Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
1
[PLEASE DO NOT QUOTE OR CIRCULATE WITHOUT PERMISSION]
Context Principle and Contextual Definitions
in Jeremy Bentham
Silver Bronzo
11/24/08
I
The earliest formulation of the context principle is usually located in Frege’s
Grundlagen, where we are told that we should never “ask for the meaning of a word in
isolation, but always in the context of a proposition”, and that “[o]nly in the proposition
have the words really a meaning”.1 According to some authors, however, a version of the
context principle substantially similar to that endorsed by Frege is already present in the
work of Jeremy Bentham. The most illustrious of such authors is Quine.2 In a number of
different essays, he attributes to Bentham two crucial innovations in semantics: (1) the
“recognition of contextual definitions” and (2) the idea that sentences, rather than terms,
are “the primary vehicle[s] of meaning”. Quine presents these two ideas as intimately
connected with one another, and as jointly constituting a genuine “revolution” in
semantics that had far-reaching consequences for the developments of twentieth-century
analytic philosophy:
1
Frege, Foudations of Arithmetics, p. x, 71.
Other authors who, like Quine, see in Bentham some anticipations of Frege’s context principle are P.M.S
Hacker (“The rise of Twentieth Century Analytic Philosophy”, p. 259) and R. Harrison (Bentham, London:
Routledge and Kegan Paul, 1983, p. 64-68).
Quine probably knew about Bentham’s logical writings through C.K. Ogden’s book, Bentham’s Theory of
Fictions, which he explicitly refers to. Ogden’s book consists in a long introduction and in excerpts from
Bentham’s collected works. Ogden, between the two wars, was proud of contributing to the “rediscovery”
of Bentham’s logic and theory of meaning. He also recognized the similarities between Bentham’s thought
and some development of twentieth century analytic philosophy, most notably between Bentham’s theory
of fictions and Russell’s appeal to the notion of “incomplete symbols”. In the same years J. Wisdom wrote
a short book entirely dedicated to the discussion of this topic (Interpretation and Analysis in Relation to
Bentham’s Theory of Definition).
2
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
2
Bentham’s [innovation] was the recognition of contextual definition, or what he
called paraphrasis. He recognized that to explain a term we do not need to specify
an object for it to refer to, nor even specify a synonymous word or phrase; we
need only show, by whatever means, how to translate the whole sentences in
which the term is to be used. […] This idea of contextual definition, or
recognition of the sentence as the primary vehicle of meaning, was indispensable
to the ensuing developments in the foundations of mathematics. It was explicit in
Frege, and it attained its full expression in Russell’s doctrine of descriptions as
incomplete symbols.3
Contextual definitions precipitated a revolution in semantics. […] The primary
vehicle of meaning is seen no longer as the word, but as the sentence. Terms, like
grammatical particles, mean by contributing to the meaning of the sentences that
contain them. […] It was the recognition of this semantic primacy of sentences
that gave us contextual definitions, and vice versa. I attribute this to Bentham.
Generations later we find Frege celebrating the semantic primacy of sentences,
and Russell giving contextual definitions its fullest exploitation in technical
logic.4
In the second passage, Quine suggests that the idea of contextual definitions and the idea
of the semantic primacy of sentences mutually support one another; in the first passage,
he even seems to present the two ideas as basically equivalent. Moreover, in these
passages Quine draws a clear line of continuity between Bentham’s use of paraphrastic
definitions, Frege’s “celebration of the semantic primacy of sentences”, and Russell’s use
of contextual definitions in his theory of descriptions. It is evident from other passages
that Quine wants to place some regions of his own work—such as his “virtual theory of
classes”5—along this line of continuity.
In this paper I’ll try to answer to two questions. First: Did Bentham recognize and
employ contextual definitions in basically the same way in which, for example, Russell
and Quine recognized and employed contextual definitions? And second: Did Bentham’s
endorse (a sufficiently close version of) Frege’s “context principle”? I will argue that we
should give a clearly positive answer to the first question, but a negative answer to the
second one. Bentham does acknowledge the role of contextually defined expressions; but
he embraces a much stronger (and much more questionable) view about the “semantic
3
Quine, “Epistemology Naturalized”, in Ontological Relativity and Other Essays, p. 72.
Quine, “Five Milestones of Empiricism”, in Theories and Things, pp. 69-70. See also Quine, “Two
Dogmas of Empiricism”, in From a Logical Point of View, p. 39.
5
See Quine, Philosophy of Logic, pp. 68-72; Set Theory and its Logic, especially pp. 15-27.
4
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
3
primacy of sentences” than the view that is encapsulated in Frege’s dictum. For Frege, I
will maintain, the meanings of words are conceptually dependent on the meanings of the
sentences of which they are parts; but words, at the same time, are genuine semantic units
that articulate the contents of sentences. Bentham, on the other hand, by taking
contextually defined terms as a model for all sub-propositional expressions, ends up with
a view according to which complete sentences are devoid of any semantic articulation:
words have no meaning of their own and give no semantic contribution to the expression
of the contents of the sentences of which they are grammatical parts.
