INQUIRY, 2017
http://dx.doi.org/10.1080/0020174X.2017.1310669
Must we measure what we mean?
Nat Hansena,b
a Department of Philosophy, University of Reading, Reading, UK; b Humanities Center, Stanford
University, Stanford, CA, USA
ABSTRACT
This paper excavates a debate concerning the claims of ordinary language
philosophers that took place during the middle of the last century. The debate
centers on the status of statements about ‘what we say’. On one side of the
debate, critics of ordinary language philosophy argued that statements about
‘what we say’ should be evaluated as empirical observations about how people
do in fact speak, on a par with claims made in the language sciences. By that
standard, ordinary language philosophers were not entitled to the claims that
they made about what we would say about various topics. On the other side of
the debate, defenders of the methods of ordinary language philosophy sought
to explain how philosophers can be entitled to statements about what we would
say without engaging in extensive observations of how people do in fact use
language. In this paper, I defend the idea that entitlement to claims about what
we say can be had in a way that doesn’t require empirical observation, and I
argue that ordinary language philosophers are (at least sometimes) engaged in a
different project than linguists or empirically minded philosophers of language,
which is subject to different conditions of success.
ARTICLE HISTORY Received 18 February 2017; Accepted 3 March 2017
KEYWORDS Ordinary language philosophy; experimental philosophy of language; J.L. Austin; Stanley
Cavell; linguistic meaning
1. The historical scene of the debate
In 1957, there was an official confrontation between philosophers with
radically opposed estimations of the significance of ordinary language
philosophy. On one side was Stanley Cavell, then still working on his
Harvard dissertation but already teaching at UC Berkeley. Cavell had been
deeply impressed by Austin’s practice of ordinary language philosophy
while attending his 1955 classes at Harvard (Cavell 2010, 322–326), and
he lectured on Austin in the spring term of 1957 at Berkeley (334–335).1
CONTACT Nat Hansen
n.d.hansen@reading.ac.uk
1 Austin himself was at Berkeley in 1958–1959, giving the series of lectures that would become Sense and
Sensibilia – see Cavell (1976b, 97 n. 3).
© 2017 Informa UK Limited, trading as Taylor & Francis Group
2
N. HANSEN
On the other side of the confrontation was his colleague Benson Mates,
a logician and historian of ancient philosophy. Cavell says, ‘My insistence
on the treasures I was finding Austin to have brought to philosophy was
getting on the nerves of some accomplished teachers in and around my
senior colleagues in the Berkeley department’, and he was invited (or
‘ordered’ [360]) to defend the methods of ordinary language philosophy
at the 1957 Christmas Pacific APA at Stanford. His opponent was Mates,
who himself taught a seminar on Austin’s ‘A Plea for Excuses’ in 1957.
Mates’ seminar was attended by several visiting advocates of ‘empirical semantics’, a midcentury version of experimental philosophy that originated
in Norway. The empirical semanticists rejected ‘anti-empirically oriented
armchair [philosophy]’ (Naess 1938, viii; quoted in Murphy 2015, 325) and
conducted surveys of how philosophically significant expressions were
used by non-philosophers. Mates and the empirical semanticists in his
seminar were skeptical about Austin’s informal methods of investigating
ordinary language, and Mates’ contribution to the APA symposium grew
out of notes from his seminar (Murphy 2015, 350–352).
Cavell reports that his response to Mates at the APA was ‘well enough
received’, meaning ‘that those who were pulling for me thought I had
answered Mates’s objections, and those who were not pulling for me
thought I had lived to fight another day’ (Cavell 2010, 361). Mates and
Cavell published their contributions to the symposium in the first volume
of the new Norwegian journal Inquiry, founded by Arne Naess, which also
went onto publish experimental work by the empirical semanticists (see
Chapman 2011).
Cavell’s defense of the methods of ordinary language philosophy against
Mates’ criticisms, ‘Must We Mean What We Say?’, was, Cavell says, ‘the
first text I published that I still care about’ (Cavell 2010, 361), and Mulhall
(1999, 1) says that ‘Cavell’s response to Mates underpins everything else
[Cavell] has written’. Cavell also reports that, according to Bernard Williams,
Austin had ‘pushed to have graduate students and younger dons at Oxford’ read it (Cavell 2010, 149). But it also attracted intense criticism from
advocates of the new, scientifically minded philosophy of language being
developed at MIT (Fodor and Katz 1963). Cavell, recalling the message of
Fodor and Katz (1963), says they asserted
. . . (I believe I can recall the exact words) that the articles [‘Must We Mean
What We Say’ and ‘The Availability of Wittgenstein’s Later Philosophy’] were
‘deleterious to the future of philosophy’. (Cavell 2010, 442)
INQUIRY
3
Cavell is slightly misremembering here. What Fodor and Katz actually
say is:
The position Cavell advocates in [the two articles] seems to us, however, to be
mistaken in every significant respect and to be pernicious both for an adequate
understanding of ordinary language philosophy and for an adequate understanding of ordinary language. (Fodor and Katz 1963, 57)
There were later defenses of Cavell (Henson 1965; Bates and Cohen
1972), and attempts to defend the methods of ordinary language philosophy against Mates and Fodor and Katz, whether or not such defenses would
be endorsed by Cavell (Searle 1969; Vendler 1967). And the Mates–Cavell
debate was reprinted in two volumes focusing on the methods of ordinary
language philosophy (Chappell 1964; Lyas 1971). But the debate eventually
fizzled out for whatever reasons philosophical debates end even when
the philosophical problems motivating the debate haven’t been resolved
(lack of interest, change of fashion). Recently, however, questions about the
proper methodology of philosophy of language and the relevance of formal
experiments to the investigation of meaning have been posed again by
twenty-first century advocates of experimental philosophy and their critics,
and by new advocates of ordinary language philosophy.2 Reexamining the
1957 debate permits an assessment of the success or failure of the original
arguments for and against ordinary language philosophy from a sufficiently
long historical distance to provide a new angle on the old debate.
2. Mates’ challenge
Mates wonders ‘how one would go about verifying’ the statements made
about language by ‘the so-called ordinary language philosophers’ (Mates
1958, 161). He takes as his target Ryle’s remarks on ‘the ordinary sense
of ‘voluntary’, [and] ‘involuntary’ (Ryle 1949, 59). Ryle’s discussion takes
place in the context of his attack on ‘The Will’ and the idea of ‘volitions’
understood as ‘special acts, or operations “in the mind” . . . which somehow
puts my muscles into action’, which are part of the ‘myth of the ghost in the
machine’ (50).
Ryle objects to the idea that intentional action involves a mental ‘volition’
partly on the grounds that ‘no one, save to endorse the theory, ever describes his own conduct, or that of his acquaintances, in the recommended
2 For examples of neo-ordinary language philosophy see Baz (2012), DeRose (2005) Fischer (2014), and
Laugier (2013). Hansen (2014) surveys varieties of contemporary ordinary language philosophy. For
comparisons of the methods of experimental philosophy and ordinary language philosophy, see Sandis
(2010) and Hansen and Chemla (2015). For historical context with a focus on ‘empirical semantics’, see
Murphy (2015).
4
N. HANSEN
idioms’ (51). He goes onto describe the way that ‘ordinary folk’ apply the
words ‘voluntary’ and ‘involuntary’, in contrast to the ‘stretched’ sense that
philosophers who propose a theory of volitions do, in which an action is
voluntary just in case it is caused by a volition, and involuntary just in case
it is not. Here is how Ryle characterizes the ordinary use of ‘voluntary’ and
‘involuntary’:
In their most ordinary employment ‘voluntary’ and ‘involuntary’ are used, with
a few minor elasticities, as adjectives applying to actions which ought not to be
done. We discuss whether someone’s action was voluntary or not only when the
action seems to have been his fault . . .
In this ordinary use, then, it is absurd to discuss whether satisfactory, correct or
admirable performances are voluntary or involuntary. (56)
Philosophers ‘stretch’ the ordinary sense of ‘voluntary’ and ‘involuntary’
when they maintain that those expressions are jointly exhaustive descriptions of human action, which leads to a ‘tangle of largely spurious problems,
known as the problem of the Freedom of the Will’ (57). In order to avoid that
problem, according to Ryle, we need to remind ourselves of the ordinary
uses of ‘voluntary’, ‘involuntary’, and related expressions, and see how they
do not require positing any unverifiable inner acts of will.3
How, Mates wonders, would we go about verifying whether Ryle is right
about the way ‘voluntary’ and ‘involuntary’ are ordinarily used? The most
obvious approach would be to look at how those expressions are in fact
used. Mates is doubtful that the ‘ordinary man’ only applies ‘voluntary’ and
‘involuntary’ to actions that ought not to be done, but he doesn’t undertake
such a study (Mates 1958, 163). He points out that even if it were the case
that the expressions are only applied to actions that ought not to be done,
that wouldn’t establish that doing so was an essential part of the meaning
of those expressions – it might be the case, for example, that ‘some other
factor, such as perhaps a disposition on the part of the ordinary man to talk
more about things of which he disapproves than about things of which he
approves’, could account for the fact (if it were a fact) that ‘voluntary’ and
‘involuntary’ are ordinarily used as Ryle describes.4
But Ryle thinks that the obvious approach to verifying statements about
ordinary use is ‘philosophically pointless’. He distinguishes what he calls
the ‘use’ of an expression and the ‘usage’ of that expression:
3 Austin (1962) argues in a similar way that philosophers have ‘stretched’ the meaning of ‘directly’ in the
phrase ‘directly perceive’ to the point of making the expression meaningless.
