Philosophical Perspectives, 22, Philosophy of Language, 2008
COMPREHENDING SPEECH
Guy Longworth
University of Warwick
Paying attention to language per se is difficult.
–Merleau-Ponty (1973: 77)
1.
Consider attempting to make out the words of an interlocutor in a noisy
bar. By carefully attending to the sounds that they are producing and perhaps
also to the accompanying movements of their mouths, faces, and hands you are
able to make out what they are saying—to comprehend their speech. You thereby
come to entertain a thought that your interlocutor has expressed, for example,
the thought that they would prefer a pint of Adnams. And you may also come
to associate that thought with their production of words that express it and
with their having so expressed it. (Indeed, if the thought that your interlocutor
expressed involved some forms of indexicality—for instance, if it was the thought
that they (the speaker) would now (the time of speaking) prefer Adnams—then
your comprehending the thought would appear to be dependent upon your
taking a view as to its provenance in a particular production of speech.) Typically,
you are able to achieve this even though the bar is noisy and even though the
background noise is partly constituted by other productions of English words
that overlap with your interlocutor’s. In fact, it would be possible for you to
achieve as much even if another speaker had produced simultaneously a matching
sound—if they had asked for a pint of Adnams in a voice that would in the same
context be indiscriminable from your interlocutor’s—just as long as you were
in a position to discriminate in your perceptual context amongst the different
sources of those sounds and so to direct your attention to your interlocutor’s
production rather than the interloper’s. Proper exercise of these abilities can play
an important part in shaping the success of consequent thought and action, not
only in determining that you buy the right drink for the right person, but more
340 / Guy Longworth
widely in determining that you respond correctly to the thoughts and wants that
are expressed in your presence.
Although we are typically in a position correctly to identify the distal sources
of our acts of comprehension, through perception of productions of speech,
it appears that we need not always be in that position. Descartes makes the
following observation about the role of speech perception in comprehension:
Words, as you well know, bear no resemblance to the things they signify, and yet
they make us think of those things, frequently even without our paying attention
to the sound of the words or to their syllables. Thus it may happen that we hear
an utterance whose meaning we understand perfectly well, but afterwards we
cannot say in what language it was spoken. . .But perhaps you will say that our
ears really cause us to perceive only the sound of the words. . .and that it is our
mind which, recollecting what the words. . .signify, represent their meanings to
us at the same time. (Descartes, 1664/1985: 79)
Descartes’ interest in linguistic understanding is due to two apparent similarities
with perceptual cognition more generally.1 First, Descartes takes the relation of
the immediate deliverances of the sensorium—sensations, say—to the normal
objects of perceptual cognition—ordinary objects, properties and activities—to
be, like the relation between words and what they express, somewhat arbitrary.
Second, and related, he takes the epistemological role of sensation in mediating
perceptual cognition to be akin to the role of speech perception in mediating
understanding. Although Descartes does not draw definite conclusions, one
might be tempted to reason as follows. Since the comprehension of content
expressed in an episode of speech can be indifferent to our attending to, making
judgements about, or remembering the perceptually revealed details of that
episode, it is held that the perception of speech can play the role, not of supplying
justificatory support for knowledgeable uptake, but rather of being a mere
trigger to immediate comprehension. Similarly, since the perceptual cognition of
ordinary objects, properties, and activities can be indifferent to our attending to,
cognizing, or remembering the details of sensory activity, it is held that awareness
of sensation does not function as an element in its justification. Both points of
similarity leave open the nature of the triggered states or episodes: whether they
are, as Descartes has his interlocutor say, episodes or states purely of the mind
or understanding, or whether they bear the imprint of their causal pedigree, so
are episodes or states partly of sensory awareness.
Whatever Descartes’ own intentions were in drawing attention to the
apparent possibility of comprehension in the absence of conscious awareness of
certain aspects of speech, acknowledgement of that possibility might be thought
to point to a way of distinguishing ordinary cases of sensory perception from
cases of comprehension. For it might be held that belief or judgement based
upon sensory perception is dependent epistemically upon its basis, so that a
belief, or a judgement so based would count as knowledgeable only if it were
Comprehending Speech / 341
properly integrated with, and not merely caused by, its sensory basis. By contrast,
it might be argued that knowing or understanding what has been said in one’s
presence depends only upon the proper epistemic functioning of the faculties
responsible for one’s facility with content. And on that basis, it might be argued
that understanding what has been said, and the epistemic standing of further
cognitive work based upon that achievement, differs from sensorily based belief
and judgement in that it is dependent at most causally on operations of one’s
sensorium.
As well as its intrinsic interest, the issue bears on a wide range of questions
in the philosophies of language, mind, and knowledge. To what extent is
comprehension a perceptual, as opposed to a cognitive achievement? Is it possible
to acquire a priori knowledge on the basis of others’ testimony? In what, if any,
ways does the comprehension of speech invoke constitutive connections with
other speakers and their activities, for instance through perceptual contact? Of
particular interest to the present author is the question whether normal awareness
of what others have said—normal understanding or comprehension—depends
upon awareness of an association between their words and the contents they
express through the use of those words. For I have argued elsewhere that, if we
assume that normal understanding does depend upon such awareness, then it
is puzzling how we should account for the understanding apparently exhibited
by speakers that are apparently unable to conceptualize such associations—for
instance, small children and some people on the autistic spectrum.2
In what follows, I shall pursue the question whether sensorily based belief
can be distinguished from comprehending, or knowing what a speaker has said,
in the way proposed. I shall take as my stalking horse the most sophisticated
available presentation of a view according to which the two types of output
can be so distinguished, that developed in a series of important papers by Tyler
Burge. A happy consequence of this approach is that it provides an opportunity
to offer a summary of some of Burge’s subtle and wide-ranging work on this
topic. Burge’s view is both attractive and complex, and I think that many of its
subtleties have not been fully appreciated.3 However, I shall argue that one of
the major epistemic functions of comprehension pursued by Burge, its role in
sponsoring the transmission of knowledge, is not well served by his account.
2.
What, then, is the epistemological role of speech perception in acquiring
knowledge from what others say? In seeking an answer, one might reason as
follows. One can only acquire knowledge from another’s words if one understands
what they say. And understanding what they say is dependent upon perceiving an
episode of speech.4 Hence, one might conclude, the perception of speech supplies
an essential element in the justification of, or entitlement to, knowledge acquired
from what others say.5
342 / Guy Longworth
Although the initial steps of the reasoning are plausible, they fail without
supplementation to sustain its conclusion. Supplementation is required at two
points. First, no reason has been supplied for viewing the role of understanding
an utterance to be that of supplying an essential element in an entitlement to
what one comes thereby to know. For all the reasoning shows, it might be
that understanding an interlocutor’s utterance that such-and-such serves, not to
supply an element in an entitlement to believe that such-and-such, but rather
to put one in a position to enjoy whatever entitlement the interlocutor then
possessed to believe that such-and-such. Second, no reason has been supplied
for viewing the role of speech perception to be that of supplying an essential
element in an entitlement to one’s understanding of the interlocutor’s utterance.
For all the reasoning shows, it might be that the perception of speech serves, not
to supply an essential element in one’s entitlement to a cognition delivered by
the understanding, but rather to causally trigger the generation of episodes of
comprehension with epistemological statuses independent of their merely causal
pedigree.
Although both lines of resistance rely upon a distinction amongst the
epistemological functions of elements responsible for the epistemic status of a
piece of cognition—between, on one hand, triggering, sustaining, preserving, or
enabling conditions and, on the other, essential components in an entitlement—
they are independently capable of undermining the initial line of reasoning in
favour of a justificatory function for the deliverances of speech perception.
One could resist the reasoning by holding that speech perception plays a
justificatory role with respect to understanding, but that understanding itself
plays a merely enabling role, allowing the subject of understanding to share
in their interlocutor’s epistemic status. And one could resist the reasoning by
holding that understanding plays a justificatory role with respect to knowledge
gained from testimony, but that understanding is not reliant for its justification
upon the sensory perception of episodes of speech.
A guiding theme in Burge’s work on the topic of knowledge transmission is
his attempt to secure the possibility of the transmission of a priori knowledge as
a priori. That is, Burge aims to explain, not only how someone who has a priori
knowledge that such-and-such can communicate that knowledge to another, but
also how the other’s knowledge can retain a priori status despite the transmission
of knowledge depending upon the perception of speech. He explains his notion
of a priori knowledge as follows:
I understand ‘a priori’ to apply to a person’s knowledge when that knowledge
is underwritten by an a priori justification or entitlement that needs no further
justification or entitlement to make it knowledge. A justification or entitlement
is a priori if its justificational force is in no way constituted or enhanced by
reference to or reliance on the specifics of some range of sense experiences or
perceptual beliefs. (Burge, 1993a: 458)
Comprehending Speech / 343
Since Burge accepts that ordinary knowledge transmission is dependent upon
speech perception, he draws a distinction, of the sort alluded to above, between
two sorts of dependence. On one hand, we have dependence of a piece of
cognition for its justificational force upon the specifics of what is sensorily
experienced. On the other, we have dependence of a piece of cognition for its
acquisition or preservation on a mere enabling or triggering condition. Having
drawn the distinction, it would be open to Burge to attempt to explain the
transmission of a priori knowledge by neutralising the justificatory role of
speech perception in either of the two ways mentioned above. Burge chooses the
second way, and seeks to argue that, while the products of comprehension play a
justificatory role with respect to the knowledge got from others, speech perception
has only a triggering or enabling function in underwriting the cognition of
content, and the preservation of the justificatory status that content has for
the interlocutor.
Burge’s account of the comprehension of speech, then, must provide that
the epistemological role of episodes of comprehension can be indifferent to
their sensory pedigree.6 That is, it must demonstrate, not only the possibility of
forms of comprehension whose epistemological status ‘is in no way constituted
or enhanced by reference to or reliance on the specifics of some range of sense
experiences or perceptual beliefs’. It must show that such forms of comprehension
are able to fulfil their epistemic function in furnishing justificatory support for
pieces of knowledge acquired on their basis. Crucially, such forms of comprehension must not rely upon sensory perception to supply required elements in their
inferential justificatory support. And, of equal importance, neither must those
forms themselves be forms of sensory perception. Or, more precisely, their being
forms of sensory perception must not be essential to their capacity to play their
required role in mediating further epistemic statuses.