The theoretical ambition of the paper is to show, contra Quine, that the
recognition of contextual definitions and the endorsement of Frege’s context principle are
two separate and independent issues. By conflating the two issues with one another—that
is, by taking contextually defined expressions as a paradigmatic exemplification of the
context principle—we loose sight of the specificity of Frege’s view about the relationship
between the meanings of propositions and meanings of their sub-propositional
components.
II
I will start with a presentation of the aspects of Bentham’s philosophy that
support the claim that he recognized and employed contextual definitions.
Throughout his long career, and in connection not only with his main intellectual
interests—that is jurisprudence and the philosophy of law—but also with several other
areas of philosophy, Bentham engaged in a sustained discussion of a problematic class of
words that he called “names of fictitious entities”. These words present themselves,
grammatically, as nouns, just as the members of the contrastive class of words, which
Bentham calls “names of real entities”. In virtue of their grammatical form names of
fictitious entities seem to have the linguistic function of naming something; but the
impression, according to Bentham, is deceptive.6 A fictitious entity, as he puts it, “is an
6
The following passage brings out nicely how Bentham’s notion of names of fictitious entities is linked to
the misleading nature of superficial grammar: “Wherever there is a word, there is thing: so says the
common notion […]. Wherever there is a word, there is a thing: hence the almost universal practice of
confounding fictitious entities with real ones—corresponding names of fictitious entities with real ones.
[…] Identity of nomenclature is certificate of identity of nature: diversity of diversity:—how absurd, how
inconsistent to make the certificate a false one!” (Memorandum Book, Bowring XI, p. 73)
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
4
entity to which, though by the grammatical form of the discourse employed in speaking
of it, existence be ascribed, yet in truth and reality existence is not meant to be ascribed.”7
When we employ names of fictitious entities, in spite of misleading grammatical
suggestions, we do not really intend to name objects; on the other hand, our
understanding of what it is that we really want to do with these words is cloudy, and
needs clarification. According to Bentham, the only method of clarification that can be
helpful in this connection is what he calls “paraphrasis”—a method that he regards as one
of his main inventions.8 Bentham gives, in different places, various descriptions of
paraphrasis, which are not always identical to one another.9 The description contained in
the following passage, however, can be regarded as representative:
The paraphrasis consists in taking the word that needs to be expounded—viz. the
name of a fictitious entity—and, after making it up into a phrase, applying to it
another phrase, which, being of the same import, shall have for its principal and
characteristic word the name of the corresponding real entity. In a definition, a
phrase is employed for the exposition of a single word: in a paraphrasis, a phrase
is employed for the exposition of an entire phrase, of which the word, proposed to
be expounded, is made to constitute the principal or characteristic word.10
In order to clarify the meaning of a name of a fictitious entity, we need, first of all, to
form an entire proposition of which it is the “principal and characteristic word”—an
operation that Bentham calls phraseoplerosis (“filling-up-into-a-sentence”);11 then we
need to translate this sentence into another sentence, with exactly the same “import” or
content, which has for its “principal or characteristic word” a name of a real entity
(paraphrase, Bentham remarks, literally means “giving phrase for phrase”12). Here is the
example that Bentham himself chooses for illustrating his doctrine of paraphrasis—the
paraphrasis of the term “obligation”:
7
Fragment on Ontology, Bowring, VIII, p. 197.
Radical Reform Bill, Appendix, Bowring III, p 594.
9
In addition to the passage quoted below, see Essay on Logic, Bowring VIII, p. 246; Radical Reform Bill,
Appendix, Bowring III p. 594; Fragment on Government, p. 495, footnote.
10
Chrestomathia, Bowring VIII, p. 126.
11
See for example Essay on Logic, Bowring VIII, p. 247.
12
Radical Reform Bill, Appendix, Bowring III p. 594
8
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
5
An obligation (viz. the obligation of conducting himself in a certain manner), is
incumbent on a man (i.e. is spoken of as incumbent on a man), in so far as, in the
event of failing to conducting himself in that manner, pain, or loss of pleasure, is
considered as about to be experienced by him.13
This means that when I say, for example, the I have an obligation to give your money
back to you, what I really mean is that I expect to suffer pain if I don’t do it. In the new
and (according to Bentham) equivalent sentence, the problematic word “obligation” has
disappeared; moreover, there is no mention of cognate words that Bentham’s regards as
equally problematic such as “right”, “duty” or “entitlement”. The talk about obligations
has been unmasked as a talk about people and their sensations of pleasures and pain,
which Bentham regards as real entities. The first kind of talk is, in Bentham’s words, a
mere “representative” or “succedaneum” of the corresponding talk involving reference to
real entities: if we make true and significant statements about our obligations, rights,
titles, etc., that is so only in so far as we are making statements about the pain that we
expect to suffer in case we perform, or fail to perform, certain actions.14 As Bentham
nicely puts it, the talk that involves names of legal fictitious entities such as “obligations”
stands to its paraphrased equivalent as paper currency stands to its gold equivalent:
These fantastic denominations [i.e., names of legal fictitious entities] are a sort of
paper currency: if we know how at any time to change them and get sterling in the
room, it is well; if not, we are deceived, and instead of being masters of so much
real knowledge as by the help of them we mean to supply ourselves with, we
possess nothing but sophistry and nonsense.15
Bentham’s conception of paraphrasis and of names of fictitious entities really
invites a comparison with some more recent employments of the technique of contextual
definitions. I want to consider, in particular, Quine’s “virtual theory of classes” and
Russell’s “theory of descriptions”.