4 This worry is an early example of what has become a standard response to the arguments of the ordinary
language philosophers – offering a pragmatic explanation of a purportedly semantic phenomenon. See
Grice (1961) for another early example of this type of response.
INQUIRY
5
Usage . . . can be local or widespread, obsolete or current, rural or urban, vulgar or
academic. There cannot be a misusage any more than there can be a miscustom or
a misvogue. The methods of discovering usages are the methods of philologists.
(Ryle 1953, 174–175)
(. . . philologists, that is, and not philosophers.)
‘Use’, in contrast, is a ‘way of operating with . . . a word’, a ‘technique,
knack, or method’. It is possible to misuse an expression, indeed, it might
be the case that popular usage of an expression substantially diverges
from the use of that expression. The ‘appeal to prevalence [of usage] is
philosophically pointless, besides being philosophically risky’ (Ryle 1953,
177). Ryle therefore says he is interested in use, not usage.5 Mates glosses
Ryle’s methodology of ‘use’ over ‘usage’ as indicating that there is something ‘normative’ about ordinary use, where he understands ‘normative’ to
mean that a recommendation is being made, or that a ‘use is sanctioned by
authority’ (165):
[Ryle’s discussion] appears to indicate that for him there is some sort of normative
element in assertions about ordinary use. If the opposite of use is misuse, then
use must be somehow right, proper, or correct. (Mates 1958, 164)
But Mates dismisses the idea that there is a normative element in assertions about ordinary use for ordinary language philosophers, even though
Ryle says that there is: ‘Despite Ryle’s own explanations, I am reluctant to
believe that the expression “ordinary use” is really a normative term for the
ordinary-language philosophers’. Instead, Mates proposes that ‘ordinary
use’ is a ‘rough descriptive term’ (Mates 1958, 164) where the standards
for evaluating statements about ordinary use are those appropriate for
evaluating any other descriptive statement, including statements about,
e.g., whether some particular usage (in Ryle’s terms) of an expression
occurs more frequently than some other usage. And Mates argues that the
‘armchair’ methods of the ordinary language philosophers do not entitle
them to claim that language is ordinarily used in the way they claim it is
ordinarily used, on the grounds that ‘even relatively careful authors are
often not reliable reporters of their own linguistic behavior, let alone that
of others’ and ‘the intuitive findings of different people, even of different
experts, are often inconsistent’ (Mates 1958, 165).
As an example of expert disagreement about ordinary language, Mates
cites Austin discussing the expressions ‘involuntary/involuntarily’ in a way
that seems to contradict Ryle’s observations about the ordinary use of
5 Contrast Ryle’s view with Horace’s in Ars Poetica: ‘Many words will be born again that have now sunk into
oblivion, and many will die that are now held in respect, if that’s what usage chooses – usage, which has
the power over the judgment, the law, and the rule of speech’ (translation from Lynch 2009, 34).
6
N. HANSEN
‘voluntary’ and ‘involuntary’, as applying only to actions which ought not
to be done:
. . . for example, take ‘voluntarily’ and ‘involuntarily’: we may join the army or
make a gift voluntarily, we may hiccough or make a small gesture involuntarily
. . . . (Austin 1956–1957, 17)
So there is an apparent disagreement between Ryle and Austin, and, as
Mates puts it, ‘If agreement about usage cannot be reached within so
restricted a sample as the class of Oxford Professors of Philosophy, what are
the prospects when the sample is enlarged?’ (Mates 1958, 165).6
Mates also makes a positive proposal for how one should go about
verifying statements about the ordinary use of expressions. He discusses
two complementary approaches to verifying descriptive statements about
how words are used: the ‘extensional’ approach, and the ‘intensional’
approach.
In the extensional approach one observes a reasonably large class of cases in
which the subject applies the word, and then one ‘sees’ or ‘elicits’ the meaning
by finding what is common to these cases . . . In the [intensional approach], one
asks the subject what he means by the given word or how he uses it; then one
proceeds in Socratic fashion to test this first answer by confronting the subject
with counterexamples and borderline cases, and so on until the subject settles
down more or less permanently upon a definition or account. (Mates 1958, 165–
166)
According to Mates, both the extensional and intensional approaches are
equally relevant to verifying whether someone uses a word in a given way
or with a given sense, and there is no guarantee that the two approaches
yield the same results.7 It is possible, for example, to observe a speaker
using a word in a way that she says she would never use it. 8
6 Note, however, that Ryle and Austin are not, strictly speaking, disagreeing in the passages Mates cites
because Ryle and Austin are discussing different words: the adjectives ‘voluntary’ and ‘involuntary’ in
Ryle’s case, and the adverbs ‘voluntarily’ and ‘involuntarily’ in Austin’s case.
7 See Hampton and Passanisi (2016) for a defense of the divergence of intensions and extensions in a
contemporary theory of concepts.
8 See Labov (1975) and Labov (2006) for examples of mismatches between speakers’ judgments about how
they speak and how they do in fact speak. For example, Labov discusses Philadelphia speakers who say
they never use ‘anymore’ in positive sentences, to mean roughly ‘nowadays’, who then go onto use it
that way (e.g. ‘Do you know what’s a lousy show anymore? Johnny Carson’) (Labov 1975, 107). And New
Yorkers who were asked whether they pronounced ‘cards’ with [r] or without [r] (cahds) overestimated
how much of their own pronunciation was r-ful (Labov 2006, discussed in Gordon 2013, 69, from which
the following passage is drawn):
Labov recounts an especially poignant case of this phenomenon. He interviewed a middle-class
mother and her daughter who offered strong opinions about many of the phonological variables
and ridiculed speakers from the subjective reaction test, including one who dropped an [r] in one
word. They insisted that their own speech was impeccably r-ful. Labov ‘unwisely’ played back part
INQUIRY
7
Mates says that ordinary language philosophers ‘tend toward an armchair version of the extensional method’ of verifying statements about
language, and suggests that they neglect the intensional method. Even
if philosophers apply ‘voluntary’ and ‘involuntary’ in different ways than
ordinary folk, the two groups may give the same ‘intensional’ account of
the word’s meaning. So there may be evidence that pulls in the direction
of thinking that philosophers use ‘voluntary’ and ‘involuntary’ differently
than ordinary folk, and there may be equally significant evidence that pulls
in the opposite direction:
Thus, even if Professor Ryle had determined that ordinary folk in fact apply the
word ‘voluntary’ only to actions which ought not to be done, while philosophers
apply it to meritorious actions as well, he would be far from establishing that
philosophers and ordinary folk apply the word ‘voluntary’ in different ways, i.e.
attach different senses to it. (Mates 1958, 166)
Mates’ challenge to ordinary language philosophy can be summarized
as follows:
(1) Statements about language made by ordinary language philosophers are descriptive (evaluable as true or false), not normative (i.e.
not advice on how to use language).
(2) Skepticism about the ordinary language philosophers’ descriptive
claims about language is warranted by the fact that (i) speakers
are often not reliable reporters of their own linguistic behavior,
and (ii) there is disagreement even among practitioners of ordinary
language philosophy (Ryle and Austin, for example) about how expressions are used.
(3) There are two basic approaches to verifying descriptive statements
about language: the extensional approach, which looks at the use of
expressions, and the intensional approach, which involves eliciting
the beliefs of speakers about the meaning or use of expressions.
Ordinary language philosophers ‘tend toward an armchair version
of the extensional approach’, ignoring the intensional approach. The
two approaches may yield conflicting results, so relying only on one
method will not give a complete picture of ordinary use.
Before diving deep into Cavell’s defense of the methods of ordinary
language philosophy, I want to explore an alternative response to Mates’
of their interview to demonstrate that they regularly dropped [r] too. This left them ‘disheartened
in a way that was painful to see’. (Labov 2006, 314)
8
N. HANSEN
challenge that there are good reasons to reject. The failure of that defense
will then show what Cavell’s defense has to avoid in order to be successful.
3. How not to defend ordinary language philosophy: conventionalism and ‘legitimate use’
Both Searle (1969) and Vendler (1967) attempt to defend the idea of a priori
knowledge of language against a Mates-style challenge. Both attempt to
address point 2 in my reconstruction of Mates’ challenge, by giving reasons
to reject skepticism about the reliability of armchair claims about how
expressions are used. Both Searle and Vendler respond to that kind of
skepticism by drawing analogies between our knowledge of how to play
games (baseball in Searle’s case and chess in Vendler’s) and our knowledge
of how to use linguistic expressions.