The issue dividing Burge from his opponents concerns the respective
epistemic functions, in supporting the communication of knowledge, of the
deliverances of two faculties: a faculty of sensory perception, marked as such by
its special dependence upon the sensorium in sustaining causal and informational
connections with particular concrete objects, properties, and events; and a faculty
of understanding, marked as such by its role in sustaining ordinary thinking
capacities. The role of ancillary faculties—e.g. those responsible for deductive
or inductive reasoning and, as we shall see, certain forms of memory—is taken
to be neutral, except insofar as they take inputs from the other faculties. Burge
considers two conceptions of comprehension and argues that neither should be
accepted. According to the first, episodes of comprehension are outputs of the
understanding, under the inferential, justificatory guidance of sensory perception
of episodes of speech. According to the second, episodes of comprehension
are exhaustively determined by the operations of sensory perception, so that
comprehension just is a form of sensory perception. I shall expound Burge’s
arguments against the two positions in order.
344 / Guy Longworth
3.
Burge’s argument against the first position has three stages. In the first stage,
Burge presents a model of epistemological entitlement. In the second stage, the
model is applied to communication in a way that reveals the products of speech
perception to be inessential to one’s entitlement to knowledge gained from others.
In the third, independent considerations are offered against the view that speech
perception provides essential elements in such an entitlement.
The first stage is driven by reflection on cases where it might seem plausible
that a distinction should be drawn amongst necessary conditions for knowledge.
The aim is to motivate a distinction between those elements that must be
mentioned in describing an adequate justification or warrant for belief and those
elements that serve as mere enabling or triggering conditions. The operation of
the latter elements may be required in acquiring or sustaining possession of the
justification constituted by the former elements; but they are otherwise inessential
to epistemic status.
Consider the role of memory in sustaining one’s capacity to follow the steps
of a proof—the function of what Burge calls preservative memory. Plausibly, the
proper functioning of memory here is essential to the epistemic status of any
belief formed on the basis of following through the proof. If one has forgotten
or mis-remembered an earlier step, then that can affect critically the epistemic
status the conclusion of one’s reasoning has for one. More generally, the course
of one’s cognitions as one follows a proof must be suitably interconnected,
hence underwritten in part through the proper functioning of memory. But it is
plausible that one’s justification for belief in the conclusion need not mention the
fact that one’s memory functioned properly. What is required is that the thoughts
that one entertains in following the proof are suitably connected. Although that
may depend on the proper functioning of one’s memory, one need not in addition
entertain beliefs about the functioning of memory. Thus, we might distinguish
between the justificatory role of the things one remembers in following through
the proof—the propositions constituting the proof—and the enabling role of
memory, in preserving one’s ability to think the propositions involved in the
proof as one works through it and in preserving justificatory support from earlier
to later steps despite the earlier steps no longer being before one’s mind.
Burge attempts to underpin the distinction between essential elements in a
justification or entitlement and mere enabling conditions through appeal to what
he calls the Acceptance Principle:
A person is a priori entitled to accept a proposition that is presented to him as
true and that is intelligible to him, unless there are stronger reasons not to do
so. (1993a: 469)
The Acceptance Principle is a central plank in Burge’s project of domesticating
putatively Externalist determinants of epistemological status. In Burge’s view,
Comprehending Speech / 345
appeal to such determinants—for instance, reliability—must be underwritten
through appeal to the functional nature of the faculties, resources, and episodes
that they condition. The (prima facie) entitlement constituted by the reliability
of a faculty, resource, or episode cannot outstrip the autonomous status that
would be afforded by exercises and their products that optimally fulfil the
proper function of the faculties, resources, or episodes that they involve. Where
the operations of a faculty provide pro tanto entitlement to cognition with
some epistemic status, the optimal functioning of the faculty must provide a
guarantee of cognition with that status—i.e. must suffice for that status. This
aspect of Burge’s entitlement-based epistemology is crucial to his overall account
of comprehension. Burge holds that non-optimal operations of a faculty can
derive epistemic status from optimal operations of that faculty, so that the
output of such non-optimal operations can be shaped by factors external to
the non-optimal operation just insofar as those factors would impact upon
its optimal operation, and so be internal to its optimal, status guaranteeing
operation. Hence, in one direction, non-optimal operations of comprehension
can exploit the powers of optimal operations, and derive epistemic status from
possible optimal operations. And, in the other direction, the possible impact of
operation external factors is constrained by the powers of optimal operation:
factors that would have no impact upon optimal functioning—i.e. factors that
go beyond what optimal operations can suffice to determine—can play no role in
determining the epistemic status of imperfect operations. The epistemic status of
any of a faculty’s operations cannot outstrip the epistemic status of that faculty’s
optimal operations. Although I shall not argue it here, I think that this represents
a minimal fix of Externalist epistemology. I shall assume that any adequate
epistemology must be subject to at least that degree of Internalist constraint.
Those who find that assumption controversial should assess the remainder as
conditional upon its acceptability.7 My argument against Burge’s view aims
to exploit his view that the epistemic powers of a faculty cannot outstrip its
optimal, status guaranteeing function. The aim is to demonstrate that, on the
(shared) assumption that the optimal operations of the faculties responsible for
comprehension can sustain the transmission of knowledge, those faculties must
include sense-perceptual elements.
The entitlement articulated through the Acceptance Principle is only pro
tanto, a default but defeasible epistemic status. The a priori, default status of
the principle is taken by Burge to be a regress-halting component of properly
functioning justificatory practice:
A presupposition of the Acceptance Principle is that one is entitled not to bring
one’s source’s sincerity or justification into question, in the absence of reasons
to the contrary. This too is an epistemic default position. (1993a: 468; see also,
e.g., 463)
Further support for the principle comes from reflection upon the functional
nature of mental or epistemic faculties—in particular, constitutive principles
346 / Guy Longworth
of rationality and charity—and the idea that one is entitled to presume upon
the proper functioning of rational systems, faculties, and, derivatively, their
products.8 Such an entitlement may be defeasible through empirical information.
But it is a default, in the sense that it can sustain an entitlement sufficient
to confer positive epistemic status on a subject without requiring additional
premises to the effect that there are no defeaters in a particular case.
One’s entitlement depends, not only upon whether a proposition that is
presented as true in fact derives in the right way from an earlier cognition, but
also upon whether (some-) one was entitled to that earlier cognition.9 As Burge
explains the latter sort of dependence,
We would not be entitled to the belief if it were preserved from unwarranted
acquisitions that we had forgotten: we cannot, I think, become warranted
by forgetting the poor grounds we originally had, and then relying on the
remembered belief. (1997: 39–40, Burge’s emphasis)
Correlatively, then, there are two sorts of defeat to which a piece of cognition is
subject: defeat of an entitlement to enjoy the entitlement of an earlier cognition;
and defeat of the entitlement to the earlier cognition.10 Moreover, the basic
entitlement to a piece of cognition, and hence its epistemic classification as a
priori or aposteriori, depends upon the extended entitlement that includes the
entitlement to the earlier cognition:11
Memory is no more intrinsically an empirical faculty than it is a rational
faculty . . . . Even in empirical reasoning, memory has a purely preservative
function that does not contribute to the force of the justification, but simply
helps assure the proper working of other cognitive capacities over time. (1993a:
463–4)
If the entitlement for a piece of cognition is a priori, memory can preserve
its a priori status; mutatis mutandis if the cognition is aposteriori. As Jim
Edwards emphasizes, the Acceptance Principle functions as a sort of defeasible
rule of inference. In undefeated cases, it facilitates the transfer of warrant or
justificatory status without itself playing the role of a premise.12 Unlike genuine
elements in a justification or entitlement, which must work together to ensure
the epistemic status of a piece of cognition, absence of defeat—so, for instance,
proper functioning of memory—is a mere enabling condition for the status and
proper linking of genuine elements of the entitlement.
Before moving on to consider the second stage in Burge’s argument, the
extension of the basic account to knowledge gained from others, we should pause
to note two points about his account as it applies to preservative memory. The
first is that the account given thus far is neutral between the two lines of resistance
to a perceptually based epistemology of comprehension outlined earlier. That
is, one might view the account as treating memorial presentations-as-true
Comprehending Speech / 347
as epistemologically insulated from their memorial pedigree, but as playing a
justificatory role with respect to later cognitions. Or, one might view the account
as requiring that memorial cognitions must be marked, or indexed, as such—
that the subject’s epistemic economy must be sensitive to their status as putatively
retained—in order for them to fulfil their later epistemic functions, while allowing
that their memorial status itself plays no further justificatory role.
The second point is an explanation of the first. The reason the account
can afford to be neutral about whether a memorial presentation-as-true must be
marked, or indexed, as such is that its being so-marked would have no effect on
the epistemological classification of cognitions in whose entitlement it serves—
that is, in determining whether they are a priori or aposteriori. Such marking
would only have such an effect if it carried empirical information concerning particular subject matters—particular objects, properties, or episodes. But
preservative memory carries no such information. An aspect of the Acceptance
Principle is that one is a priori—though defeasibly—entitled to presume upon the
proper functioning of one’s faculties. As long as the relevant features of proper
functioning do not depend upon contingent features of particular exercises of
a faculty, particular applications of the Principle need not import empirical
elements into an entitlement that they help to constitute. One is therefore
a priori entitled to presume upon information concerning particular subject
matters that is fixed solely by completely general requirements on the proper
functioning of one’s faculties. But the proper functioning of a subject’s memory
with respect to any given presentation-as-true ensures, quite generally, that the
presentation is preserved from an earlier cognition of the subject.13 So despite
their carrying information about a particular, contingent subject matter, exercises
of preservative memory can retain their epistemological neutrality because the
information they carry is determined by general conditions on proper functioning
of any such exercise.
4.
Burge seeks to extend his basic account to knowledge acquired through
communication. Burge outlines his account as follows:
In interlocution, perception of utterances makes possible the passage of propositional content from one mind to another rather as purely preservative memory
makes possible the preservation of propositional content from one time to
another. (1993a: 481)
As in his account of preservative memory, Burge exploits the distinction
between enabling or triggering conditions and justificatory elements. Again,
the Acceptance Principle plays a key role. But in this case the role is, not to
preserve the justificatory status of cognized content intra-personally across time,
348 / Guy Longworth
but rather to preserve it inter-personally. Like preservative memory, preservative
interlocution functions to facilitate the recipient’s access to whatever warrant is
already possessed by their interlocutor and not to generate afresh a warrant for
the recipient. As Burge puts it,
The default entitlement to believe propositions one received in interlocution presumes a more primary epistemic warrant somewhere in the chain of interlocutors.
(1997: 44, fn. 2)
Hence, the epistemological classification of a piece of cognition acquired from
another in accord with the Acceptance Principle is determined by the classification it had at source:
Sometimes, the epistemic status of beliefs acquired from others is not empirical.