Quine’s virtual theory of classes purports to give a reduction of part of set theory
to first order logic. The reduction is accomplished by providing (what Quine calls) a
“contextual definition” of the distinctive expressions of set theory, namely the term
13
Essay on Logic, Bowring VIII, p. 247.
Essay on Logic, Bowring VIII, p. 246.
15
Of Laws in General, p. 251.
14
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
6
“class” and the predicate “is a member of”. Such a contextual definition consists in taking
the whole sentence a ∈ {x: Fx} (that is “a is a member of the class of Fs”) and translating
it into the equivalent sentence Fa (that is “a is F”), which contains only expressions that
belong to the language of first order logic. The first sentence, according to the virtual
theory, is just a potentially misleading notational variant of the second sentence; it just
says, “in disguise”,16 what the second sentence says. The class talk that figures in the first
sentence merely “simulate”17 the language of first order logic. Even though our
grammatical form of expression suggests that we are invoking a particular class of
objects, namely “classes”, and a particular predicate of membership into classes, this is a
“sham invocation”18, a mere “manner of speaking to be paraphrased away at will”.19
Turning now to Bentham, we see that that these same descriptions apply quite well to his
theory of names of fictitious entities. The language of obligations, for example, is for
Bentham a mere simulation of a language that does not contain legal terms and mentions
only people, their actions and their sensations of pain; when we talk of obligations we
just say, in disguise, what we could very well say by means of a more parsimonious
language that lacks this legal vocabulary completely; and even though, when we talk
about obligations, we seem to invoke a class of “legal objects”, that is just a sham
invocation, a manner of speaking to be paraphrased away at will.
Let’s now consider Russell’s theory of definite descriptions. This theory analyzes
sentences containing definite descriptions into sentences in which these expressions have
been made to disappear. To take Russell’s famous example, “The present King of France
is bald” should be analyzed as “There is one and only one x such that x is the King of
France and x is bald”. Expressions like “The present King of France” are, according to
Russell, “denoting phrases”, and the theory that he proposes “gives a reduction of all
propositions in which denoting phrases occur to forms in which no such phrases occur.”20
Denoting phrases present themselves, grammatically, as expressions that have the logical
function of naming objects—an impression that is misleading and that the theory of
descriptions can help to dispel by representing in a more perspicuous way what we are
16
Quine, Philosophy of Logic, p. 72.
Ibid., p. 71.
18
Ibid., p. 69.
19
Ibid., p. 73.
20
Russell, Logic and Knowledge, p. 45.
17
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
7
actually saying when we employ these expressions in discourse. As David Kaplan has
put it, Russell’s “contextual definitions”21 of denoting phrases may be treated “as rules
for translating ordinary, logically imperfect language into a logically perfect
symbolism.”22 In a similar fashion, we may regard Bentham’s paraphrastic exposition of
the term “obligation” as a translation rule for moving from sentences of ordinary
language to sentences (not so much of a “logically perfect symbolism”, but rather) of a
restricted version of ordinary language that has been purified of all names of legal
fictitious entities—a restricted version of ordinary language that is supposed to present in
a clearer and more perspicuous way what we want to say when we normally come up
with sentences that involve those problematic expressions.
Even after a first survey, therefore, there seem to be strong similarities between
Bentham’s paraphrasis and contextual definitions as used by authors such as Russell and
Quine. It seems that we are dealing with the same technical device: a rule of translation
for rewriting sentences in a more perspicuous form, getting rid of problematic
expressions that present a misleading superficial grammar. In order to be in a more
proper position to access the similarities, however, we need to enter further into
Bentham’s philosophy and examine the use that he makes of the technique of paraphrasis.
What does he thinks he can achieve by paraphrasing away certain sorts of expressions?