Vendler explains how certain kinds of knowledge about chess don’t
involve conducting any empirical investigations:
Suppose that while watching a game of chess I see two Pawns of the same
color standing in the same column. Then I say: ‘One of them must have taken an
opposing piece in a previous move’. How do I know this? Is it sufficient to say that
in all chess games we ever witnessed this correlation held? No, given the rules of
the game, the relation holds a priori; the contrary is not something unusual or
unlikely: it is inconceivable. (Vendler 1967, 17)
The linguist, according to Vendler, is analogous to an observer of chess
who wants to describe its rules by observing what players of chess regard
as admissible moves. Discovering the existence of the rules is an empirical
project, but the rules lay down necessary requirements on movements
in the game, and given knowledge of the rules, one can derive a priori
knowledge about the possibility of particular configurations of pieces on
the board. For Vendler, speakers of a language are entitled to statements
about ordinary language on the same grounds that one who knows the
rules of chess is entitled to the claim, made of two pawns of the same
color standing in the same column that ‘one of them must have taken
an opposing piece in a previous move’. The entitlement is not based on
observation, but derives from the speaker’s knowledge of the rules of
the language, which can be combined to derive knowledge of unobvious
conclusions.
Similarly, Searle 1969 writes that his knowledge of particular facts about
the language he speaks is analogous to his knowledge that after hitting a
fair ball in baseball, ‘the batter runs in the direction of first base, and not
in the direction, say, of third base or the left field grand stand’ (14). Searle
INQUIRY
9
asks, ‘Now what sort of knowledge is this? On what is it based? How is it
possible?’ He answers:
My knowledge is based on knowing how to play baseball, which is inter alia
having internalized a set of rules. I wish to suggest that my knowledge of linguistic
characterizations is of a similar kind.
If this is correct, then the answer to the philosopher’s question, ‘What would
we say if. . . ?’ is not a prediction about future verbal behavior but a hypothetical
statement of intention within a system of rules, where mastery of the rules dictates
the answer . . .
The ‘game-like’ account of how one knows statements about one’s native
language without having to observe any linguistic behavior goes as follows:
(1) A speaker of a language knows a set of rules governing the language.
(2) One can derive particular pieces of a priori knowledge about the
language by applying those rules.
The ‘game-like’ explanation of a priori linguistic knowledge proposed by
Searle and Vendler is not radical; in fact it is a way of stating one of the basic
commitments of standard compositional theories of semantics. Speakers
of a language are able to derive the meaning of an infinite number of novel
sentences from their knowledge of a finite stock of lexical items and rules
for assembling those items into complex wholes. Such an account is meant
to explain how speakers know the meaning of sentences they have never
observed in use before; in that sense the standard semantic picture is a
story about a form of a priori linguistic knowledge.
While this ‘game-like’ explanation of how speakers can acquire knowledge of unobvious linguistic facts about the language they speak without
leaving their armchairs looks promising, Jackman (2001) argues that such
an account doesn’t do the work the ordinary language philosopher needs
it to do (Jackman calls this account ‘conventionalist’, and I will adopt his
terminology). The first prong of Jackman’s attack involves driving a wedge
between a speaker’s a priori knowledge of her own idiolect and knowledge
that anyone else speaks the same language. Jackman cites Searle saying
‘That my idiolect matches a given dialect group is indeed an empirical
hypothesis (for which I have a lifetime of ‘evidence’), but the truth that in
my idiolect ‘oculist’ means eye doctor is not refuted by evidence concerning
the behavior of others’ (Searle 1969, 13). The conventionalist story of how a
priori linguistic knowledge (or knowledge of the rules of a game) is possible
explains only how a speaker can have a priori knowledge of her own idiolect
(or the rules of the game that she is following), but any move beyond the
individual to knowledge of linguistic facts about a community of speakers
10
N. HANSEN
(how ‘we’ ordinarily use an expression) involves an ‘empirical hypothesis’.
The conventionalist therefore falls short of justifying statements of the sort
that interest the ordinary language philosopher, which concern what we
mean when we say certain things.
The second prong of Jackman’s attack on the conventionalist account
of a priori linguistic knowledge targets the assumption that speakers have
a priori knowledge even of their own idiolect. Drawing on examples from
Putnam (1975) and Burge (1979), Jackman argues that ‘a speaker’s discriminatory capacities often underdetermine or even misidentify what one is
talking about’ (318). For example, Burge describes a situation in which a
competent speaker consistently misapplies the expression ‘arthritis’, which
reveals the possibility that speakers may have internalized mistaken linguistic rules. Recognizing the possibility of linguistic ignorance of the Putnam
and Burge varieties ‘comes with a corresponding uncertainty about the
mastery of one’s own language’ (319).
The two prongs of Jackman’s attack amount to a dilemma for someone
who wants to defend the idea that we have a priori entitlement to claims
about what ‘we’ mean:
(1) If speakers have a priori knowledge of what the expressions of their
language mean, then they aren’t entitled to generalize from their
own language to a public language.
(2) If the language that speakers know is shared, then they cannot have
a priori knowledge of the meaning of expressions of the language.
The upshot of the dilemma is that there cannot be a priori knowledge that
one’s language (or the rules one is playing by) is shared – but that is just
what it seems ordinary language philosophers need in order to respond to
Mates’ challenge.
Some recent defenders of the methods of ordinary language philosophy
rely on versions of the conventionalist response to Mates that sidestep
worries about idiolects by assuming that certain practitioners of ordinary
language philosophy have access to ‘objective facts about linguistic norms’.
For example, Sandis (2010) distinguishes ordinary language philosophy
from recent experimental investigations of topics of philosophical interest
by claiming that ordinary language philosophy makes observations about
‘proper linguistic usage’ and ‘legitimate use’:
OLP [ordinary language philosophy], by contrast [with experimental philosophy
of language], is not interested in what the majority of people happen to think at
any given time and place but, rather, in objective facts about linguistic norms.
(185)
INQUIRY
11
When OLP talks of what we would say, it refers not to the latest up-to-the-minute
statistics about what happens to be the case but to well-established, legitimate
use, distinguishing further between paradigm and peripheral cases. (186)
The invocation of ‘correct’ or ‘legitimate’ use (or simply ‘the use’ – see
Hanfling (2000)) recalls Ryle’s distinction between use and usage, where use
is explicitly normative, while usage is not. When he considers this aspect of
Ryle’s view, Mates dismisses it, and says, ‘What authority deems it wrong
to use the word ‘voluntary’ as the philosopher does [as opposed to the
ordinary speaker]?’ (Mates 1958, 164) That is a centrally important question
for the defender of ordinary language philosophy who invokes ‘correct’
or ‘legitimate’ use. It is a question that Cavell attempts to answer, and his
answer is strikingly different than the conventionalist response to Mates’
challenge. I think V.C. Chappell (1964) is still correct (more than 50 years
later) in his assessment that Cavell gives ‘the most detailed explanation and
defense of the procedures of ordinary language philosophers that has yet
appeared’. In the remainder of this paper, I will describe Cavell’s response
to Mates’ attack and show how it avoids the problems that arise for the
conventionalist defense of ordinary language philosophy.
4. Cavell’s defense of the methods of ordinary language philosophy in ‘Must We Mean What We Say?’
As Bates and Cohen (1972) observe, explicating Cavell ‘is no picnic’, but
the basic outlines of what Cavell argues in his (1958) defense of ordinary
language philosophy are clear: He denies point 1 in Mates’ challenge, by
arguing that there is a normative element to statements about ordinary
use, he disputes point 2 by arguing that native speakers of a language are
authoritative regarding ordinary uses of expressions in that language, and
he claims that Mates’ discussion of intensional and extensional methods is
‘irrelevant’ to the concerns of ordinary language philosophy as he defends
it (Cavell 1958, 131).
4.1. On the reliability of judgments about ordinary language
Cavell distinguishes two types of statements about ordinary language:
• Type 1: ‘Statements which produce instances of what is said in a
language (‘We do say . . . but we don’t say – ’)’
• Type 2: Instances of what is said in a language ‘accompanied by
explications – statements which make explicit what is implied when
12
N. HANSEN
we say what statements of the first type instance us as saying (“When
we say . . . we imply (suggest, say) –”)’.
The clash between Austin and Ryle involves statements of different
types: Cavell says Austin produces a statement of type 1 (‘we may make a
gift voluntarily’) while Ryle makes a claim of type 2, about what ‘voluntary’
and ‘involuntary’ mean – that when we describe an action as voluntary or
involuntary, we imply that the action ought not to be done.9 Cavell allows
that Ryle’s type 2 statement is wrong, but he rejects Mates’ charge that
Ryle lacks good evidence for making such a statement (whether or not it
is correct). Ryle has good evidence for his claim simply in virtue of being a
native speaker of English:
Such speakers do not, in general, need evidence for what is said in the language;
they are the source of such evidence . . .