In particular, it is not empirical just by virtue of the fact that the beliefs are
acquired from others. (1993a: 466, Burge’s emphasis)
It should be obvious how the Acceptance Principle can mediate vicarious
justification via cognizance of a genuine presentation-as-true. But it may be less
obvious how the account sustains an entitlement to comprehension of genuine
presentation that is not constituted or enhanced by perception of an interlocutor’s
speech. In order to bring the initial stage of knowledge transmission within the
purview of his account, Burge makes explicit a strengthening of the Acceptance
Principle:
We are a priori prima facie entitled to accept something that is prima facie
intelligible and presented as true. For prima facie intelligible propositional
contents prima facie presented as true bear an a priori prima facie conceptual
relation to a rational source of true presentations-as-true. . . (1993a: 472; see also
1997: 45, fn.4; 1999: 243ff)
Because the proper function of the understanding, in application to speech
episodes, is to deliver awareness of genuinely intelligible propositional contents
together with their true force—for our purposes, their having been presented
as true—one is a priori entitled to take prima facie awareness of such to be
genuine.
5.
We can use a simple example to make clear how an entitlement might be
articulated in a particular instance. Kim believes that there is no largest prime.
Her belief has an entitlement that is sufficient for it to count as knowledge and
that is not constituted or enhanced by products of sensory perception. So, Kim
has a priori knowledge that there is no largest prime. Within earshot of Jo,
Comprehending Speech / 349
Kim utters the sentence, ‘There is no largest prime’. Jo’s auditing those words
causes her understanding to produce in her a cognition of the propositional
content that there is no largest prime, with what appears to her to be the force
of a presentation-as-true received from without. Jo comprehends the expressed
content as received and as having been presented as true, thereby accepting
the immediate deliverance of her faculty of understanding. Although she makes
that transition automatically, she is entitled to do so because, first, one is in
general a priori, albeit defeasibly, entitled to accept as genuine an apparently
received presentation-as-true and, second, there are in this case no defeating
considerations.14 Crucially, as explained above, the explanation for why she is
entitled to accept the deliverances of her faculty of understanding is, not an
element in her entitlement or justification, but rather a sort of meta-entitlement
to have elements in her first-order entitlement. Thus, the fact that making explicit
the absence of defeat for Jo’s basic, a priori entitlement would involve mention
of the proper operation of her sense perceptual faculty does not in itself suffice
to make her entitlement empirical.
That completes the first stage in the articulation of Jo’s entitlement: the
account of her entitlement to accept as having been presented as true the
propositional content that there is no largest prime. The second stage articulates
her entitlement to accept the propositional content itself, to believe that there
is no largest prime. Here again, Jo may make the transition to acceptance
automatically. Two sorts of entitlement are involved: first, one’s entitlement to
accept what is presented to one as true, and so to partake in whatever extended
entitlement that presentation possesses; second, the extended entitlement possessed by the presenter-as-true, one’s interlocutor. In the present case, the first
sort of entitlement is one’s general, a priori, entitlement to presume on the proper
epistemic functioning of one’s interlocutor; the second consists in whatever
elements and meta-entitlements constitute Jo’s interlocutor’s entitlement—that
is, Kim’s extended entitlement to believe that there is no largest prime. Again,
both sorts of entitlement function only because they are not subject to defeat.
Again, the absence of defeaters need not be mentioned in an entitlement sufficient
for Jo’s belief to count as knowledge. Jo’s first-order entitlement contains no
empirical elements: her knowledge, like Kim’s, is a priori.
The two features of Burge’s account of one’s entitlement to cognition
gained through interlocution of most importance to the remainder are the
following. First, as noted above, optimal epistemic status cannot, on Burge’s
account, outstrip optimal fulfilment of proper function. Second, episodes of
comprehension play an essential role, as elements with their own proprietary
justificatory force, in the constitution of any extended entitlement one can gain
through an interlocutor.
6.
The upshot of the first two stages of Burge’s argument is that an account can be given of the knowledge-sufficing entitlement to some cases of
350 / Guy Longworth
comprehension—crucially, cases that are able to function as mediators in the
transmission of knowledge—that makes no mention of the products of speech
perception. As noted above, the first two stages depend upon the claim that the
faculty responsible for comprehension can function optimally to guarantee (i.e.
suffice for) knowledge transmission without dependence upon sense perception.
In the third stage, Burge argues that an account of the sort he presents should
be preferred to one on which comprehension is the upshot of inference from—so
inferentially entitled on the basis of—the perception, or perceptual cognition, of
speech. Since I agree with Burge that inferential models of the role of speech
perception in comprehension are unacceptable, I shall not comment in detail
on the third stage of his argument.15 However, the considerations have some
bearing on the credentials of the more straightforwardly perceptual models to be
considered in the remainder and so deserve an airing. Aside from observing that
such inferences seem, at best, sub-personal, and that ‘explicating [such inferential
entitlements] is well-nigh impossible. . .even for philosophers’ (1999: 240)—points
that Burge takes to be indecisive—he offers three main considerations.
The first consideration harks back to Descartes:
Understanding depends on perceptual awareness and implicit memory of words
in something like the way that perception depends on utilization of sensations
or perceptual representations in forming perceptual beliefs. But forming beliefs
about the properties of words is no more the aim of understanding than forming
beliefs about sensations or perceptions is the aim of perception. Understanding
of speech may well precede an ability to conceptualize and form beliefs about
a distinction between words and the objects they indicate or the meanings they
convey. It is well known that memory of the properties of words is less reliable
and more ephemeral than memory of what one understands through the words.
(Burge, 1999: 241)
The consideration has two related aspects. The first is that the epistemic statuses
of instances of comprehension can be greater than their supposed inferential
bases. The second is that the aim—read: function—of comprehension is not
the formation of beliefs about the properties of episodes of speech. I take
the first consideration to weigh heavily against some inferential models of the
epistemology of understanding. But the second is less compelling. It is not
obvious that the function of understanding does not involve sustaining awareness
of associations between speech episodes and expressed content. That depends
upon the precise character of that function; for present purposes, it depends
upon the role of speech perception in knowledge communication and, hence, on
the probity of Burge’s positive account.
The second consideration has a more recent provenance:
When communication runs smoothly, the question of justifying one’s understanding does not seem to arise. It is no more in place to ask someone who is a
perfectly competent language user to support his or her presumed understanding
Comprehending Speech / 351
of someone who says “push-button telephones are more common than rotary
ones” than it is to ask a normal perceiver how he or she justifies a perceptual
belief that that is a brown lectern, when he or she is looking at one in good
light. These questions are philosophers’ questions. Addressing them well requires
giving weight to the fact that they do not arise in that form in ordinary life. (1999:
241)
This consideration has force against many substantive forms of inferential
conception. However, it is compatible with conceptions according to which
answering justificatory questions is too difficult for ordinary folk because
the requisite inferential transitions are sustained inaccessibly, as a matter of
mere entitlement. And it is compatible with conceptions wherein answering
justificatory questions is in ordinary cases too easy to warrant response. Consider
the following exchange. Kim: ‘George said that the train is delayed’. Jo: ‘How do
you know? I didn’t hear him speak and even if I had, what he says sounds to me
like mere noise.’ Kim: ‘I know what he said because, first, I did hear him speak
and, second, I have/speak/know French, so I can understand the “mere” noises
that he makes.’ Of course the need for such response is rare. But that appears
to be, not because stakes or relevant alternatives have somehow been raised so
that ordinary epistemic practices require supplementation, but rather because the
relevant responses are ordinarily so obvious to all parties. However, the second
consideration does carry weight against inferential conceptions according to
which ordinary, personal level accounts of warrant must reconstruct a cogent
inference from mere sounds, or even sentences, to claims about expressed
content.
The third consideration is a thought experiment:
Suppose that we could not perceive words others speak. Suppose that the
stimulus effects of the words nevertheless affected us by some natural causal
process in such a way that we reliably understand their sense. . .as received—
rather than as initiated. Suppose that we could not directly know or even reliably
guess anything about the words whose effects were thus injected. Suppose that
the word sounds. . .called up understanding of conceptual content. . .by bypassing
the perceptual system, but triggering the same central mechanisms by which we
understand our own speech . . . . The words might become perceptible when but
only when something in the context provides grounds to doubt the standing
comprehension of what the interlocutor is saying. Understanding, however,
remains as good as ever. (Burge, 1999: 244)16
Again, much depends upon the function of understanding. It is at least plausible
that one might reliably entertain propositional contents on causal exposure to
speech episodes that express those contents, despite being wholly unaware of
those episodes. It is somewhat less plausible that, absent meta-level reflection
on the provenance of one’s entertaining, one would be in a position to reliably
entertain content as received, rather than as [self-] initiated.17 Be that as it may. As
352 / Guy Longworth
I shall argue shortly, the function of comprehension in knowledge-transmission
demands more.
7.
That completes Burge’s case against models of comprehension according
to which speech perception supplies an essential element in its inferential
entitlement. The second account Burge considers is one according to which
episodes of comprehension are sense-perceptual due to their being the sole
responsibility of exercises of the sensorium. Burge objects to this account—
decisively, in my view—that comprehension requires the entertaining of expressed
content, so exercise of the understanding:
We do not perceive the contents of attitudes that are conveyed to us; we understand them. We perceive and have perceptual beliefs about word occurrences.
We may perceive them as having a certain content and subject matter, but the
content is understood, not perceived. (Burge, 1993a: 478)
8.
The way in which the dialectic was established involved two key assumptions.
The first assumption, made explicit above, is that the issue concerns the roles of
two faculties—sense perception and the understanding—in producing episodes
of comprehension. The second assumption—one that has thus far remained
implicit—is that sustaining each episode is the responsibility of either a senseperceptual faculty or the faculty of understanding. For, I shall argue, the considerations Burge presents against a sensory model of comprehension are powerless
against a model involving both sensory and intellectual components. Given the
second assumption, an episode of comprehension is either sense-perceptual—the
product of the perceptual faculty—or a product of the understanding. If it is the
latter, it can at most depend upon independent sense-perceptual episodes. Since
comprehension is not purely sense-perceptual and since, according to Burge’s
arguments, it need not depend for its justificational force upon sense-perceptual
episodes, the only role left for speech perception to play is an enabling role.
Given his assumptions, Burge has exhausted the options. So the cogency of
Burge’s argument depends upon the virtue of those assumptions. I shall focus
here upon the second assumption.
Recall that Burge’s entitlement epistemology has the following core feature.