And what does he actually achieve by paraphrasing them away? I will address these
questions in two steps. In the next section, I will consider exclusively what I will call the
central application of Bentham’s theory of fictions, i.e. the case of names of fictitious
entities that can be paraphrased away. More specifically, I will focus on the most
representative sub-species of this kind of names of fictitious entities, that is the names of
legal fictitious entities.23 Then, in section IV, I will focus on what I will call the extended
application of Bentham’s theory of fictions, i.e. the case of names of logical fictitious
entities—which, according to Bentham, are intrinsic to articulate language as such and
cannot be paraphrased away. By discussing these two different sorts of cases I will
21
This is the term that Kaplan himself uses for describing Russell’s theory of descriptions (“What is
Russell’s Theory of Descriptions?”, p. 233).
22
Ibid., p. 234.
23
Another class of paraphrasable names of fictitious entities is, for Bentham, the class of names of modal
properties (see Fragment on Ontology, p. 221).
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
8
present two different and opposite interpretations of the motivations that drive Bentham’s
use of paraphrasis and his appeal to notion of names of fictitious entities.
III
Throughout his career, Bentham presents his paraphrastic expositions of legal
terms as instruments of intellectual and moral liberation. They should function as tools
that help us to detect and overcome intellectual confusions—confusions that, in the cases
that Bentham is most interested in, serve “sinister interests” and are exploited, with a
greater or lesser degree of self-consciousness, in support of the perpetration of moral
mischief and political abuse. Paraphrasis is meant to serve as a medicine against
“sophistry and nonsense”. In order to see how this is supposed to work, we will consider
Bentham’s criticism of natural rights.
Bentham contrasts “natural rights” with what he calls “political rights”. The latter
kind of rights, he thinks, constitutes the primary and proper sense in which rights are
spoken of. According to his analysis, for each political right there is a corresponding
obligation, which is in turn paraphrased in terms of expectations of pain. I have, for
example, a political right to health care if some people belonging to the relevant
institutions have the obligation to provide me with health care assistance; and these
people, in turn, have such an obligation only if they can expect to suffer some form of
pain, inflicted by the designated institutions, in case they don’t provide me with the due
assistance. According to Bentham, when we assert the existence of a right, in this
political and primary sense, we assert a matter of fact—the fact that the people who have
the correspondent obligation are liable to be punished, and can reliably expect to be
punished, if they don’t comply with it. But something different happens when we start
talking about natural rights:
If I say a man has a right to this coat or the land, meaning a right in the political
sense of the word—what I assert is a matter of fact […]. If I say a man has a
natural right to the coat or the land—all that I can mean, if it mean any thing and
mean it true, is, that I am of the opinion he ought to have a political right to it;
that, by the appropriate services rendered upon occasion to him by the appropriate
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
9
functionaries of government, he ought to be protected and secured in the use of it
[…].24
It might seem that when we move from political rights to natural rights, we simply start
talking about another species of rights. But the difference is for Bentham much more
radical than what we are initially inclined to realize. When we talk abut natural rights, as
Bentham puts it in a discussion of a similar case, we start using the word “right” in a
“figurative and improper” sense, and we should not expect that “the same conclusions
[are] to be drawn from any proposition in which it is used in this sense, as might be
drawn from them if it were used in the other sense, which is its proper one.”25 Bentham’s
paraphrasis can help us to realize how radical the projection of the word “right” here at
issue is, and to warn us against the tendency to draw from its projected application the
same conclusions that we draw from its primary use. It is clear, indeed, that the projected
application of the word is not paraphrasable in the same sort of way in which we can
paraphrase its primary use. Who is going to be punished in case our “natural rights” are
not respected? And by whom? By Nature? For Bentham it is clear that, when we start
speaking about natural rights, we don’t want anymore to assert a matter of fact, but to
make a normative statement, to the effect that someone ought to be granted a political
right. The projected use of the word “right” requires a completely different paraphrasis.
The technique of paraphrasis can help us to recognize this difference in use, against a
tendency to confuse the two uses and to exploit the confusion for the attainment of
questionable ends. Here is a passage in which Bentham diagnoses the motivation that
typically leads people to talk about natural rights:
To engage others to join with him in applying force for the purpose of putting
things into a state in which he would actually be in possession of the right, of
which he thus pretends to be in possession, is at bottom the real object and
purpose of the confusion thus endeavoured to be introduced into men’s ideas, by
employing a word in a sense different from what it had been wont to be
employed, and from thus causing men to accede in words to positions from which
they dissent in judgment.26
24
Pannomial Fragments, Bowring III, p. 218.
Fragment on Government, p. 496-498. The remark refers to the difference between the duties of the
governed and the so-call “duties” of a supreme governor.
26
Pannomial Fragments, Bowring III, pp. 218-219.
