 'pain = c
Debates within philosophy of mind often begin still execute the same function, program, or
by examining the relationship between minds algorithm, philosophers thought that thinking
and bodies. Since minds and bodies seem so and feeling could also take place in a number of
different, it is natural to see them as different different materials. Putnam argued that whether
sorts of substances. This view, called mind-body some entity counts as a mental state is not
dualism, has been so associated with Rene Des determined by what it is made out of, but rather
cartes (1641, 1649) that it is often called 'Carte by what it does, i.e. its functional role in the
sian dualism'. These Cartesian dualists cannot mind. For example, anything that performs the
see how a merely physical thing could, inter alia, same function as pain - being caused by tissue
think, talk, exhibit consciousness, exhibit damage, causing further mental states, leading
rationality, or see itself and others as having to overt behaviour, etc. - counts as being a pain.
minds. Hence, something distinct must serve as Since creatures like the octopus, with nervous
the seat of these mental features - a mental systems very different from that of a human
substance. Although initially appealing, this view have parts that play functionally isomorphi~
has two major drawbacks. First, it fails to explain roles to pain in humans (not to mention robots
how mental substances are any more capable of or Martians!), pain cannot be identified as c-fibre
having these features than physical substances firing. Since pain and other mental states are, in
it simply asserts that the former substance has Putnam's words 'multiply realizable' in different
them. Second, dualists portray mental features sorts of things, mental states should be identified
as being so distinct from physical features that an by their functional role and not by what they are
obvious fact - that the mind and body interact made out of.
becomes deeply mysterious. Dualists must either Despite being the most widely held theory
deny mind-body interaction or generate some about the nature of mental states (for varieties
suitable explanation as to how such radically see Block 1980a, 1980b), functionalism has been
distinct things can interact with one another. criticized for being too liberal in its articulation
Each of these options has sustained critiques so of n1.entality. Block (1978) argues that there is no
severe that many have been led to adopt some way for a functionalist to identify the functional
form of monism - either everything is mental roles of mental states in such a way that avoids
(idealism) or the more widely held view that having complex systems that lack a mind (e.g.
_____________________________ 1