The epistemic powers of a faculty cannot outstrip the powers of its optimal
function, so that unless the optimal functioning of a faculty would guarantee
cognitions with some status, the sub-optimal operations of that faculty cannot
have that status. It is only because optimal functioning of a faculty can
Comprehending Speech / 353
guarantee the obtaining of conditions determinative of epistemic status that
those conditions can play a role in shaping the status of non-optimal operations
even where the conditions are external to—not guaranteed by—those nonoptimal operations, considered independently of their optimal counterparts. In
what follows, I shall argue that that feature of Burge’s epistemology has as
a consequence that, on the shared assumption that operations of the faculty
responsible for comprehension are able to underwrite the transmission of
knowledge, the optimal operations of that faculty must involve the operations of a
sense-perceptual faculty. More specifically, Burge’s epistemology requires that the
optimal functioning of the faculty responsible for comprehension can guarantee
the status of certain elements required for the transmission of knowledge partly
grounded in its operations. Otherwise, the ordinary sub-optimal operations of the
faculty cannot deliver elements with that status even pro tanto. And, as I shall
argue, the requirement on optimal functioning can be met only if the faculty
responsible for comprehension is a partly sense-perceptual faculty. It is a faculty
whose optimal operations are the unified upshot of sense perception and the
understanding, and cannot be decomposed into independent operations of those
two sub-faculties.
There appears to be no difficulty of principle in understanding how items of
cognition can be the unified, non-decomposable upshot of the exercise of more
than one faculty or cognitive resource.
Consider, for one sort of example, Jo and Kim. Jo and Kim know how
to drive to London, although neither possesses this knowledge in isolation:
only Jo knows how to state directions to London; only Kim knows how to
drive. We might think that, since someone might know what Jo knows without
being paired with a driver and someone might know what Kim knows without
being paired with a navigator, it is possible to decompose Kim and Jo’s state of
communal knowledge into individualistic components. However, things are not
so straightforward. In the sense in which Jo’s state of knowledge is separable,
George also knows how to state directions to London. But unlike Kim and Jo,
George speaks French. There is no way to recompose what Kim and Jo know
by appeal only to the things we have said that they each know separately.18
Another sort of example is provided by the range of partly perceptual
episodes involved in some experiences of pictorial representations. Consider, for
instance, seeing some lines as a (pictured) duck, or seeing a (pictured) duck in
some lines. Awareness of the (pictured) duck can rely, not only upon sensory
perception, but also upon faculties of memory and imagination. But aspects of
one’s experience that register the representational features of the lines cannot be
enjoyed independently of sensory presentation of the lines. And it is far from
clear that the specific sort of experience of the lines one enjoys in seeing a pictured
duck therein can be enjoyed without seeing the (pictured) duck.19
A third sort of example, of especial interest in the present context, is the
integration, in speech perception, of sub-components of the sense perceptual
faculty responsible for audition and vision. The effects of this sort of integration
354 / Guy Longworth
become apparent, for instance, through the ventriloquist effect, wherein a heard
voice can appear to emanate from the silent mouthing of a dummy.20
There appear to be numerous examples of types of psychological states
and episodes whose most fundamental nature is constituted by the operation
of what we might think of as a super-faculty composed by sub-faculties. They
are not mere congeries of more fundamental types of experience or attitude.
Rather, they are individuated by the combination of a single experiential or
attitudinal typing—say, their typing as an episode of perception or a state of
knowledge—together with a single content clause. Their distinctive feature is
that their content clause marks them out—indexes them—as due to exercises of
multiple faculties, each faculty responsible for only some of the aspects of their
content. In short, they are fundamentally unified psychological episodes—or, as
I shall sometimes say, cognitions—with more than one aspect. Without wishing
to prejudge the question whether the various aspects of cognitions due to such
integrated operations can be disentangled in any particular case, let us extend
the use of ‘aspect’ to apply also to the products of independent operations of
different faculties. It is then apparent that we can classify conceptions of episodes
of comprehension along two dimensions. Along the first dimension, we can ask
whether or not a conception treats episodes of comprehension as exhaustively
perceptual or perceptually justified, where an affirmative answer requires that
all aspects of those episodes are taken to be perceptual or perceptually justified.
Along the second dimension, we can ask whether or not a conception treats
all aspects of episodes of comprehension as belonging, indissolubly, to a single
episode type or cognition. We can tabulate answers to the two questions as in
figure 1:
Are all aspects of comprehension perceptual
or perceptually justified?
Are all aspects
involved, indissolubly,
in episodes of
comprehension aspects
of a single episode type
or cognition?
Yes
No
Yes
No
Pure perceptual
model
Perceptual
justification model
Burge’s triggering
or enabling model
Figure 1.
As figure 1 reveals, Burge fails to consider a fourth conception, according
to which an episode of comprehension is a single cognition with two aspects, the
co-operative output of exercises of sense perception and the understanding. For
all his argument shows, episodes of comprehension might involve the operation
of distinct faculties—speech perceptual faculties and the understanding—and
yet fail to decompose into elements that are the sole responsibility of speech
Comprehending Speech / 355
perceptual faculties and elements that are the sole responsibility of the understanding. On such a view, episodes of comprehension are viewed as the
responsibility of a super-faculty able to recruit and integrate the operations
of both sensory and intellectual faculties. And such a conception of episodes
of comprehension has some independent plausibility. Consider a description
of episodes of comprehension that Burge appears willing to endorse: ‘We may
perceive [word occurrences] as having a certain content and subject matter.’21
Prima facie, the description applies to a single episode type or cognition—a
sort of perception—involving the joint exercise of two faculties: sense-perceptual
registration of words together with understanding registration of content and
subject matter. But, as we have seen, Burge must reject any such characterization
of the episodes of comprehension upon which the communication of knowledge
depends. Since Burge holds that episodes of comprehension supply genuine
elements in the entitlement of later cognitions, he must deny that later functions
of those episodes depend upon their having perceptual aspects.
It is as yet unclear how Burge’s positive account compares with the revealed
conception, or how the latter fares in the face of his negative considerations.
Over the course of the following three sections, I shall argue that the conception
enjoys critical advantages over Burge’s in accounting for the communication of
knowledge.
9.
To reiterate: the question at issue is not whether comprehension invariably
involves—or, even, must involve—aspects of cognition sustained by the faculty
of sense perception. From the outset, I have allowed that episodes of some
forms of comprehension can be free of perceptual taint. And it should be
apparent, equally, that it is open to Burge to accept a role for sense perception in
comprehension, so long as that role has no justificatory function. The question
dividing Burge and me is the following: must the episodes of cognition delivered
by the optimal, status guaranteeing operations of the faculty responsible for comprehension have perceptual aspects if they are to guarantee comprehension based
elements involved in the communication of knowledge? For if an affirmative
answer is given to that question, then the faculty responsible for comprehension
able to sustain knowledge transmission, even pro tanto as in sub-optimal cases,
must be a partly sense-perceptual faculty. Since Burge’s account has it that
the cognitions constituted through the operations of comprehension have a
justificatory function in sustaining the transmission of knowledge, he is required
to give a negative answer. I shall now argue for an affirmative answer: in order to
fulfil their epistemic function in facilitating the transmission of knowledge from
one person to another, the cognitions constitutive of comprehension must have
a perceptual aspect.
To a good first approximation, the integrated account views comprehension
as, in Timothy Williamson’s terms, prime.22 That is, the integrated account denies
356 / Guy Longworth
that, consistently with its epistemic status, episodes of comprehension can be
factorized into sub-episodes, with each sub-episode the responsibility either of
the sensorium or the understanding. Rather, according to the integrated account,
the epistemic status of episodes of comprehension depends, not only upon the
proper functioning of the sensorium and the understanding, but also upon the
proper functioning of a super-faculty, consisting of both sub-faculties operating
in a properly integrated way.
Williamson is concerned primarily with what we might call vertical primeness. He argues that subject-internal and subject-external factors in some examples of environment-involving cognition must be properly integrated. To take
one example, he argues that states of knowledge about the external environment
cannot be factorized into subject-internal factors—perhaps including belief—and
subject-external factors—including truth. Although states of knowing depend
upon both subject-internal and subject-external factors, they depend also upon
the proper integration of those factors, upon the ways in which the subjectinternal and subject-external factors fit together. By contrast, our present
concern is with what we might call horizontal primeness. Here, the question
at issue concerns putative requirements, not on the proper integration of internal
and external factors in world-involving cognition, but rather on the proper
integration of the operations of different faculties involved in multiple-faculty
cognition—operations that, taken individually, may or may not be vertically
prime. For present purposes, it can be agreed that the optimal functioning of
comprehension requires the optimal functioning of both sensory and intellectual
faculties. At issue is whether, in addition, the optimal functioning of comprehension requires the optimal functioning of what is in effect a super-faculty
involving both the sensorium and the understanding. The defender of Burge’s
triggering account of comprehension must argue that optimally functioning
comprehension can be factorized horizontally into sensory and intellectual
components; the defender of the integrated account must argue that this is not
possible.
In a little more detail, and following Williamson’s discussions of vertical
primeness, we can think about the issue in the following way. A type episode of
comprehension, C—say the type: comprehending an utterance of ‘There is no
largest prime’—involves the following:
(i) A pure product of the understanding, U: a type of episode that can
be fully characterized without appeal to faculties other than the understanding.
(ii) A pure product of perception, P: a type of episode that can be fully
characterized without appeal to faculties other than sensory faculties.
It is open to both the triggering and the integrated accounts to hold that the
occurrence of a specific type of episode of comprehension of type C entails the
occurrence of a specific U and a specific P. They can also agree that there must
Comprehending Speech / 357
be a causal connection between the U and the P, so that for instance the P causes
the U. The accounts disagree, however, about the converse entailment.
The triggering account holds that the epistemic status of an episode of
comprehension is determined just by the epistemic status of components U and
P, perhaps in conjunction with their causal connection. The integrated account
holds that the epistemic status of comprehension is dependent also upon the way
the faculties responsible for U and P work together to constitute a super-faculty:
optimally functioning comprehension requires the optimal functioning of the
super-faculty.
Since the integrated account holds that episodes of comprehension, given
their epistemic status and function, can be prime, it holds that cases can pattern
in the following way. Optimally functioning comprehension occurs in two cases, α
and β. In both cases, since comprehension episodes entail them, episodes purely
of the understanding and episodes purely of perception occur, with suitable
epistemic statuses and functions to sustain optimally functional comprehension.
However, at a more specific level, it is possible to distinguish the types of
comprehension episode that occur. And it is possible at the more specific level
for the entailed types episodes of understanding to fail to be exactly alike, and
for the entailed types of episodes of perception to fail to be exactly alike.
For instance, case α might differ from case β in perceptual respects due
to slight differences in the volume of the target utterance; and the cases might
differ in understanding respects due to different causal effects of that perceptual
difference. More generally, instances of a generic comprehension episode type
can vary in physical (or other) respects, so that we can distinguish more specific
types of comprehension episodes. And that variation can make for failures of
two such specific type of occurrence to be alike sensorily or intellectually.