[F]or a native speaker to say what, in ordinary circumstances, is said when, no
such special information is needed or claimed. All that is needed is the truth of
the proposition that a natural language is what native speakers of that language
speak. (174–176)
Cavell is right that it would be absurd to think that native speakers of a
language need ‘special information’ or have to ‘count noses’ before they
know how to use some expression in the language they speak. There have
to be native speakers saying things for any data about ordinary use to exist.
But it’s a different question when speakers go beyond simply speaking
their language to make meta-linguistic claims about how expressions are
ordinarily used. Fodor and Katz (1963 60) criticize Cavell for missing the
distinction between using an expression and making statements about
how that expression is used:
What Cavell misses is the distinction between what a native speaker says (the
utterances he produces in the course of speaking) and what he says about what
he and other native speakers say (the metalinguistic comments he makes when
the reflective mood is upon him). There can be no doubt but that most (though
definitely not all) of the utterances of a native speaker are utterances of the
speaker’s language. This truth is guaranteed by the truism that a natural language
is what a native speaker of that language speaks. However, the statements that a
native speaker makes about his language, his metalinguistic claims, need not be
true in order for the linguist to have noses to count.
Fodor and Katz overextend their criticism of Cavell here – though a
restricted version of their objection is important. It is standard practice in
9 Cavell misdescribes Austin’s remark, which is not metalinguistic – it doesn’t mention ‘what we say’, it
simply produces an example of something that does make sense. The fact that it makes sense is evidence
against Ryle’s Type-2 statement, however, in the way Cavell describes.
INQUIRY
13
Figure 1. Every boy that stands next to a girl holds her hand.
linguistics and philosophy of language to make use of meta-linguistic judgments of speakers as evidence for and against theoretical hypotheses.10 In
syntax, theorists make use of speakers’ meta-linguistic judgments about
the acceptability of various constructions, like the following11 :
Please indicate which of the following two sentences sounds more acceptable:
(1) Ginny remembered to have brought the beer.
(2) Ginny remembered to bring the beer. (Sprouse, Schütze, and Almeida,
2013, 237)
And in semantics and pragmatics, speakers are often asked to make truth
value judgments about claims made in various contexts, or judgments
about entailment relations. For example, as part of the evaluation of the
semantics of ‘donkey sentences’, Geurts (2002 135–136) asked participants
to judge whether the statement ‘Every boy that stands next to a girl holds
her hand’ is a true description of the situation depicted in Figure 1.
The use of speakers’ meta-linguistic judgments about truth and acceptability reflects a central assumption of contemporary research in the
language sciences, namely that such judgments are products of speakers’ underlying (tacit) linguistic knowledge. While speakers’ meta-linguistic
judgments about acceptability and truth are widely regarded as useful tools
in the investigation of syntax, meaning, and use, it is a further, much more
contentious question whether individual speakers’ judgments about how
a group of speakers would use a particular expression are reliable.
Mates is skeptical that ordinary speakers are reliable when asked to make
these kinds of judgments. Only recently have there been any systematic
10 This is pointed out, in response to Fodor and Katz, by Bates and Cohen (1972).
11 ‘The data syntacticians predominately appeal to are speaker/hearers’ intuitive judgments of acceptability
and interpretability (suffice it to say, any relevant data are admissible; intuitions are simply readily
available, not obviously misleading, and have no serious competitor as a data source)’ (Collins 2007,
887). For a summary of discussions of the role of meta-linguistic intuitions in syntax, see Sprouse (2013).
14
N. HANSEN
examinations that bear on whether ordinary speakers are reliable judges
about other speakers’ linguistic behavior. The studies that have been conducted have yielded results that, while they do not support general skepticism about the reliability of group-directed meta-linguistic judgments, also
do not support Cavell’s claim that the ordinary language philosopher has
‘good evidence’ for statements of type 1 or type 2 simply in virtue of being
a native speaker of English.
Experimental investigations in both linguistics and philosophy have
indicated that meta-linguistic judgments about acceptability made by ordinary speakers and those made by experts are highly correlated with one
another, which is evidence that generalizations from one’s own judgment
to judgments of the group are reliable (Culbertson and Gross 2009; Myers
2009; Sprouse, Schütze, and Almeida 2013). These findings might be taken
to support the ordinary language philosophers’ method of relying on
armchair meta-linguistic judgments, since (in the acceptability judgment
tasks that have been experimentally examined) those armchair judgments
correlate with ordinary speakers’ judgments in experimental conditions.
But it is a further question whether ordinary and expert meta-linguistic
judgments about semantic and pragmatic facts, like what particular names
or natural kind terms refer to, correlate in the same way. And there is
evidence from experimental investigations that, in certain cases, there
is substantial interpersonal variation in meta-linguistic judgments about
semantic and pragmatic phenomena.12
In summary, given the existing experimental evidence, there is reason
to think that armchair meta-linguistic judgments about certain linguistic
properties (acceptability) reliably generalize, while armchair meta-linguistic
judgments about other properties (reference, satisfaction conditions for
adjectives) may not. Overall, that is bad news for both general skepticism
about the reliability of armchair metalinguistic judgments and for Cavell’s
optimism about the reliability of such judgments. It will take much more
experimental work to determine the exact contours of ordinary speakers’
ability to make reliable judgments about the linguistic behavior of groups
they belong to.
12 For evidence concerning proper names, see the debate surrounding Machery et al. (2004). Genone (2012)
and Hansen (2015) provide summaries of that debate. For evidence concerning natural kind terms, see
Braisby, Franks, and Hampton (1996), Jylkkä, Railo, and Haukioja (2009) Genone and Lombrozo (2012),
and Nichols, Ángel Pinillos, and Mallon (2016). Hansen (2015) provides a summary of the experimental
evidence for natural kind terms. For related evidence concerning adjectives, see Hansen and Chemla
(2017).
INQUIRY
15
4.2. Categorial declaratives
Cavell’s response to Mates’ challenge that invokes the fact that native
speakers of a language are reliable judges of how words are ordinarily
used does not stand up to scrutiny. But there is a different line of response
suggested by Cavell that isn’t subject to the same worries. I will develop
that response in the remainder of this paper.
Cavell (1958) is centrally concerned with the relation between what
someone says and what she means by what she says. The significance
of the question in the title of the essay, ‘Must We Mean What We Say?’, is
whether there is any logical (or otherwise necessary) connection between
‘what you (explicitly) say and what saying it implies or suggests’ (196). Cavell
revises Ryle’s type 2 statement about ‘voluntary’ to accommodate Austin’s
observations, and produces the following statement, which he refers to as
‘S’:
S ‘When we ask whether an action is voluntary we imply that the action
is fishy’
Cavell says that ‘Something important will be learned if we realize that
we do not know what kind of assertion S is’ (181). S is ‘obviously’ not
analytic, Cavell says, because ‘it is perfectly true that ‘voluntary’ does not
mean (you will not find set beside it in a dictionary) ‘fishy’)’ (181). And
yet Cavell expresses the feeling that S expresses a necessary truth, and he
assumes that if a statement expresses a necessary truth then it has to be
known a priori. That combination of necessity and a prioricity with nonanalyticity is what poses the problem Cavell tries to solve:
When (if) you feel that S is necessarily true, that it is a priori, you will have to
explain how a statement which is obviously not analytic can be true a priori.
(181)13
It is important to note that Cavell is writing at a time before widespread
acknowledgement of two conceptual distinctions that became very important to analytic philosophy in the latter half of the 20th century:
• Necessary truths can be known a posteriori.
• There is a distinction between what a statement necessarily implies in
virtue of its (context-invariant) meaning and what making a statement
implicates, which is context dependent.
13 Cavell was teaching the Critique of Pure Reason at Berkeley around the time that he was writing ‘Must We
Mean What We Say?’ (Cavell, 2010, 277).
16
N. HANSEN
Acknowledging either distinction would help explain what kind of assertion
S is. First, supposing that S does express a necessary truth, it need not
be knowable a priori. Informative identity statements, like ‘Hesperus is
Phosphorus’ are necessarily true, but can be a posteriori discoveries (Kripke,
1980, Lecture II). Even if Cavell were right that S expresses a necessary truth,
it could be discovered via an empirical scientific investigation of language.
Second, the connection between what is explicitly said and what is
suggested or conveyed by what is said is of central importance to the
ordinary language philosophers who want to argue that philosophers
are ‘stretching’ or distorting the ordinary meaning of expressions. Ryle’s
discussion of the way philosophers distort the meaning of ‘voluntary’ and
‘involuntary’, for example, relies on the idea that there is some regular
connection between those expressions and the actions they describe being
something that someone ought not to do. That is, Ryle thinks that saying
that some action is voluntary will convey that the action ought not to be
done. When philosophers wonder whether actions of a certain type are
voluntary or not when there is no question of whether or not the action
ought not to be done, then they are distorting the ordinary meaning of
‘voluntary’.