25
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
10
By asserting the existence of a natural right, we want to make a normative statement; we
want to convince our interlocutor that things ought to be in a certain way. But by masking
this assertion as a statement of a matter of fact, we manage to subtract ourselves from the
genuine requirements of a normative discussion. Our interlocutor, confused by the form
of our assertion, will not properly challenge our statement by asking and evaluating the
moral or political grounds for our claim. At the same time, he won’t simply take our
assertion as the statement of a matter of fact, as when we talk about political rights. If he
did so, he could argue very easily that our assertion is false, by pointing out that there is
in fact no power (political or otherwise) that is enforcing the right that we are talking
about. He definitely understands that this objection would be out of order; that’s not what
we are talking about. On the other, what it is that we are actually talking about is unclear
to him; and it is in virtue of this very confusion that we manage to lead him to assent,
unwillingly and unwittingly, to our normative positions. Bentham writes that natural
rights have been given birth by the “weakness of the understanding” and their adoption in
discourse has been fostered by “the heat of argument” and the “force of passions”: people
appeal to the natural rights talk when they want to persuade us to assent to positions that
are ungrounded, or whose grounds they are unable to effectively articulate.27 Paraphrasis,
according to Bentham, can help us to dispel these confusions and deceptions, thus serving
as an instrument of clarification.
This kind of clarificatory employment of the technique of paraphrasis is not an
isolated moment in Bentham’s work and is not confined to his criticism of natural
rights.28 It seems rather to constitute the function that Bentham’s officially assigns to
paraphrasis. We can doubt, of course, that paraphrasis, in the case of natural rights or in
other particular cases, actually fulfils the function that Bentham has assigned to it. Is it
true, for example, that the discourse about natural rights is confused and deceitful? And if
it is, does Bentham’s employment of paraphrasis actually help for dispelling these
27
Ibid., pp. 219-220.
We can recognize a similar use of paraphrasis in other central regions of Bentham’s thought, such as his
criticism of Common Law and, more specifically, his criticism of the peculiar juridical procedures that
were called by Common Law judges “legal fictions” (e.g., the procedure of “ejectment”). In such cases,
Bentham employs paraphrasis in order to unmask the surreptitious and illegitimate arrogation of legislative
powers by Common Law judges. (See for example Pannomial Fragments Bowring, III, pp. 223-224;
Constitutional Code, Bowring IX, pp. 77-78; and for a helpful discussion, R. Harrison, Bentham, pp. 2446.)
28
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
11
confusions and deceptions? We can have, however, also a more radical doubt. We can
look with suspicion at the very motivations that guide Bentham’s employment of
paraphrasis, and argue that, in spite of his good intentions, Bentham in fact appeals to this
technique in order to subtract from a critical examination some of his most
controversial—and ultimately unsatisfying—philosophical assumptions. Paraphrasis,
according to this alternative interpretation, far from being an instrument in the service of
clarity, is a device that systematically prevents the achievement of genuine clarity and
stable philosophical satisfaction. The reasons for this suspicious way of looking at
Bentham’s use of paraphrasis will become clear in the next section, where I try to
uncover the philosophical assumptions that inform Bentham’s use of paraphrasis and
discuss the way in which these assumptions lead him to the extended application of the
doctrine of names of fictitious entities.
IV
One of Bentham’s fundamental philosophical assumptions is that words genuinely
have meaning by naming some kind of entity: “The only parts of speech which is
perfectly simple in its import, and at the same time integrally significant, is the nounsubstantive. […] [A] noun-substantive is a name […]. The entity of which it is a name,
belongs either to the class of real entities, or to the class of fictitious entities.”29 We have
already seen that, for Bentham, names of fictitious entities are just sham names; so the
only things that can genuinely be named are “real entities”. Moreover, Bentham has a
specific view about what real entities are—and therefore, about what can constitute the
genuine meaning of a word. The paradigmatic case of a real entity is for Bentham an
object that we are immediately aware of through the testimony of the senses; more
specifically, an object of the external world that we can perceive and point to (such as
persons and tables), and “objects” that we perceived through our inner sense, that is
sensations of pleasure and pain. When Bentham is actually engaged in applying the
technique of paraphrasis to particular cases, he seems quite satisfied with this basic
account of the nature of real entities; but when he approaches the issue from a more
theoretical perspective, he elaborates a partially different view. So, on the one hand,
29
Chrestomathia, Bowring VIII, p. 188-189.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
12
Bentham is inclined to enlarge the class of real entities so as to admit the possibility of
inferential real entities, such as the “Almighty Being” or “the human soul, conceived in a
state of separation from the body”.30 These are not entities that we can be immediately
aware of through the testimony of the senses, but we can infer their existence on the basis
of what we are immediately aware of—that is, on the basis of our acquaintance with
perceptible real entities.31 On the other hand, Bentham shares with other classical
empiricists the impulse to retrieve from the external word and to relegate ordinary objects
such as people and chairs to the class of inferential real entities. The only genuine real
entities would be first-personal mental items, i.e. feelings of pleasure and pain and ideas
of external objects.32 However, Bentham is reluctant to pursue with consistency this latter
line of thought, and tries to reduce the distance between our ideas and the objects of the
external world by observing the strong, necessary and irresistible character of the
inference here at issue, and by offering a sort of “pragmatic proof” of the existence of the
external world—a proof that, in all its crudity, amounts to the following: Suppose that the
objects that correspond to your sensible ideas do not exist, act accordingly, and “the
perception of pain will at once bear witness against you”.33
These assumptions about the nature of meaning and the nature of real entities
determine which words are genuinely meaningful; and some words are inevitably going
to appear problematic. Take for instance the word “obligation”. On the one hand, the
word seems to be meaningful, since we employ it in apparently meaningful discourse; on
the other hand, the word does not satisfy Bentham’s criteria of meaningfulness, since it
doesn’t name any “real entity”. The technique of paraphrasis, in this and other similar
cases, can work as a good accommodating device: it allows us to get rid of the word
“obligation”—which will be regarded thereafter as a “name of a fictitious entity”—and to
introduce in its place words that seem to satisfy Bentham's criteria of meaningfulness.