Let C 1 and C 2 be two specific types of comprehension episode involving
comprehension of an utterance of ‘There is no largest prime’. The understanding
component of C 1 is U 1 ; and the understanding component of C 2 is U 2 . The
perceptual component of C 1 is P 1 ; and the perceptual component of C 2 is
P 2 . So U 1 and U 2 are physically (or otherwise) different specific ways for the
understanding component of a generic comprehension episode-type to occur;
hence, U 1 and U 2 are intellectually dissimilar. And P 1 and P 2 are physically
different ways for the sensory component of a generic comprehension episode
to occur; hence P 1 and P 2 are sensorily dissimilar. Now case α involves
comprehension episode type C 1 , involving the conjunction of understanding and
perceptual components (U 1 & P 1 ). And case β involves comprehension episode
type C 2 , involving (U 2 & P 2 ).
On the assumption that these components are independent, we can now
consider cases γ that are like α with respect to the understanding (so intellectually
dissimilar from β) and perceptually like β (so perceptually dissimilar from α).
Thus, we can consider a case involving the understanding component of C 1
and the perceptual component of C 2 , so (U 1 & P 2 ). Let’s call this episode
type C ? .
358 / Guy Longworth
According to the integrated account, comprehension episodes with particular epistemic statuses are prime. Hence, on that account, it’s possible for there
to be such a γ case—an episode of type C ? —in which no episode of optimally
functioning comprehension occurs. That is, the integrated accounts holds that it
is possible for there to be such a case in which no episode with the epistemic
status of C 1 or C 2 occurs. To a first approximation (induced by the need to add
in a causal connection between understanding and perceptual components), the
defender of the triggering account must deny that such a case is possible.
Crudely, the idea here is as follows. Since C 1 and C 2 are both optimally
functional, so are their U and P components. Hence, if the optimal function of
comprehension can be factorized into the optimal function of components, plus
causation, we should be able to bring together the understanding component
of C 1 with the perceptual component of C 2 in order to produce an optimally
functional comprehension episode type. The triggering account predicts that, so
long as its components are causally linked in the right way, C ? is guaranteed be
a specific type of optimally functional comprehension. The integrated account
denies this.
If the only function of speech perception was to trigger operations of the
understanding, to cause us to entertain content, then I think the triggering
account might be acceptable. But speech perception has another function: it
serves to determine the identities of our particular sources, the particular sources
for the contents we come to entertain in communication. And this function for
speech perception is essential to knowledge transmission via testimony. As we
shall see, the triggering account is unable to sustain that function.
10.
I shall now sketch an argument that comprehension is epistemically prime.
I shall be continuing to work on the assumption that optimally functional
comprehension must be sufficient, all else being equal, to underwrite the
testimonial transmission of knowledge.
Let me begin with an outline of the sketch. The argument for the epistemic
primeness of (epistemically optimal) comprehension has two stages. The first
stage is an argument that the acquisition by a subject of knowledge via testimony
requires the subject’s optimal functioning to determine the identity of their
sources. It is in effect an argument for the vertical primeness of chains of
testimonial knowledge transmission. Consider a testimonial chain involving
independent subject components and source components. It does not suffice
for knowledge transmission that the source components function properly and
the subject components function properly. It is also required that the testimonial
chain involving subject and source functions properly. For instance, consider a
case, α, in which a subject, A, is optimally functioning and their source, B, is
optimally functioning. Now consider a case, β, in which A and B are optimally
functioning, but A takes testimony from dysfunctional C, rather than from B. In
Comprehending Speech / 359
β, A’s epistemic status insofar as it is independent of the testimony received from
C can be as good as it was in α. And B’s epistemic status can also be as good
as it was in α. Yet, since A acquires testimony from dysfunctional C, rather than
from optimally functioning B, A’s status in β is sub-optimal. But now suppose
that A’s optimal functioning in α is consistent with A’s being in case β—in
particular, that it is consistent with A’s having taken testimony from C rather
than from B. In that case, the optimal functioning of A, even conjoined with the
optimal functioning of the individual who was A’s de facto source, B, would fail
to determine that A was optimally functioning with respect to the cognition A
acquired via testimony. And in that case, optimal functioning of each link in the
testimonial chain would not suffice for the transmission of knowledge.
In short, something about the optimal functioning of the subject must
determine that their source is one particular individual rather than any other,
so that the optimal functioning of the subject is incompatible with their
having another source. Determining the identity of a particular source involves
(a) identifying the source using perception and (b) identifying the perceived
individual as the source for the subject’s entertaining of a piece of communicated
content.
The second stage is an argument that the triggering account can’t meet the
required condition on proper functioning of testimonial chains. A subject could
meet the conditions that the triggering account imposes on proper functioning
consistently with the subject entertaining content on the basis of its presentation
by any of a variety of sources. According to the triggering account, the subject’s
cognition may present an entertaining of content as derived from one or
another source; but optimal functioning of the subject is compatible with that
presentation of source failing to sustain knowledge transmission. That completes
the outline. I shall now run through the two stages of the argument in more detail.
The first stage of argument is designed to show that testimonial knowledge
transmission is possible only if the optimal functioning of some (possibly
improper) part of the testimonial chain determines the identity of the source
component. Let’s assume an account on which the proposed condition fails and
consider the consequences.
We saw above that one function of comprehension of a presentation-as-true,
as of preservative memory, is to furnish its subjects with access to some extant
entitlements to accept the content so presented, if such entitlements are available.
We also saw that not just any extant entitlements are accessible by such a route.
Being told that there is no largest prime by someone whose only right to believe
it is that they have found some fairly large primes cannot furnish one with access
to the entitlements of those who have proven the proposition. Rather, one gains
access only to the entitlements that accrue to the cognition at the source of the
presentation-as-true made available to one through comprehension.
To that extent, comprehension and preservative memory are on a par. But
there are differences—differences that may be obscured by focus upon cases
involving a single potential interlocutor. Unlike memory, which enables access
360 / Guy Longworth
only to one’s own prior entitlements, properly functioning comprehension can
potentially provide access to entitlements residing in any of a variety of sources.
Suppose that Kim and Jo both attempt to present as true the proposition that
there is no largest prime. Kim’s right to believe that proposition takes the form
of a very weak induction, whilst Jo has proved it. In such a situation, the status
of one’s extended entitlement to accept the proposition through comprehension
of its presentation is dependent upon which particular presentation-as-true is the
source of one’s comprehension. So the status of an entitlement gained through
comprehension is hostage to a particular feature of the circumstance in which
that comprehension is produced, the specific source to which that comprehension
is responsive. Not only that, but the particular feature to which one’s entitlement
is hostage is a contingent feature of one’s circumstance, the sort of feature that
it is a proprietary function of perception to reveal.23
We noted earlier that preservative memory also carries information for
its subject concerning contingent features of particular subject matters. And
we saw that its carrying such information does not affect the epistemological
classification of the cognitions it preserves. The reason for this is that the
general function of a subject’s memory is to provide access to only one extended
entitlement, that accruing to the subject’s own earlier cognition. Hence, the
proper function of an exercise of preservative memory determines, independently
of features of the circumstance extrinsic to the memorial cognition or features
specific to that particular cognition, the particular source of the extended
entitlement to which it allows access. Since the Acceptance Principle presupposes
an a priori entitlement to rely upon the proper function of cognitive resources
and faculties, it sustains an a priori entitlement to presume upon its deliverances
having a source in one’s own earlier cognitions. It serves, thereby, to provide a
non-empirical link with one’s own extended entitlement despite the fact that, in
so doing, it carries information about a particular, contingent subject matter.
By contrast, the function of the understanding in facilitating the interpersonal preservation of entitlement fails to determine which particular extended
entitlement its operations preserve. The understanding may function perfectly
in delivering to one cognizance of a presentation-as-true sourced either with
Kim or with Jo, or indeed with anyone else. In the case of comprehension,
then, the reach of a priori entitlement is limited to sustaining a right of access
to whichever extended entitlement happens to reside with its source. Which
source that is, and hence which extended entitlement, is beyond its purview. Of
course, the Acceptance Principle provides an entitlement to presume upon the
proper functioning, hence entitlement, of one’s interlocutor, whoever they are.
The present point is only that which person the cash value of that transpersonal
entitlement depends upon, hence which extended entitlement, cannot be fixed by
the proper functioning of the understanding.
In assessing the present account with respect to the first stage of the
argument, let us begin by reverting once more to the analogy with memory.
In that case, it is indifferent whether we view what fixes one’s source as a suitable
Comprehending Speech / 361
sort of causal connection or rather as the properly functioning perspective of the
subject. This is because both potential determinants converge on a single source,
the subject themselves. As we have seen, the case of communication is quite
different. Here, each of the various presentations that the subject comprehends
may have a different causal pedigree. But nothing in the subject’s own perspective
discriminates the sources of those various presentations. From the subject’s
perspective, it is as if each presentation has the same source. In effect, the account
makes the subject’s perspective on communicated presentations an outwardfacing analogue of memorial, or otherwise first-personal, presentations: for the
subject, it is as if each of her presentations is either received, so communicated
from without, or self-initiated. No more fine-grained discrimination need be
made available to them by the optimal functioning of the faculty responsible for
comprehension.
The potential divergence between perspectival determination of source and
merely causal determination of source—that is, between determination that is
internal, and determination that is external, to optimal functioning—might not
matter in a world in which interlocutors were guaranteed to function perfectly.
In such a world, any presentation-as-true received by a subject would have
the same epistemic status—would be knowledgeable—and the only epistemic
differences between sources would reside in their knowledge or ignorance of
particular facts. There would be no risk, for example, of conflict amongst the
deliverances of comprehension and, so, of a need to privilege some of those
deliverances over others. In a world like that, distinctions amongst sources would
be irrelevant to the status of the various extended entitlements one gained access
to via comprehension. But our world is not like that. In a world like ours,
populated by interlocutors that can be faulty or capricious, responsible epistemic
practice often demands sensitivity to epistemic differences amongst one’s sources.
Unlike the inhabitants of a world of epistemic angels, we are in some danger of
being confronted, through comprehension, by conflicting presentations. For our
purposes, then, it is crucial that we should be able, at least in principle, to
differentiate our sources and, where appropriate, exploit sensitivity to conditions
defeating an initially global presumption of trust in order to privilege one source
over another.