Cavell calls the connection between what is explicitly said and what is
suggested or conveyed by what is said a ‘necessary’ connection.14 But that
is to overstate the kind of regularity that exists between what is explicitly
said and what is conveyed or suggested by what is explicitly said. Consider
Cavell’s own example of a ‘necessary’ truth, namely S. It is doubtful that
S expresses a necessary truth, because there are contexts in which one
can ask whether an action is ‘voluntary’ without suggesting that there is
anything fishy about the action (bracketing the fact that it’s not really clear
what a ‘fishy’ action would be). Consider the following questions, which
are based on naturally occurring sentences obtained through the Corpus
of Contemporary American English (COCA):
(3) Was the contraception proposed by the Gates Foundation as part of
the Family Planning 2020 initiative going to be voluntary? (http://
goo.gl/gkzpHH)
(4) Was the administrative fee taken out during the first three years of
employment voluntary? (http://goo.gl/axQqCp)
It is unclear what the ‘fishy’ suggestion either of these questions would
be. The first question concerns whether the contraception is voluntary, as
14 See Mulhall (1999, 13–14) for further discussion.
INQUIRY
17
opposed to mandatory, and the second question concerns whether the
administrative fee can be waived by the employee or not.
A better way of capturing the relation between what is said and what is
conveyed by what is said is introduced by Grice’s (1975) ‘logic of conversation’, which wouldn’t have been available to Cavell at the time of writing
‘Must We Mean What We Say?’.15 Conversational implicatures arise out of a
combination of the content of what is explicitly said and the ‘cooperative
principle’ – a governing principle of conversation. For example, if I ask
whether some action is voluntary, I implicate (though I do not explicitly
say) that I don’t know whether the action is voluntary. Assuming that I am
making a cooperative contribution to the conversation, my audience can
(with reason) infer that I don’t know whether the action was voluntary or
not. That they draw that inference is something that I can expect if I understand the norms of conversation. I can exploit this expectation by asking
whether some action was voluntary when I know that it was voluntary; that
will generate the false suggestion that I don’t know whether the action
was voluntary. It is the norms governing cooperative conversations, rather
than anything about the meaning of ‘voluntary’, that generate expectations
regarding the question. Expectations about what I have implicated are
not completely under the speaker’s control, but they also do not reflect
necessary connections between what is explicitly said and what is conveyed
or suggested. To the contrary: conversational implicatures differ from the
context-invariant implications associated with what is explicitly said in that
they are context-dependent.
The unavailability of those two conceptual distinctions constrains the
range of options open to Cavell in his discussion of statements about
ordinary language, and he recognizes that the conceptual tools available
to him are inadequate to his task:
At this point the argument has become aporetic. ‘Statements about ordinary
language’ . . . are not analytic, and they are not (it would be misleading to call
them) synthetic (just like that). Nor do we know whether to say they are a priori,
or whether to account for their air of necessity as a dialectical illusion, due more
to the motion of our argument than to their own nature. Given our current
alternatives, there is no way to classify such statements; we do not yet know what
they are. (184)
15 The lectures that were the basis of Grice (1975), ‘Logic and Conversation’, were delivered in 1967. See
Grice (1975, 41). It is very doubtful that Cavell himself would endorse this way of characterizing the
relation between what is said and what is conveyed by what is said.
18
N. HANSEN
Even given these limitations, part of Cavell’s proposal for understanding
‘statements about ordinary language’ is worth salvaging. In the remaining
part of this section I will spell out what I think that part is.
Cavell begins his discussion of how it is possible to be entitled to
make S-type statements without conducting empirical observations by
distinguishing S from a similar-sounding statement T, which he says ‘does
seem obviously synthetic’:
T ‘Is X voluntary?’ implies X is fishy.
He says:
But S and T, though they are true together and false together, are not everywhere
interchangeable; the identical state of affairs is described by both, but a person
who may be entitled to say T, may not be entitled to say S. Only a native speaker
of English is entitled to the statement S, where as a linguist describing English
may, though he is not a native speaker of English, be entitled to T. What entitles
him to T is his having gathered a certain amount and kind of evidence in its favor.
But the person entitled to S is not entitled to that statement for the same reason.
He needs no evidence for it. . . The question of evidence is irrelevant. (182)
The interesting suggestion in this passage is that one can be entitled to
make a statement in virtue of having the right kind of status as a member
of a certain group – in this case, the group of native speakers of English.
S, unlike T, uses the first person plural ‘we’, which presupposes that the
speaker is included in the group (or ‘plural individual’) denoted by ‘we’
(Schlenker, 2003, 5). If the presupposition is satisfied, it can be the case that
the speaker is entitled to ‘speak for the group’, and introduce, or reinforce,
a rule.
Cavell observes that statements that share the same surface form as
type-T (or type-S) statements can be used to perform a different kind of
speech act16 – instead of purporting to state a fact, they can be used
to introduce a rule, ‘a way of doing or saying something which is to be
followed’:
Whether remarks . . . ‘about’ ordinary language, and equally about ordinary
actions – are statements or rules depends on how they are taken: if they are
taken to state facts and are supposed to be believed, they are statements; if they
are taken as guides and supposed to be followed, they are rules. (184)
Statements about what ‘we’ do can play this role: ‘[I]f someone is tempted
not to fulfill a promise, you may say, ‘Promises are kept’, or ‘We keep
our promises (that is the sort of thing a promise is)’, thus employing a
16 Bates and Cohen (1972, 11) observe that ‘Cavell is concerned with something like speech acts [in his
discussion of categorial declaratives]’.
INQUIRY
19
rule-description – what I have called a categorial declarative’ (195). A
categorial declarative would not be challenged by citing evidence that
the group referred to by ‘we’ does not always conform to the proposed
rule. Consider the following sentence:
(5) We send thank-you cards in this family.
The utterer of (5), understood as a categorial declarative, is not entitled to
it on the grounds of observing the past behavior of her family. It may even
turn out that members of the family have not usually sent thank-you cards.
But that fact does not mean that the speaker is not entitled to utter (5),
and pointing it out would not constitute a refutation of (5). Understood as
a categorial declarative, (5) could be challenged in the following way:
(6) That’s not for you to say.
The entitlement to the utterance (5) in the situation described does not
stem from observation, but from occupying a certain position of authority in
the group spoken for. As Cavell says about statements of type S, understood
as categorial declaratives, the question of evidence is irrelevant for one’s
entitlement to assert (5).
The parallel between (5) and the case of claims about what ‘we’ say is
that speakers can occupy an analogous position of authority among fellow
speakers of a language. Statements like S can be used to perform two
different kinds of speech act. Critics of ordinary language philosophy (Mates
and Fodor and Katz, for example) have focused exclusively on the act of
describing, and wonder where the entitlement for such a claim comes from
if not from observation of actual linguistic behavior. But Cavell’s central
claim in ‘Must We Mean What We Say?’, as I understand it, is that the
rule-giving aspect of statements like S has been overlooked. Entitlement
to the rule-giving kind of speech act is derived from occupying a position
of suitable authority with respect to those included in the group referred
to with ‘we’, and challenging that kind of speech act requires a different
kind of challenge than presenting evidence that ‘we’ do not in fact act
as the speaker seems to be saying that we do. Challenging the speaker’s
entitlement to the rule-giving aspect of the categorial declarative involves
contesting the speaker’s entitlement to speak on behalf of the relevant
group (as in [6]).
If this way of understanding Cavell’s proposal is correct, it would provide
a way of responding to Mates’ challenge that doesn’t run into objections
from the spotty reliability of armchair meta-linguistic judgments about how
we speak. Recall that Mates rejects the idea that the statements of ordinary
20
N. HANSEN
language philosophers about language are ‘normative’, even though Ryle
distinguishes what he calls ‘use’ and ‘misuse’ from the descriptive patterns
of ‘usage’. Mates’ criticisms are directed at the statements of ordinary
language philosophers on the assumption that they are (non-normative)
descriptions. But if my interpretation of Cavell is correct, then Cavell should
be seen as rejecting Mates’ starting assumption and insisting that there is a
central normative component to the statements of the ordinary language
philosophers – statements about what we say are not just descriptions,
they are proposals as to how words should be used.17
4.3. Aesthetic judgments and claims about ordinary language
In ‘Must We Mean What We Say?’, Cavell does not discuss in detail what
would entitle someone to make S-type statements about ordinary language. He says ‘An examination of what does entitle a person to the
statement S would be required in any full account of such statements. Such
an examination is out of the question here’ (Cavell 1958, 182). But he returns
to the topic and enriches his discussion with a new metaphor in ‘Aesthetic
Problems of Modern Philosophy’ (Cavell 1976a, originally published in
1965). His new way of thinking about ‘the characteristic claim. . . of ordinary
language philosophers’ is to compare them with aesthetic judgments:
I will suggest that the aesthetic judgment models the sort of claim entered
by these philosophers, and that the familiar lack of conclusiveness in aesthetic
argument, rather than showing up an irrationality, shows the kind of rationality it
has, and needs. (Cavell, 1976a, 86)
Cavell uses different patterns of disagreement and reason-giving to
illustrate the difference between two types of aesthetic judgment – a mere
judgment of ‘sense’ vs. a judgment of ‘reflection’ – and a non-aesthetic
judgment (whether there is a goldfinch in the garden). The patterns of
disagreement that distinguish the two types of aesthetic judgment go as
follows:
(1) A: Canary wine is pleasant.