At a closer examination, however, we will find out that not only a restricted class
of words such as legal terms, but a lot other words fail to comply with Bentham’s
assumptions. We shouldn’t be surprised, therefore, to find Bentham trying to deal with an
30
Chrestomathia, Bowring VIII, p. 126.
Fragment on Ontology, Bowring VIII, pp. 195, 196; Chrestomathia, Bowring VIII, p. 126.
32
Chrestomathia, Bowring VIII, p. 126; Fragment on Ontology, p. 196.
33
Fragment on Ontology, Bowring VIII, p. 197. See also Chrestomathia, Bowring, VIII, p. 188.
31
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
13
explosive proliferation of “names of fictitious entities”. We shall focus, in particular, on
Bentham’s treatment of logical fictions—a kind of fiction that he regards as an intrinsic
feature of human language.
Bentham comes to regard as fictitious “all those entities that will be found
included in Aristotle’s […] Ten Predicaments, the first excepted”, that is “2) Quantity, 3)
Quality, 4) Relation, 5) Places, 6) Time, 7) Situation, 8) Possession, 9) Action, 10)
Passion or Suffering” (the first Predicament, which Bentham excludes from the list of
fictitious entities, is “Substance”).34 More generally, Bentham is led to think that any time
we utter a meaningful proposition, even the simplest one, we are already involved in
naming fictitious entities:
Among names of fictitious entities, the foremost, and those the designation of
which is of the most immediate necessity to mind-expressing converse, are
qualities.
[Fn: Quality being taken in the largest sense of which the word is susceptible, in
that which, in its import, is co-extensive with the applicability of the word so
much used in the Aristotelian Logic school, predication.]
Taking the word proposition in its simplest acceptation, by every proposition the
existence of some quality in some subject is asserted. A proposition is any portion
of discourse by which the existence of some quality in some subject is asserted.
The name of the substance is the noun-substantive. The name of the quality is the
noun-adjective. The word by which the relation between the quality and the
substance is asserted, viz. the existence of the one in the other, is by logicians
called the copula.35
All qualities “in the largest sense of the term” are names of fictitious entities. Any time
we predicate a quality of a substance—that is, every time we say something of
something—we are already slipping into fiction. Here is Bentham’s discussion of an
illustrative example, which I report at some length:
That apple is ripe. Apples are sweet. Apples are good. An apple is a real entity; in
saying that apple exists,—the existence of which, I express my opinion, is a real
entity. But that apple is ripe; of what is it that, in addition to that of the apple, I
express my opinion of the existence? It is the existence of the quality of ripeness
in the apple.
34
35
Fragment on Ontology, Bowring VIII, p. 199. See also Essay on Logic, Bowring VIII, p. 234-235.
Chrestomathia, Bowring VIII, p. 189.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
14
But the quality of ripeness, is it a real entity? Different from apples, and
everything else that is susceptible of it, has this quality, or any quality, any
separate existence?
[…] In saying this apple is ripe, what is it that I affirm? It is, that in this apple is
the quality of ripeness. The two expressions are equivalent. But,—in this apple is
the quality of ripeness, in the assertion thus made, what is the image that I bring to
view? It is, that the apple is a receptacle; and that, in this receptacle, the quality of
ripeness, the imaginary, the fictitious entity called a quality is lodged.
[…] Thus it is that, in the use made of language, fiction, at the very first step that
can be taken in the field of language, fiction, in the simplest, or almost the
simplest case, in which language can be employed, becomes a necessary
resource.36
Bentham just cannot conceive of words having a genuine meaning otherwise than by
naming some kind of “real entity”. So, when he considers a very simple propositions such
as “That apple is ripe”, he cannot but be puzzled by the words in the sentence that do no
even appear to name something. So he assumes—with a quick and supposedly innocent
move—that the proposition “That apple is ripe” is equivalent to “Ripeness is in the
apple”. Bentham regards the second sentence as a way of explaining the import expressed
by the first sentence, a more perspicuous rendering of its content.37 Indeed, he explicitly
argues that the simplest proposition contains, in spite of grammatical appearances, at least
four elements: a “noun-substantive” that names a quality, the copula, the preposition “in”,
and a name of a real entity, “in which” the existence of the quality is asserted.38 But once
we have so rephrased our predicative proposition, we may ask what it is that the nounsubstantive that is now the grammatical subject of the proposition is a name of. What
does the word “ripeness” name? Can we point to ripeness, as we can point to that apple?