In order to fix ideas, assume that the optimal functioning of the subject
fails to determine the identity of their source and consider the following oversimplified cases. First, assume normal conditions of access to the sources of
presentations and consider the following pairs of cases. In case α, one hears
Jo say that there is no largest prime and one hears Kim say that there is a
largest prime. One’s initial entitlement to accept what both Jo and Kim present
is undermined. But one is in a position to assess one’s sources. In particular,
one is in a position to see whether one’s pro tanto entitlement to accept the
presentation of either source is subject to defeat and so to regain entitlement to
accept the other source’s presentation. In case β, one hears Jo say that there is no
largest prime and then say that there is a largest prime. In this case one’s initial
362 / Guy Longworth
entitlement to trust Jo is undermined. But in this case, there is no question of
exploiting conditions defeating one’s entitlement to trust Jo in order to win back
one’s entitlement to trust Jo. Of course, that assessment is too simplistic; there are
various ways of differentiating the statuses of different presentations stemming
from the same source. But the structural point survives into more realistic
cases.
Second, then, assume that the particular sources of different received
presentations are inaccessible from one’s perspective—that is, that they are not
determined by one’s optimal functioning. It should be obvious that if one were
in those conditions, then it would be for one as if one were in a case like β
whether or not one were in fact in a situation like α. There would be no way for
one to exploit conditions defeating one’s entitlement to accept a presentation
in order to regain entitlement to accept the other deliverances of receptive
comprehension. Within the simplified conception of cases we are currently
working with, the following situation is possible. One’s comprehension faculty is
functioning perfectly, so that its products reveal to one genuine presentations-astrue. One comprehends a presentation-as-true by an impeccable source so that,
were one in a position rationally to accept that presentation, one would thereby
acquire knowledge. Yet one also comprehends a conflicting presentation and so,
because one has no means of differentially assessing the two presentations, one
cannot attain a position in which one is entitled to accept either. One’s own
faculties are functioning perfectly and so are (one of) one’s interlocutors’. From
the perspective of the present account, one is in the optimal position sustained by
the functions of internal and external components of an extended entitlement. In
spite of this, one’s principled ignorance of source renders one unable to exploit
that position.
The structural problem here is due to the way that the account we
are considering decomposes an extended entitlement into internal—subjectcentred—components and external—interlocutor-centred—components, so that
only a causal connection holds the components together. Still assuming these
conditions, consider another case. Jo is perfectly functioning and knows that
there is no largest prime; Kim functions imperfectly and, while believing the
proposition, does so on poor inductive grounds. Both Jo and Kim attempt to
present one with the proposition that there is no largest prime. One’s comprehension faculty works perfectly, and delivers to one the right presentation. As luck
would have it, the causal source of one’s comprehension of the proposition that
there is no largest prime is Jo’s attempt, rather than Kim’s. Hence, on an account
that sustains the factorization of testimonial chains into subject and source
components, one should be in a position to know that there is no largest prime.
But compatibly with the perfect functioning of the internal component of one’s
extended entitlement—one’s comprehension faculty—and the perfect functioning
of what is, in fact, the external component of one’s extended entitlement—
Jo’s understanding, reason, and presentational faculties—one might easily have
lacked the extended entitlement one in fact possesses. For it might easily have
Comprehending Speech / 363
been the case that Kim’s presentation trumped Jo’s in serving as the causal
source of one’s comprehension. At the centre of Burge’s optimal function based
epistemology, causal happenstance, external to the optimal functioning of the
faculties involved, would be required to play a critical role.24
That completes the sketch of the first stage of argument for the horizontal
primeness of comprehension: the testimonial chain consisting of the source of
testimony and the subject who aims to acquire knowledge from that testimony
cannot be factorized into independent components. Rather, something about the
participants in the chain, and their interaction, must determine the identity of
the chain, by determining the identity of the source. Plausibly, responsibility for
this resides with the optimal functioning of the subject. The second stage of the
argument is designed to show that the responsibility cannot be met according to
the triggering account of optimal functioning. I shall begin by outlining how the
integrated account can meet the required condition.
The integrated account sustains the following account. Comprehension
derives from the operation of a super-faculty involving both perception and
the understanding. Optimal functioning of the super-faculty determines an
association between a speech episode and an entertained content, an association
fixed by the episodic output of exercises of the super-faculty. If the super-faculty
responds to A’s assertion by enabling a cognition associating A’s utterance
with the content that A thereby expressed, then it is incompatible with the
optimal functioning of the super-faculty that the understanding component—
the subject’s entertaining of the content that there is no largest prime—was
derived from B’s assertion. Determination of source is fixed, first, by optimal
function of perception and, second, by optimal function of a super-faculty
involving perception and the understanding. Perception serves to identify a
particular individual as source; the super-faculty informs the subject that the
perceived individual is responsible for the content that they come to entertain
through comprehension. The outputs of the super-faculty are cognitions to the
effect that that particular utterance gave expression to a particular entertained
content, perhaps taking the form of knowledge that that utterance expressed
this content. To be sure, puzzles remain concerning the way in which sensory
perception underwrites the identification of particular sources, and the proper
characterization of integrated outputs.25 But there is no immediate reason to
think that these puzzles indicate problems of principle rather than questions of
detail. How does Burge’s triggering account fare with respect to the determination
of source?
I shall assume, for reasons sketched in the first stage of the argument, that
the understanding cannot, without supplementation from sensory perception,
determine the sources of the presentations-as-true whose entertaining it enables.
However, the triggering view has a variety of other resources at its disposal.
In particular, the triggering account can appeal to the individual optimal
functioning of sensory perception and the understanding, respectively, in order
to cope with some types of problem case.
364 / Guy Longworth
Consider a first problem case, involving an operation of one’s understanding
induced by a sensory hallucination as of an appropriate source, but where there
is no appropriate external source.26 Suppose, for instance, that one hallucinated
an utterance of ‘There is no largest prime’, perhaps at the end of a very
long mathematics conference. Here, there would be no genuine source for the
operation of the understanding, which would therefore trivially fail to sustain
the transmission of knowledge. And yet the imagined case might appear to be
one in which one’s faculty of understanding is optimally functioning, so that at
the very least, the functioning of sensory perception would have to play a role in
determining the epistemic status of the output.
However, it is plausible that the triggering account has the resources
to respond to this type of case without appealing to the status of sensory
functioning. It is at least consistent with the triggering account that the only
power of sensory perception that the understanding lacks is the power to
determine for the subject the particular individuals and properties with which
they are confronted. So, although the proper functioning of the understanding
cannot sustain the identification of any particular source, since that would
be a perceptual task, it can involve accurately depicting the presentation of
content by an appropriate source. That is, it is open to the triggering account
to view the optimal functioning of the understanding as determining whether
or not there is at least one suitable presentation of content to which its
operations are responsive. Hence, the triggering account has the resources
to show that an entertaining of content based upon this sort of perceptual
hallucination would involve a malfunction, and in particular a malfunction of the
understanding.27
The triggering account also has resources to deal with a second sort of case.
Suppose that the operation of the understanding, the entertaining of content,
was induced by a veridical perceptual hallucination? That is, suppose the subject
has a hallucination as of an individual uttering ‘There is no largest prime’ and, on
the basis of that hallucination, entertains the content. And suppose that, in fact,
the hallucination was caused, via some tortuous route, by someone’s uttering the
words ‘There is no largest prime’. In that case, we could not trace the problem
to a malfunction of the understanding, since there was in fact an appropriate
source, and yet it would be a case in which the source was not determined in an
epistemologically appropriate way.
Again, the triggering account has resources to deal with this sort of case.
The triggering account is allowed to appeal to the optimal functioning of each
of the faculties responsible, causally or constitutively, for episodes of seeming
comprehension, considered independently of their roles in the system as a whole,
or a super-faculty thereof. In this case, the episode of seeming comprehension was
triggered by the operations of the sensorium, via a sensory hallucination. And
although such hallucination need involve no malfunction in the understanding,
it would obviously involve a perceptual malfunction. So, the triggering account
has the resources to deal with the second sort of case.28
Comprehending Speech / 365
The capacity of the triggering account to cope with the first two cases
depends upon its being possible to trace the malfunctions involved in those
cases to malfunctions in the operations of one or another of the individual
faculties involved in the production of seeming comprehension. That suggests
that we might be able to present a problem case of a similar sort involving the
connections between the understanding and perception such that the triggering
account cannot explain the malfunction it involves.
Let’s call any piece of perception or cognition that fails to be suitably (i.e.
epistemically) responsive to its causal source an hallucination. There appears
then to be space for a purely intellectual hallucination. This would involve the
understanding producing an entertaining of content (as though received from
perceived speech), but where that entertaining was insensitive to operations of
perception in something like the way that perceptual hallucination is insensitive
to the external world. Of course, if no appropriate source were present in
the subject’s environment, then this would amount to a malfunction in the
understanding. And as we have seen, the triggering account has the resources
to deal with cases that can be traced to such malfunction. However, veridical
intellectual hallucinations also appear to be possible. These would be intellectual
hallucinations wherein an entertaining of content in fact responded to the
expression of that content in the environment, and perhaps to the sensory
perception of an utterance that expresses that content, but in a way that was
insensitive (or insufficiently sensitive) to the particular perceptual cause of the
entertaining. Perhaps, for example, such cases might involve veridical perception
of an appropriate source together with an appropriate entertaining of content, but
where the perceived source diverged from the causal source of the entertaining.
For instance, if we follow Burge in allowing that an entertaining of content might
be caused in a way that bypasses the typical perceptual route, we might construct
such a case by simply reinstating perceptual access to the causal source, without
fixing up an appropriate connection between the perception and the entertaining.
And intellectual hallucination of that sort would be consistent with optimal
functioning according to the triggering account, since it would be consistent with
perfect sensory and intellectual functioning. But then optimal functioning would
be consistent also with the entertaining having any of a variety of perceived, or
non-perceived, sources. Hence, it would be consistent with a failure of optimal
functioning to determine the particular source of the entertaining and so, given
the first stage of the account, with a failure to acquire knowledge from testimony
in conditions of optimal functioning at all links in the testimonial chain.
That completes the sketch of the second stage of the argument and so the
argument as a whole. In the context of an account of epistemology based upon
optimal functioning, the upshot is that episodes of comprehension suitable to
underwrite ordinary cases of testimonial transmission of knowledge—suitable,
that is, independently of special epistemic crutches like inductive inference—are
prime. Their epistemic status cannot be factorized into the independent statuses
of intellectual and sensory components.
366 / Guy Longworth
Thus far, I have been working on the simplifying assumption that optimal
functioning of a faculty ought to supply a guarantee of maximal epistemic
status (so, for instance, must guarantee truth in order to guarantee factive
epistemic statuses like knowledge). However, the basic complaint against the
triggering account is structural, and so survives into a variety of alternative
conceptions of optimal functioning. For instance, it is a fairly straightforward
matter to show that a recipient of testimony and the causal source of that
testimony can both meet reliability or safety conditions with respect to their
functioning independently of their involvement in any particular testimonial
chain while the recipient fails to meet those conditions with respect to beliefs
acquired through the testimony, due to a lack of reliability or safety in their
connection with the reliable source. One sort of case would involve a recipient
of knowledgeable testimony who might easily have taken testimony from a
competing source whose testimony was unreliable or non-veridical. Similarly, it
is a fairly straightforward matter to show that a subject’s sensory and intellectual
faculties might be individually reliable whilst their joint operation was unreliable.