B: How can you say that? It tastes like canary droppings.
A: Well, I like it.
17 Some early commentators on ordinary language philosophy detected the normative component in
statements about ordinary language, but without developing the idea in any detail: ‘We strongly suspect
that cases of putative ordinary language analysis are, in fact, disguised reformations’ (Maxwell and Feigl
1961, 489); ‘There are innumerable cases in the philosophy of language where we realize that what we
off hand may have interpreted to be a language hypothesis is. . . more readily understood as a verbal
recommendation, as a convention, or as any other type of normative statement, say a proposal for how
to use a given linguistic expression . . . ’ (Tennessen 1962, 507).
INQUIRY
21
(2) A: He plays beautifully, doesn’t he?
B: Yes, too beautifully. Beethoven is not Chopin.
Or he [B] may answer:
B: How can you say that? There was no line, no structure, no idea what
the music was about. He’s simply an impressive colorist (91).
Whereas it is unremarkable to defend a mere judgment of sense by
responding to a challenge with ‘Well, I like it’ (as in dialogue 1), it would
be ‘feeble’ to respond to the challenge in dialogue 2 by saying ‘Well, I
liked it’. Moreover, while A doesn’t have to engage with B by giving some
reasons to defend the assessment of the beautiful playing, if he doesn’t,
Cavell says ‘there is a price he will have to pay in our estimate of him’ (92).
That is a difference in the obligations that one undertakes when making a
judgment of reflection rather than a mere judgment of sense. Cavell then
uses the same strategy to distinguish both types of aesthetic judgment
from a judgment about a (more) straightforward matter of fact – whether
there is a goldfinch in the garden:
A:
B:
A:
B:
A:
There is a goldfinch in the garden.
How do you know?
From the color of its head.
But goldcrests also have heads that color.
Well, I think it’s a goldfinch (it’s a goldfinch to me).
In the goldfinch discussion, it’s not open to A to retreat to a personal
judgment without giving up his claim to be a ‘competent interlocutor in
matters of knowledge (about birds?)’ (92).
In addition to the different patterns of challenge and retraction that characterize judgments of reflection, Cavell ties the success of such judgments
not to their accuracy in stating the facts, but to their ability to convince the
audience to see and acknowledge the relevant fact. The story of Sancho
Panza’s relatives Hume relates in ‘Of the Standard of Taste’ is used to
illustrate this criterion of success: they are ridiculed for their evaluation
of the quality of a supposedly excellent hogshead of wine on the grounds
that one says it tastes of iron and the other of leather.18 But when the
barrel is empty, a rusty iron key on a leather thong is found at the bottom,
apparently vindicating their judgments. Cavell’s reading of this episode is
18 Presumably those were not accepted terms for describing the taste of quality wine in 17th century Spain;
maybe a more contemporary equivalent would be saying that a wine tastes of ‘too many tramlines’. See
the ‘Winesmanship’ section of Potter (2005) for discussion.
22
N. HANSEN
that it is an example of the wrong model of understanding how aesthetic
judgments are vindicated:
It dissociates the exercise of taste from the discipline of accounting for it: but
all that makes the critic’s expression of taste worth more than another man’s
is his ability to produce for himself the thong and key of his response; and his
vindication comes not from his pointing out that it is, or was, in the barrel, but in
getting us to taste it there. (87)
How should this model be applied to S-type statements about ordinary
language? It gives a fuller picture of the thin nature of authority that is
required for someone to be entitled to S-type statements about ordinary
language that remains underdescribed in ‘Must We Mean What We Say?’.
While the authority that is required to make such statements consists in
merely being a speaker of the relevant language, the success conditions
of such statements involve a very demanding, and fickle, component:
successful categorial declaratives require getting one’s audience to acknowledge their correctness by adopting the proposed rule of use. Those
elements clearly distinguish categorial declaratives from statements about
ordinary use made by linguists: entitlement to such statements does require
possessing evidence of how people do in fact use language, and the success
of such statements does not require the audience to embrace any rule of
use.
4.4. Austin at criticism
For Cavell, Austin represents the ‘purest version of ‘ordinary language philosophy’ (Cavell 1976b, 98), and he compares Austin’s practice of drawing
fine distinctions and making illuminating comparisons to the methods of
an art critic:
The positive purpose in Austin’s distinctions resembles the art critic’s purpose
in comparing and distinguishing works of art, namely, that in this crosslight the
capacities and salience of an individual object are brought to attention and focus.
(Cavell 1976b, 103)
Examining how Austin distinguishes and compares the meaning of
expressions will make Cavell’s comparison of Austin to an art critic – and
more generally, his comparison of statements about ordinary language to
aesthetic judgments – more concrete.
Austin’s most famous lexical distinction concerns the expressions ‘mistake’ and ‘accident’. He says that the choice to apply one or the other of
these expressions to describe a situation can ‘appear indifferent . . . Yet a
story or two, and everybody will not merely agree that they are completely
INQUIRY
23
different, but even discover for himself what the difference is and what
each means’ (Austin 1956–1957, 10–11). To distinguish the meaning of
these expressions, Austin tells two contrasting ‘donkey stories’, which make
it seem clear that ‘by mistake’ better describes the action in one situation,
and ‘by accident’ better describes the action in the other, thereby providing
evidence that the meanings of the two expressions are indeed distinct:
You have a donkey, so have I, and they graze in the same field. The day comes
when I conceive a dislike for mine. I go to shoot it, draw a bead on it, fire: the
brute falls in its tracks. I inspect the victim, and find to my horror that it is your
donkey. I appear on your doorstep with the remains and say – what? ‘I say, old
sport, I’m awfully sorry, &c., I’ve shot your donkey by accident?’ Or ‘by mistake’?
Then again, I go to shoot my donkey as before, draw a bead on it, fire – but as
I do so, the beasts move, and to my horror yours falls. Again the scene on the
doorstep – what do I say? ‘By mistake’? Or ‘by accident’? (Austin 1956–1957, 11
n. 4)
Those who read Austin’s donkey stories tend to reach the same conclusion
about which term to apply in which situation: ‘by mistake’ better suits the
first story, and ‘by accident’ the second.19 Backed by the donkey stories, we
can issue two S-type statements that have achieved the kind of vindication,
in their acknowledgment by those who read Austin’s donkey stories, that
Cavell says is an aim of aesthetic judgments:
(7) In the first story, we say you shot your neighbor’s donkey by mistake.
(8) In the second story, we say you shot your neighbor’s donkey by
accident.
A comparison with a less successful application of Austin’s method
shows, by way of contrast, how clearly the donkey stories compel us
to acknowledge a difference in meaning between doing something ‘by
mistake’ and ‘by accident’. The go-cart story from Austin (1966) is intended
to distinguish the meaning of ‘intentionally’ and ‘deliberately’:
I am summoned to quell a riot in India. Speed is imperative. My mind runs on the
action to be taken five miles down the road at the Residency. As I set off down
the drive, my cookboy’s child’s new go-cart, the apple of her eye, is right across
the road. I realize I could stop, get out, and move it, but to hell with that: I must
push on. It’s too bad, that’s all: I drive right over it and am on my way. In this case, a
snap decision is taken on what is essentially an incidental matter. I did drive over
the go-cart deliberately, but not intentionally – nor, of course, unintentionally
either. It was never part of my intention to drive over the go-cart. At no time did I
intend to drive over it. It was incidental to anything I intended to do, which was
simply to get to the scene of the riot in order to quell it. However ‘odd’ it may
19 See, for example, Gustafsson (2005, 368), and Hanfling (2000, 64).
24
N. HANSEN
sound, I feel little doubt that we should say here that we did run over the go-cart
deliberately and that we should not care to say we ran over it intentionally. We
never intended to run over it. (432)
One crucial difference between this story and the donkey stories is that
Austin provides a ‘gloss’ on how to describe the action of driving over the
go-cart: he says that it was done deliberately, but not intentionally, because
‘we never intended to run over it’. Prompted by the feeling that this story
didn’t compel agreement to the same degree as the donkey stories, and the
fact that it has been argued that doing something intentionally does not
require intending to do it (see Harman 1976), Hansen and Chemla (2015)
constructed a formal experiment in which the go-cart story was presented
to experimental participants in three conditions: (i) with Austin’s original
gloss, (ii) with a reversed gloss that made the case that the go-cart was
driven over intentionally, but not deliberately, and (iii) with no gloss either
way. In Hansen and Chemla’s experiment, it was easy to reverse participants’ judgments about the go-cart story by reversing the glosses, and
in the ‘no-gloss’ condition, where participants were asked to say whether
‘deliberately’ or ‘intentionally’ better described the action of driving over
the go-cart, no preference was observed either way. (Figure 1, Hansen and
Chemla 2015). The experiment lends support to the thought that the story
itself (without a gloss) is insufficient to generate agreement with Austin’s
judgment that we should say that we ran over the go-cart deliberately, but
not intentionally.