36
Essay on Language, Bowring VIII, p. 330-331.
Bentham makes this point explicit just a few lines after the passage that I have quoted: “[…] to the
explanation of the import of the word ripe, the word ripeness may thus be rendered subservient […].”
On this point I disagree with the otherwise extremely helpful reconstruction offered by Ross Harrison. He
argues that, for Bentham, qualities are problematic only to the extent that they are referred to my means of
abstract substantive ( “ripeness”, “redness” etc.). Bentham, according to Harrison, uses paraphrasis to move
from this kind of sentences (“Ripeness is in that apple”) to ones that contain correspondent adjectives
(“That apple is red”), which Bentham would regard as unproblematic (see Bentham, pp. 86-87). But the
passages I am discussing suggest that Bentham uses paraphrasis in exactly the opposite direction: he wants
to take sentences containing adjectives and “explain their import” by rewriting them as sentences
containing abstract substantives.
38
Ibid., p. 337: “[S]ugar is sweet. The number of words employed here no more than three; but, in the form
of expression, an abbreviation may be observed. Sweetness (the quality of sweetness) is in sugar. […] For
the formation of a proposition […] no fewer than four objects require to be brought to view.” (See also
ibid., p. 333.)
37
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
15
According to Bentham, there is no real entity of which “ripeness” can be a name; the
word is a name of a fictitious entity, and the same holds for all words that appear to name
qualities—words that, according to Bentham, are necessarily involved in any possible
proposition. There is therefore a kind of fiction—which he calls logical fiction—that is a
constitutive feature of our language: “In the mind of all, fiction, in the logical sense, has
been the coin of necessity”;39 and again: “[F]iction […] is a contrivance but for which
language, or, at any rate, language, in any form superior to that of the language of the
brute creation, could not have existed.”40
The contrast that Bentham alludes to in this last quotation between human
language and the language of brutes is not just a rhetorical flourishing, but a theme that
he discusses explicitly and that is worthwhile to have in view. Bentham thinks that
animals definitely have a language, but different from ours in being completely
inarticulate. Like us, animals can express propositions; but, unlike us, they are incapable
of expressing propositions that are articulated in sub-propositional elements: “Brutes
have no terms—their language is all propositions; their faculties enable them not to break
them down into words.”41 Moreover, according to Bentham, a language like that of brutes
is at the origin of our own articulate language:
Of language in its origin, the parts could not have existed in a degree of
simplicity, equal to that of the most simple of those at present in use. The first
words must, in their import, have been equivalent to whole sentences, to
sentences expressive, for example, of suffering, of enjoyment, of desire, of
aversion.
Of this original language, the parts of speech called interjections are examples.
Of this nature is, and seems destined for ever to continue, the language of
quadrupeds and other inferior animals.
To form the words of which language is at present composed has been the work
of analysis. The original sentences were, as it were, broken down into words,
those words into syllables, and these syllables, with the help of written and visible
signs, into letters.
Of these elements, thus formed by analysis, those called words will now be put
together in the way of sentences.42
39
Fragment on Ontology, Bowring VIII, p. 199.
Ibid., p. 198.
41
Essay on Language, Bowring VIII, p. 322.
42
Ibid., p. 322-323.
40
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
16
At the beginning, human beings used to talk, like brutes, by means of entire unstructured
propositions. Then, when their faculties evolved, our ancestors proceeded, “by
abstraction and analysis”,43 to break down propositions into parts, giving us the form of
language that we presently master. According to Bentham, our use of “interjections” (that
is single-word sentences, such as certain kinds of calls and commands) is a trace of that
original language, as well as an element of commonality with other animals.44
Bentham has an ambiguous attitude towards the inarticulate language that he
places at the origins of human society. On the one hand, the fact that it is the sort of
language that “inferior animals”, today, still speaks, suggests that the articulation
characteristic of human language has constituted some form of progress or gain. On the
other hand, the process of “abstraction ad analysis” that gave us articulate language is
also the process that made us fall into fiction. By breaking down the proposition into subpropositional components, according to Bentham, we inevitably introduce a class of
words—“names of qualities”—that, in spite of grammatical appearances, do not name
any real entity and are therefore devoid of genuinely meaning. So there is a sense in
which the inarticulate language originally employed by mankind—and still employed by
non-human animals—is an ideal language: the only form of language that is not
grammatically misleading and that avoids any form of fiction. In an almost biblical way,
it seems that fiction was the price we had to pay for acquiring specifically human
capacities.