For instance, the subject might reliably track the production of speech sounds
and reliably entertain contents presented as true, and yet be unreliable with
respect to the association of speech sounds and presented contents. One sort
of case would involve a subject whose normal perceptually based sensitivity to
contents expressed in English was easily trumped by a non-perceptual sensitivity
to presentations in French, so that in situations involving competition between
potential English speaking and French speaking interlocutors, they would take
contents presented in French to have been presented by English utterances. The
issues here warrant further discussion, but I do not think that the discussion
would materially affect the conclusion of the argument.
The upshot of the argument is that faculty operations that are able to subserve elements of cognition required for knowledge transmission must involve
the operations of a sense-perceptual faculty. Hence, operations with that status
even pro tanto must be operations of a partly perceptual faculty. Due to the
partly perceptual nature of their source faculty, the cognitions delivered by that
faculty are therefore partly sense-perceptual.
11.
How did Burge arrive at an account that allows for the sort of fracture within
a subject’s extended entitlement detailed above? A natural diagnosis would be
divided attention. Burge’s overall account gives his Acceptance Principle two
applications: first, in entitling one to accept as genuine putative presentationsas-true; and second, in entitling one to accept genuine presentations-as-true.
Individual attention to either application does not require one to consider how
they fit together. Speech-perceptual aspects of episodes of comprehension have
no obvious function in either application of the Principle. A central function
Comprehending Speech / 367
of the perceptual aspects, I have argued, is to integrate the applications. Hence,
blinkered attention to individual applications of the Principle would not reveal
the epistemological function of the perception of speech.
Consider, in this light, Burge’s four negative arguments against perceptionbased models of comprehension. First, it should be obvious that Burge’s
argument against a pure perceptual model has no force against the integrated
model of comprehension. Second, the force of Burge’s thought experiment,
based around speech-blind comprehension, depends ultimately, as noted earlier,
upon whether, and in what ways, the epistemological functions of episodes of
comprehension require them to involve sensory aspects. Isolated consideration of
episodes of comprehension can fail to reveal such a function. Hence, from that
perspective, the case Burge asks us to consider can seem genuinely possible.
It is only when one considers the role of comprehension in facilitating the
communication of knowledge that the functional poverty of comprehension in
the case that Burge presents is revealed.
Third, there is the consideration that functional comprehension can withstand perceptual malfunction. As a version of Descartes’ observation, the
consideration is ineffective because sub-serving communication requires only
perceptual registration of a speech episode, not revelation of its detail. As an
observation about the non-perceptual function of comprehension, the response
to Burge’s thought experiment stands. Some, though not all, functions of
comprehension, are indifferent to perceptual performance.
The fourth consideration, the insouciance of ordinary epistemic practice,
actively supports a partly perceptual model. Suppose that one were to ask how
Kim knows that George would prefer Adnams, perhaps being open to the answer:
George told her. Then—unless one was in the grip of the idea that any perceptual
episode must be exhaustively so—one might find both illuminating and satisfying
the answer: She heard (or saw) him say it.
12.
I have argued that the knowledge preserving function of comprehension
requires that the episodes of cognition constitutive of comprehension have a
dual nature: they are products partly of the understanding and partly of sense
perception. I have not argued that the function of speech perception is to
supply justificatory elements in the extended entitlements to which episodes of
comprehension provide access. That result is the upshot of a particular feature
of Burge’s account, about whose credentials I am neutral. On Burge’s account,
comprehension supplies essential elements in an extended entitlement. So, if
Burge were to accept the conception of comprehension presented here, without
amending other elements of his overall account, he would be forced to allow that
sense perception is implicated, through its role in the constitution of episodes
of comprehension, in supplying elements with justificatory force. He would have
368 / Guy Longworth
to accept that the transmission of a priori knowledge as a priori is not possible
except, perhaps, amongst angels. Nonetheless, my basic disagreement with Burge
does not concern the justificatory role of perception in communication, with
respect either to episodes of comprehension or to later cognitions that gain their
epistemic status through comprehension. Our basic disagreement concerns the
nature of episodes of comprehension able to function in knowledge transmission:
I have argued, while Burge denies, that such episodes are partly the work of sense
perception.
As noted at the outset, there may be a way of retaining much of the
overall architecture of Burge’s account compatibly with a partly sense-perceptual
conception of comprehension. Rather than viewing episodes of comprehension
as adding to the justificatory force of an entitlement that they help to constitute,
one might try for an account on which the function of those episodes is itself
preservative. I have insufficient space to pursue such an account here. But I want
to conclude by noting two points in its favour.
First, we are already in the market for a distinction between access
conditions—a form of enabling condition—and elements in an entitlement. With
the distinction in hand, no obvious difficulty attends treating comprehension
as part of an access condition. Of course, the propositional content so comprehended does form part of an entitlement, the entitlement to which one’s
interlocutor has access. Hence, according to an access condition conception, the
propositional content itself might be thought of as possessing a dual role, as
both inlet to, and constituent of, an entitlement. But it would be a mistake to
think that that duality—or better, neutrality—must impact upon the function of
particular cognitions with that propositional content. In particular, an episode
of comprehension may have only the access role with respect to a subject’s
engagement with a particular entitlement, despite the comprehended content
itself playing other roles.
Second, there are independent reasons, given Burge’s characterization of
justificatory involvement in an entitlement, for refusing to give comprehended
elements such a role. According to Burge, an element forms part of an entitlement
only if it plays an essential role in constituting or enhancing the justificatory
force of that entitlement. But, in that case, to suppose that comprehension
supplies for a subject elements that constitute or enhance the force of their
entitlement—in any sense except the trivial one involved in thinking of epistemic
role as a feature of propositional contents in the abstract—is to suppose either
of two things. It is to suppose either, first, that in the absence of the comprehension, the entitlement was incomplete or, second, that the comprehension
enhanced the force of the entitlement. The first supposition is incompatible
with one’s interlocutor possessing an adequate entitlement—i.e., knowledge; the
second is incompatible with comprehension putting one in an equal, or lesser,
epistemic position than one’s interlocutor. Since neither consequence is acceptable, I hazard that comprehension may sometimes have a solely preservative
function.29
Comprehending Speech / 369
Notes
1. This is a common theme in the work of modern philosophers. See e.g. Berkeley,
1993: IV; Reid, 1997: passim. See also Locke, 1979: bk. II, ch.ix, sect. 9. For an
excellent recent discussion of Descartes’ view on the role of the sensorium and
intellect in perceptual cognition, see Simmons, 2003. For discussion of empirical
aspects of the phenomenon to which Descartes draws attention see Coltheart,
1999.
2. See Longworth, 2008, where I assume that ordinary understanding must involve
awareness of such associations.
3. For useful discussion of Burge’s view, see Bezuidenhout, 1998; Christensen &
Kornblith, 1997.
4. I shall use ‘speech’ and its analogues as a cover term for the expression of
thoughts, etc., by articulate action, including signing and facial contact, so that
sensorily perceiving here may be a matter of hearing, seeing, or feeling.
5. I follow Burge in distinguishing between entitlement and justification. Both an
entitlement and a justification consist of elements essential to the epistemic status
of a piece of cognition. But elements of an entitlement, unlike a justification,
need not be available to the subject in order that they possess the status it
constitutes.
6. I should emphasize that Burge hopes to secure only the possibility of this sort
of comprehension consistently with knowledge transmission. He does not hold
that entitlements to all instances of comprehension are free from sense-perceptual
elements.
7. See especially Burge, 2003a: 532–37, 2003b: 307. For discussion, and defence,
of this approach to a restrained Externalism, see, e.g., Dretske, 2000; Peacocke,
1999, 2002; Plantinga, 1993a, 1993b; Sosa, 1991; Williams, 2000.
8. In addition to earlier references to Burge, see his 2003a for extended discussion
of subjects’ entitlement to presume upon the proper function of their faculties.
9. Or, perhaps, something if Burge is right about knowledge acquired from certain
types of computer. See his 1998a.
10. For discussion, see Edwards, 2000; Faulkner, 2000.
11. I follow Burge in distinguishing between a subject’s full entitlement for a piece of
cognition, including contributions made by their own past, or other subjects—
what Burge calls the subject’s extended entitlement—and the subject’s own
proprietary contribution to their possession of that entitlement. See e.g. Burge,
1998a: 5–6.
12. Edwards, 2000: 130.
13. For discussion, see Burge, 2003b; Wiggins, 1992, 2001: 198–244; Martin, 2001.
14. As Burge points out, we need not view what I have labelled transitions as
genuine psychological acts: ‘[engaging]. . .commitment involved in. . .belief is not
necessarily an act, but rather the absence of a withholding of belief.’ (Burge,
2003a: 542)
15. For additional arguments against inferentialist views, see McDowell 1980, 1998a,
1998b; Recanati, 2002. For some defensive discussion, see Dummett, 1986, 1987.
16. For empirical discussion of some cases of putative comprehension without
conscious awareness of signal, see Potter, 1999; Saffran & Martin, 1999.
370 / Guy Longworth
17. Burge attempts to account for the availability to the subject of this distinction
through their capacity, through intellection, to tell whether they are the agent
of a particular presentation. See Burge, 1998a: 18; 1998b: 262–70 and O’Brien,
2005.
18. For discussion of whether knowing how should be thought of as a form of
cognition, see Stanley & Williamson, 2001; Rumfitt, 2003; Snowdon, 2004.
19. For discussion of this sort of experience, and the indissolubility of its aspects, see
Wollheim on seeing-in, in his 1980, 1987, 2003, and also Walton, 1990.
20. Such effects appear to play a crucial role in ordinary speech perception. See
Massaro, 1998.
21. Burge, 1993a: 478.
22. See Williamson, 1998, 2000, 2006.
23. As Burge puts it,
The point and function of the perceptual system is to put the perceiver in
touch with particular situations, particular objects or events, or particular
instances of properties or relations. (Burge, 2003a: 523)
24.
25.
26.
27.
28.
For discussion of this function of perception, see Burge, 1986, 1993b, 2005;
Brewer, 1999; Martin, 2002; McDowell, 1986; Soteriou, 2000.