If the criteria of success for an S-type statement of ‘what we say’ were
the same as those for an observation made by an empirically informed
linguist, then Austin’s remarks regarding how we use ‘deliberately’ and
‘intentionally’ would simply be incorrect. The go-cart experiment in Hansen
and Chemla (2015) did not find any evidence that English speakers have
a preference regarding whether ‘deliberately’ and ‘intentionally’ better
describes the action of running over the go-cart. But if we follow Cavell in
his analogy between S-type statements and aesthetic judgments, Austin’s
gloss is not a detachable component of the case for why we (should)
describe the story in the way Austin suggests. In his reading of the go-cart
story, Austin makes the case that we should describe the action of running
over the go-cart as done deliberately, but not intentionally, because we did
not intend to run it over. As mentioned above, whether that connection
exists has been contested. But the important point for assessing Austin’s
methodology is that in his account of what ‘we should say’ or ‘should not
care to say’ about the case, he is proposing that ‘deliberately’ is a better
term to use in describing the action than ‘intentionally’ – whether or not
INQUIRY
25
speakers would prefer to use one term or the other before adopting his
proposal.
4.5. Categorial declaratives and meta-linguistic negotiation
Cavell’s analogy between categorial declaratives (S-type statements) and
aesthetic judgments has provided a sketch of what is required for one to
be entitled to make such statements (merely being a speaker of the language) and what is required for such statements to be successful (audience
endorsement of the proposed use). Our understanding of how categorial
declaratives operate can be enriched by considering recent discussions in
philosophy of language of how what look like straightforward statements of
fact can be playing a normative, meta-linguistic role (Sundell 2010; Plunkett
and Sundell 2013, 2014).
Barker (2002, 1–2) describes a ‘metalinguistic use’ of ‘tall’:
Normally, (9) will be used in order to add to the common ground new information
concerning Feynman’s height:
(9) Feynman is tall.
But (9) has another mode of use. Imagine that we are at a party. Perhaps Feynman
stands before us a short distance away, drinking punch and thinking about
dancing; in any case, the exact degree to which Feynman is tall is common
knowledge. You ask me what counts as tall in my country. ‘Well,’ I say, ‘around
here, . . . ’ and I continue by uttering. (9)
Barker (2) characterizes what is communicated by an utterance of (9) in the
situation he describes as follows: ‘My purpose in uttering (9) under such
circumstances would be nothing more than to communicate something
about how to use a certain word appropriately – it would be a metalinguistic
use’.
Plunkett and Sundell enrich Barker’s picture of metalinguistic use by
adding the possibility that metalinguistic uses of expressions can debate
not only how expressions are used, but how those expressions should be
used Plunkett and Sundell (2013, 3). They illustrate this possibility with
metalinguistic disputes over the use of evaluative expressions like ‘spicy’,
where it isn’t the case that there is some antecedently settled local standard
for what makes something count as spicy. Spiciness is up for grabs in such
a situation, and the appropriate use of ‘spicy’ has to be ‘negotiated’ by
the participants in the conversation (15). A pair of statements like (10)
and (11) can function as moves in a metalinguistic negotiation about the
appropriate use of the adjective ‘spicy’:
26
N. HANSEN
(10) That chili is spicy!
(11) No, it’s not spicy at all. (Plunkett and Sundell 2013, 15)
Adopting Plunkett and Sundell’s terminology, let a ‘metalinguistic proposal’ be an individual move in a metalinguistic negotiation: it is a proposal
for how an expression should be used.
A metalinguistic proposal, like a categorial declarative, has a descriptive
form and a normative conversational role. Cavell’s central examples of
categorial declaratives involve statements about what saying something
conveys or implicates, and what is implicated by what is said should be a
good candidate for metalinguistic negotiation, since they aren’t explicitly
encoded in what is said and Grice observes that what is implicated is indeterminate (Grice, 1975, 58).20 There should therefore be room to negotiate
what is conveyed or implicated by an utterance.
Consider again Cavell’s example of a categorial declarative, S:
S ‘When we ask whether an action is voluntary, we imply that the
action is fishy’.
How can a speaker be entitled to say S without ‘counting noses’? Once
S is understood as a metalinguistic proposal for how ‘voluntary’ should be
used (as implicating that the action it describes is somehow ‘fishy’), and
not a description of the antecedently settled use of ‘voluntary’, then the
entitlement to S will come from the speaker’s status as a speaker of the
language. But the success of such a proposal will depend on the speaker’s
ability to get her audience to accept her proposal, as Cavell discusses in
his invocation of the parallel between statements about ordinary use and
aesthetic judgments.
Using the notion of a metalinguistic proposal, it is also possible to
illuminate what Austin is doing when he draws fine-grained differences
in meaning between expressions like ‘intentionally’ and ‘deliberately’. As
discussed above, Hansen and Chemla (2015) found no evidence that ordinary speakers draw a clear distinction between the meaning of doing
something ‘intentionally’ and doing it ‘deliberately’. But Austin wants to
distinguish the meaning of those two expressions, and he argues that in his
go-cart story, it is better to say that the protagonist runs over the go-cart
deliberately, but not intentionally, because the protagonist didn’t intend to
20 Grice’s claim about the indeterminacy of implicatures was brought to my attention by Gray (forthcoming).
Plunkett and Sundell (2013, 3) say that disputes over how words should be used could arise when we
are ‘resolving ambiguity, prescisifying a vague term, setting a contextual parameter, or in any other way
determining how some antecedently indeterminate matter of meaning should be settled’.
INQUIRY
27
run over the go-cart. Austin, in discussing his method for distinguishing the
meaning of ‘intentionally’ and ‘deliberately’ says:
We must . . . imagine some cases (imagine them carefully and in detail and
comprehensively) and try to reach agreement upon what we should in fact say
concerning them. (Austin 1966, 429)
Austin’s methods for reaching agreement include ‘glossing’ those cases –
making meta-linguistic proposals for how the relevant expressions should
be used.
5. Must we measure what we mean?
Cavell argues that speakers of a language can be entitled to claims about
how ‘we’ use expressions, without having to ‘count noses’. Mates (1958),
Fodor and Katz (1963), and Jackman (2001) have criticized various attempts to explain the source of that entitlement. I have argued that an
essential component of defending Cavell’s proposal requires emphasizing
the normative nature of ‘categorial declaratives’. More specifically, I have
suggested that categorial declaratives should be understood as proposals
for how expressions should be used, or for what we should understand to
be implicated by a statement. One can be entitled to such claims in virtue
of being a speaker of the relevant language without having to conduct
empirical observations of how ordinary people do in fact speak. But Cavell’s
proposal for how to understand the success conditions of such claims
is both demanding and fragile – it is analogous to the conditions of a
successful aesthetic judgment, which, according to Cavell, depends on the
audience having the appropriate kind of reaction (analogous to getting the
audience to taste the leather and iron in the wine in the story about Sancho
Panza’s relatives). That condition on the success of a categorial declarative
is hard to satisfy – Cavell’s own attempt to tie uses of ‘voluntary’ to fishiness
is not convincing, for example, for the reasons discussed in Section 4.2.
So the answer to the question must we measure what we mean? – when
that is taken as a question about the need to survey how speakers of
the language actually speak when making claims about what we mean
by the use of an expression – is no, but it is extremely difficult to make
a compelling proposal about the meaning of an expression that ignores
evidence about how the expression is in fact used. Citing experimental
(or corpus-based) evidence of a distinction in use can be an effective way
of convincing an audience that a particular difference in meaning exists,
and can play a powerful rhetorical role in producing the acknowledgment
necessary for a successful categorial declarative. Ryle referred to this as the
28
N. HANSEN
feeling of being ‘strengthened, when told that big battalions are on [your]
side’ (Ryle 1953, 177). Even though Cavell is right that it is possible to be
entitled to make claims about how ‘we’ use certain expressions without
engaging in experimental or corpus-based investigations of how words are
used, contemporary practitioners of ordinary language philosophy should
not ignore the power of the big battalions.
Acknowledgements
Thanks to Eliot Michaelson, Chauncey Maher, Dan Lassiter, Jay Elliott, Bruno Ambroise,
Zed Adams, participants in the Cavell seminar at the New School for Social Research
and the fellows’ workshop at Stanford University’s Humanities Center for very helpful
comments.
Disclosure statement
No potential conflict of interest was reported by the author.
Funding
This project is supported by an External Faculty Fellowship at Stanford University’s
Humanities Center and a Leverhulme Research Project [grant number RPH-2016-193].
ORCID
Nat Hansen
http://orcid.org/0000-0001-5074-1075
References
Austin, J. 1956–1957. “A Plea for Excuses.” Proceedings of the Aristotelian Society 57: 1–30.
Austin, J. 1962. Sense and Sensibilia. Oxford: Oxford University Press.