It would, of course, be possible to add purely Externalist—e.g. reliabilist—
elements to the system in order to shore up the division between internal and
external components. In the present context, the move would be inefficacious: absent reason to think that sensorily based belief is subject to different requirements,
we would lose the ability to distinguish it from comprehension. And, anyway, for
three reasons such a manoeuvre should be made only as a last resort. First—a
prima facie ad hominem point—, Burge’s deployment of the Acceptance Principle,
in order to underpin rational dependence upon proper function, forms part of
an attempt to domesticate such Externalist requirements, to bring them within
the purview of norms fixed by the natures of psychological and epistemological
kinds. See especially Burge, 2003a: 532–37; 2003b: 307. So allowing purely
Externalist factors to play a fundamental role in the system would be in serious
tension with Burge’s basic project. Second, the recent history of epistemology
supports the view that Burge’s project is, to that extent, well motivated. It is far
from obvious that epistemic goods can be reconstructed solely on the basis of
independent, Externalist notions of reliability, safety, and their ilk. Philosophers
have thus far been unable even to provide clearly acceptable reconstructions of
the relevant external standards. And even partial success on that score has been
marred by widespread concerns about the capacity of those external standards
to reconstruct, without some form of Internalist supplementation, the genuinely
epistemic goods that they condition. Third, the availability of a more Internalist
account of the integration of internal and external elements based around
materials made available in our ordinary comprehension of content renders the
manoeuvre otiose.
For discussion of some complexities on the sound side, see Nudds, 2001.
See Christensen & Kornblith, 1997.
Cp. Burge, 1997.
Cp. Burge, 1997.
Comprehending Speech / 371
29. Thanks to Thomas Crowther, Ken Gemes, Susan James, Mark Eli Kalderon,
Tony Marcel, Christian Nimtz, Milena Nuti, David Papineau, Sarah Patterson,
Greg Scherkoske, Gabriel Segal, Scott Sturgeon, Barry C. Smith, Neil Smith,
and Mark Textor for discussion. Thanks also to audiences at a joint session of
the SPP and ESPP, Barcelona, including Francois Recanati and David Owens,
and a Departmental Seminar at Bristol University, including Alexander Bird,
Jessica Brown, James Ladyman, Finn Spicer, and Andrew Woodfield. Special
thanks to Jennifer Hornsby, Hemdat Lerman, and Matthew Soteriou for extended
discussion.
References
Berkeley, G. (1993) Alciphron, or the Minute Philosopher in Focus, D. Burman ed., London:
Routledge.
Bezuidenhout, A. (1998) ‘Is Verbal Communication a Purely Preservative Process?’, Philosophical Review, 107, 2: 261–88.
Brewer, B. (1999) Perception and Reason, Oxford: Oxford University Press.
Burge, T. (1986) ‘Cartesian Error and the Objectivity of Perception’, in P. Pettit and J. McDowell
(eds.) Subject, Thought and Context, Oxford: Clarendon Press.
Burge, T. (1993a) ‘Content Preservation’, Philosophical Review, 102: 457–488.
Burge, T. (1993b) ‘Vision and Intentional Content’, in R. van Gulick and E. Lepore (eds.) Searle
and his Critics, Oxford: Blackwell.
Burge, T. (1997) ‘Interlocution, Perception, and Memory’, Philosophical Studies, 86: 21–47.
Burge, T. (1998a) ‘Computer Proof, Apriori Knowledge, and Other Minds’, Philosophical
Perspectives, 12: 1–37.
Burge, T. (1998b) ‘Reason and the First Person’, in B. C. Smith, C. MacDonald, and C. Wright
(eds.) Knowing Our Own Minds: Essays on Self-Knowledge, Oxford: Clarendon Press.
Burge, T. (1999) ‘Comprehension and Interpretation’, in Lewis E. Hahn (ed.) The Philosophy of
Donald Davidson, Chicago and La Salle, Illinois: Open Court.
Burge, T. (2003a) ‘Perceptual Entitlement’, Philosophy and Phenomenological Research, LXVII,
3: 503–48.
Burge, T. (2003b) ‘Memory and Persons’, Philosophical Review, 112, 3: 289–337.
Burge, T. (2005) ‘Disjunctivism and Perceptual Psychology’, Philosophical Topics, 33: 1–78.
Christensen, D. and Kornblith, D. (1997) ‘Testimony, Memory, and the Limits of the A Priori’,
Philosophical Studies, 86: 1–20.
Coltheart, V. (ed.) (1999) Fleeting Memories: Cognition of Brief Visual Stimuli, Cambridge,
Mass.: MIT Press.
Descartes, R. (1664/1985) ‘The World or Treatise on Light’, in J. Cottingham, R. Stoothof,
and D. Murdoch (trans. and eds.) The Philosophical Writings of Descartes, vol. I,
Cambridge: Cambridge University Press: 79–98.
Dretske, F. (2000) ‘Entitlement: Epistemic Rights Without Epistemic Duties?’, Philosophy and
Phenomenological Research, 60: 591–606.
Dummett, M. A. E. (1986) ‘Thought and Language’, in J. Vuillemin ed. Mérites et limites des
méthodes logiques en philosophie, Paris: J. Vrin.
Dummett, M. A. E. (1987) ‘Reply to McDowell’, in B. Taylor ed. MichaelDummett: Contributions to Philosophy, Dordrecht: Nijhoff.
Edwards, J. (2000) ‘Burge on testimony and memory’, Analysis, 60.1: 124–31.
Faulkner, P. (2000) ‘The Social Character of Testimonial Knowledge’, Journal of Philosophy,
97, 11: 581–601.
372 / Guy Longworth
Locke, J. (1979) An Essay Concerning Human Understanding, P. H. Nidditch ed., Oxford:
Clarendon Press.
Longworth, G. (2008) ‘Linguistic Understanding and Knowledge’, Nôus, 42, 1: 50–79.
Martin, M. G. F. (2001) ‘Out of the Past: Episodic Recall as Retained Acquaintance’, in C. Hoerl
and T. McCormack (eds.) Time and Memory: Issues in Philosophy and Psychology,
Oxford: Clarendon Press.
Martin, M. G. F. (2002) ‘Particular Thoughts and Singular Thoughts’, in A. O’Hear (ed.) Logic,
Thought and Language, Cambridge: Cambridge University Press.
Massaro, D. W. (1998) Perceiving Talking Faces, Cambridge, Mass.: MIT Press.
McDowell, J. (1980) ‘Meaning, Communication, and Knowledge’ in Z. van Straaten (ed.)
Philosophical Subjects: Essays Presented to P. F. Strawson, Oxford: Clarendon
Press.
McDowell, J. (1986) ‘Singular Thought and the Extent of Inner Space’, in P. Pettit and J.
McDowell (eds.) Subject, Thought and Context, Oxford: Clarendon Press.
McDowell, J. (1998a) ‘A Plea for Modesty’, in his Meaning, Knowledge and Reality, Cambridge,
Mass.: Harvard University Press.
McDowell, J. (1998b) ‘Another Plea for Modesty’, in his Meaning, Knowledge and Reality,
Cambridge, Mass.: Harvard University Press.
Merlau-Ponty, M. (1973) Consciousness and the Acquisition of Language, Evanston: Northwestern University Press.
Nudds, M. (2001) ‘Experiencing the Production of Sounds’, European Journal of Philosophy, 9,
2: 210–29.
O’Brien, L. (2005) ‘Self-knowledge, Agency and Force’, in Philosophy and Phenomenological
Research, 71, 3: 580–601.
Peacocke, C. (1999) Being Known, Oxford: Oxford University Press.
Peacocke, C. (2002) ‘Three Principles of Rationalism’, European Journal of Philosophy, 10:
375–397.
Plantinga, A. (1993a) Warrant: The Current Debate, Oxford: Oxford University Press.
Plantinga, A. (1993b) Warrant and Proper Function, Oxford: Oxford University Press.
Potter, M. C. (1999) ‘Understanding Sentences and Scenes: the Role of Conceptual Short Term
Memory’, in V. Coltheart (ed.) (1999).
Recanati, F. (2002) ‘Does Linguistic Communication Rest on Inference’, Mind & Language, 17:
105–126.
Reid, T. (1997) An Inquiry into the Human Mind on the Principles of Common Sense, D. R.
Brookes ed., Edinburgh: Edinburgh University Press.
Rumfitt, I. (2003) ‘Savoir Faire’, Journal of Philosophy, 100: 158–6.
Saffran, E. M. and Martin, N. (1999) ‘Meaning but not Words: Neuropsychological Evidence
for Very Short Term Conceptual Memory’, in V. Coltheart (ed.) (1999).
Simmons, A. (2003) ‘Descartes on the Cognitive Structure of Sensory Experience’, Philosophy
and Phenomenological Research, LXVII, 3: 549–579.
Snowdon, P. (2004) ‘Knowing How and Knowing That: A Distinction Reconsidered’, Proceedings of the Aristotelian Society, 104: 1–32.
Sosa, E. (1991) ‘Intellectual Virtue in Perspective’, in his Knowledge in Perspective: Selected
Essays in Epistemology, Cambridge: Cambridge University Press.
Soteriou, M. (2000) ‘The Particularity of Experience’, European Journal of Philosophy, 8: 173–
89.
Stanley, J. and Williamson, T. (2001) ‘Knowing How’, Journal of Philosophy 98: 411–44.
Walton, K. L. (1990) Mimesis as Make-Believe, Cambridge, Mass.: Harvard University Press.
Wiggins, D. (1992) ‘Remembering Directly’, in J. Hopkins and A. Savile (eds.), Psychoanalysis,
Mind and Art: Essays for Richard Wollheim, Oxford: Blackwell.
Wiggins, D. (2001) Sameness and Substance Renewed, Cambridge: Cambridge University Press.
Williams, M. (2000) ‘Entitlement: Epistemic Rights Without Epistemic Duties?’, Philosophy and
Phenomenological Research, 60: 607–12.
Comprehending Speech / 373
Williamson, T. (1998) ‘The Broadness of the Mental: Some Logical Considerations,’ J. E.
Tomberlin ed. Language, Mind and Ontology: Philosophical Perspectives, 12: 389–
410.
Williamson, T. (2000) Knowledge and its Limits, Oxford: Oxford University Press.
Williamson, T. (2006) ‘Can Cognition be Factorised into Internal and External Components?’
R. Stainton ed. Contemporary Debates in Cognitive Science, Oxford: Blackwell.
Wollheim, R. (1980) Art and its Objects, 2nd edn., Cambridge: Cambridge University Press.
Wollheim, R. (1987) Painting as an Art, Princeton, N. J.: Princeton University Press.
Wollheim, R. (2003) ‘What makes representational painting truly visual?’, Proceedings of the
Aristotelian Society Suppl. Vol.: 131–47.