Austin, J. 1966. “Three Ways of Spilling Ink.” The Philosophical Review 75 (4): 427–440.
Barker, C. 2002. “The Dynamics of Vagueness.” Linguistics and Philosophy 25 (1): 1–36.
Bates, S., and T. Cohen. 1972. “More on What We Say.” Metaphilosophy 3 (1): 1–24.
Baz, A. 2012. When Words Are Called for: A Defense of Ordinary Language Philosophy.
Cambridge, MA: Harvard University Press.
Braisby, N., B. Franks, and J. Hampton. 1996. “Essentialism, Word Use, and Concepts.”
Cognition 59 (3): 247–274.
Cavell, S. 1958. “Must We Mean What We Say?” Inquiry 1: 172–212.
Cavell, S. 2010. Little Did I Know: Exercepts from Memory. Stanford, CA: Stanford University
Press.
Cavell, S. 1976a. “Aesthetic Problems of Modern Philosophy.” Must We Mean What We
Say?, 73–96. Cambridge: Cambridge University Press.
INQUIRY
29
Cavell, S. 1976b. “Austin at Criticism”. Must We Mean What We Say?. 97–114. Cambridge:
Cambridge University Press.
Chapman, S. 2011. “Arne Naess and Empirical Semantics.” Inquiry 54 (1): 18–30.
Chappell, V., ed. 1964. Ordinary Language. Englewood Cliffs, NJ: Prentice-Hall.
Collins, J. 2007. “Linguistic Competence without Knowledge of Language.” Philosophy
Compass 2 (6): 880–895.
Culbertson, J., and S. Gross. 2009. “Are Linguists Better Subjects?” British Journal of
Philosophy of Science 60 (4): 721–736.
DeRose, K. 2005. “The Ordinary Language Basis for Contextualism, and the New
Invariantism.” The Philosophical Quarterly 55 (219): 172–198.
Fischer, E. 2014. “Verbal Fallacies and Philosophical Intuitions: The Continuing Relevance
of Ordinary Language Analysis.” In J.L. Austin on Language, edited by B. Garvey, 124–
140. Basingstoke: Palgrave MacMillan.
Fodor, J. A., and J. J. Katz. 1963. “The Availability of What We Say.” The Philosophical
Review 72 (1): 57–71.
Genone, J. 2012. “Theories of Reference and Experimental Philosophy.” Philosophy
Compass 7 (2): 152–163.
Genone, J., and T. Lombrozo. 2012. “Concept Possession, Experimental Semantics, and
Hybrid Theories of Reference.” Philosophical Psychology 25 (5): 717–742.
Gordon, M. J. 2013. Labov: A Guide for the Perplexed. London: Bloomsbury.
Gray, A. Forthcoming. “What’s Wrong with Indeterminate Implicature?”
Grice, H. 1961. “The Causal Theory of Perception.” Proceedings of the Aristotelian Society
35: 121–168.
Grice, H. 1975. “Logic and Conversation.” In Syntax and Semantics. vol. 3, edited by
P. Cole and J. L. Morgan, 41–58. New York: Academic Press.
Gustafsson, M. 2005. “Perfect Pitch and Austinian Examples: Cavell, Mcdowell,
Wittgenstein, and the Philosophical Significance of Ordinary Language.” Inquiry 48
(4): 356–389.
Hampton, J. A., and A. Passanisi. 2016. “When Intensions Do Not Map onto Extensions:
Individual Differences in Conceptualization.” Journal of Experimental Psychology:
Leaning, Memory, and Cognition 42 (4): 505–523.
Hanfling, O. 2000. Philosophy and Ordinary Language: The Bent and Genius of our Tongue.
London: Routledge.
Hansen, N. 2014. “Contemporary Ordinary Language Philosophy.” Philosophy Compass
9 (8): 556–569.
Hansen, N. 2015. “Experimental Philosophy of Language.” Oxford Handbooks Online.
doi:10.1093/oxfordhb/9780199935314.013.53.
Hansen, N., and E. Chemla. 2015. “Linguistic Experiments and Ordinary Language
Philosophy.” Ratio 28 (4): 422–445.
Hansen, N., and E. Chemla. 2017. “Color Adjectives, Standards and Thresholds: An
Experimental Investigation.” Linguistics and Philosophy. First online: 1–40.
Harman, G. 1976. “Practical Reasoning.” The Review of Metaphysics 29 (3): 431–463.
Henson, R. G. 1965. “What We Say.” American Philosophical Quarterly 2 (1): 52–62.
Jackman, H. 2001. “Ordinary Language, Conventionalism, and a Priori Knowledge.”
Dialectica 55 (4): 315–325.
30
N. HANSEN
Jylkkä, J., H. Railo, and J. Haukioja. 2009. “Psychological Essentialism and Semantic
Externalism: Evidence for Externalism in Lay Speakers’ Language Use.” Philosophical
Psychology 22 (1): 37–60.
Kripke, S. 1980. Naming and Necessity. Cambridge, MA: Harvard University Press.
Labov, W. 1975. “Empirical Foundations of Linguistic Theory.” In The Scope of American
Linguistics: Papers of the First Golden Anniversary Symposium of the Linguistic Society of
America, edited by R. Austerlitz, 77–134. Lisse: The Peter de Ridder Press.
Labov, W. 2006. The Social Stratification of English in New York City. 2nd ed. Cambridge:
Cambridge University Press.
Laugier, S. 2013. Why We Need Ordinary Language Philosophy. Chicago: University of
Chicago Press.
Lyas, C., ed. 1971. Philosophy and Linguistics. London: Macmillan.
Lynch, J. 2009. The Lexicographer’s Dilemma. New York: Walker & Company.
Machery, E., R. Mallon, S. Nichols, and S. P. Stich. 2004. “Semantics, Cross-cultural Style.”
Cognition 92: B1–B12.
Mates, B. 1958. “On the Verification of Statements about Ordinary Language.” Inquiry 1
(1): 161–171.
Maxwell, G., and H. Feigl. 1961. “Why Ordinary Language Needs Reforming.” The Journal
of Philosophy 58 (18): 488–498.
Mulhall, S. 1999. Stanley Cavell: Philosophy’s Recounting of the Ordinary. Oxford: Oxford
University Press.
Murphy, T. S. 2015. “Experimental Philosophy: 1935–1965.” In Oxford Studies in
Experimental Philosophy. Vol. 1, edited by T. Lombrozo, J. Knobe and S. Nichols, 325–
368. Oxford: Oxford University Press.
Myers, J. 2009. “Syntactic Judgment Experiments.” Language and Linguistics Compass
3 (1): 406–423.
Naess, A. 1938. ‘Truth’ As Conceived by Those Who Are Not Professional Philosophers. Oslo:
Kommisjon Hos Jacob Dybwad.
Nichols, S., N. Ángel Pinillos, and R. Mallon. 2016. “Ambiguous Reference.” Mind 125
(497): 145–175.
Plunkett, D., and T. Sundell. 2013. “Disagreement and the Semantics of Normative and
Evaluative Terms.” Philosophers’ Imprint 13 (23): 1–37.
Plunkett, D., and T. Sundell. 2014. “Antipositivist Arguments from Legal Thought and
Talk.” In Pragmatism, Law, and Language, edited by Hubb, G. & Lind, D., 56–75.
Routledge.
Potter, S. 2005. Lifemanship: Some Notes on Lifemanship with a Summary of Recent
Research in Gamesmanship. London: Moyer Bell and its subsidiaries.
Ryle, G. 1949. The Concept of Mind. New York: Routledge.
Ryle, G. 1953. “Ordinary Language.” The Philosophical Review 62 (2): 167–186.
Sandis, C. 2010. “The Experimental Turn and Ordinary Language.” Essays in Philosophy
11 (2): 181–196.
Schlenker, P. 2003. “Indexicality, Logophoricity, and Plural Pronouns.” In Research
in Afroasiatic Grammar II, Selected Papers from the Fifth Conference on Afroasiatic
Languages, edited by J. Lecarme, 409–428. Amsterdam: Benjamins.
Searle, J. 1969. Speech Acts. Cambridge: Cambridge University Press.
INQUIRY
31
Sprouse, J. 2013. “Acceptability Judgments.” Oxford Bibliographies. http://www.
oxfordbibliographies.com/view/document/obo-9780199772810/
obo-9780199772810-0097.xml
Sprouse, J., C. T. Schütze, and D. Almeida. 2013. “A Comparison of Informal and Formal
Acceptability Judgments Using a Random Sample from Linguistic Inquiry 2001–2010.”
Lingua 134: 219–248.
Sundell, T. 2010. “Disagreements about taste.” Philosophical Studies 155 (2): 267–288.
Tennessen, H. 1962. “Permissible and Impermissible Locutions.” In Logic and Language,
Synthese Library. Vol. 5, 220–233. Dordrecht: Springer.
Vendler, Z. 1967. “Linguistics and the a Priori.” In Linguistics in Philosophy, 1–32. Ithaca,
NY: Cornell University Press.