We need now to notice how far we have gone from the original application of the
doctrine of names of fictitious entities. In the central application of the theory,
paraphrasis serves for eliminating names of fictitious entities. In the case of logical
fictions, on the other hand, Bentham uses paraphrasis (even though in an implicit and
perhaps not completely conscious way) for introducing names of fictitious entities. The
proposition “That apple is ripe”, he argues, should be rewritten as “Ripeness is in the
apple”; we need therefore to pass from a proposition containing the adjective “ripe”,
which does not even appear to name something, to a propositions containing the abstract
noun “ripeness”, which now does seem to name something, namely a “quality”.
43
44
Ibid., p. 322.
See, in addition to the passage quoted above, Chrestomathia, Bowring VIII, p. 188.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
17
Moreover, the names of fictitious entities that we reach in this way are, for Bentham, just
bedrock: there is no way of getting rid of names of qualities by means of a further
paraphrasis—they should simply be accepted as an intrinsic feature of articulate
language. This is another significant difference between the central and the extended
application of Bentham’s doctrine. In the central application, praphrasability is a
constitutive property of names of fictitious entities; but in the extended application this
internal connection is severed through the introduction of the notion of names of
fictitious entities that are impossible to eliminate. Similarly, in the central application of
the theory, the notion of a misleading form of expression is always correlated to the
notion of a clear form of expression: at least in principle, any misleading form of
expression can be converted, through paraphrasis, into a clear form of expression. But
when we come to names of logical fictions (which, like other names of fictitious entities,
are regarded as misleading), it turns out that no clear form of expression is within our
reach: no reworking or improvement of our present language could be sufficient for
avoiding this kind of fictions. For Bentham, it would seem, the only way of freeing
ourselves from logical fictions would be to jump out of articulate language as such—to
jump out, that is, of the only kind of language that we can understand and make sense
of—and go back to the language of brutes.
Bentham’s treatment of qualities and logical fictions brings his philosophical
assumptions about meaning to their extreme consequences, and can indeed be regarded as
a reductio of those assumptions. We may think that instead of accepting the idea that we
are under the inescapable domination of fiction we should simply give up the hopeless
assumption that the only way in which a word can have a meaning is by naming a “real
entity”.
If we adopt this perspective, then we will tend to see Bentham’s theory of fictions,
and the correlative technique of paraphrasis, as an obstacle, rather than an aid, to the
attainment of clarity. We will tend to think that the theory seduces us into a series of
accommodations that have the effect of postponing the moment at which we eventually
recognize the need of putting in question Bentham’s unexamined and ultimately
unsatisfying philosophical assumptions. In the case of words such as “obligation”, as we
saw above, the accommodation works smoothly and elegantly. The situation is more
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
18
complicated when we come to realize that, no matter how many times we reiterate the
operation of paraphrasis, we will always obtain sentences that contain some recalcitrant
terms; given Bentham’s assumptions about meaning, the very phenomemon of
predication becomes problematic. The theory of fictitious entities cannot accommodate
these problems as it did with words like “obligations”; it introduces, therefore, the
acrobatic notion of names of fictitious entities that are impossible to eliminate. Arguably,
this is not an accommodation at all, but rather the point of explosion of Bentham’s
semantic assumptions. But the theory of fictitious entities can still manage to create the
impression of an accommodation here by trading on our failure to realize how radically
different the extended application of the theory is from its central application. (In virtue
of the mere fact that we are still talking about “names of fictitious entities”, we can be
deceived into thinking that we are being offered the same kind of smooth and elegant
accommodation that we were offered in the case of words like “obligation”.) In addition,
Bentham’s call for resignation can exercise some influence and fascination on our minds
(“Isn’t logical fiction a reasonable price to pay for enjoying the benefits of articulate
language?”). However, if we come to recognize the need of rejecting Bentham’s
fundamental assumptions about meaning, we can see not only that the “accommodations”
provided by the extended application of the theory of fiction are only apparent, but also
that the accommodations provided by its central application, even though unexceptional
from technical point of view, have in fact the effect of hindering philosophical progress
by hiding the problematic character of those assumptions.
This is a very different way of looking at Bentham’s theory of fictions than the
one considered earlier. In the previous section I tried to formulate and show the grounds
for a sympathetic interpretation of the motivations of Bentham’s theory of fictions:
paraphrasis is supposed to work as a clarificatory device that harness and sharpen our
reflective capacities, by making it easier to recognize some confusions that originate from
the misleading character of some expressions of ordinary language. Such an
interpretation corresponds to Bentham explicit intentions and represents the designated
function of the theory of fictions. In the present section, on the other hand, I have tried to
articulate an unsympathetic interpretation of the motivations of Bentham’s theory of
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
19
fictions: the theory, in spite of Bentham’s best intentions, works in fact as an
accommodating device in the service of specific philosophical preconceptions.
The two interpretations, even though quite opposite to one another, are not
mutually incompatible. We may think (and I do think) that there are very good grounds
for the unsympathetic interpretation; and yet this, by itself, does not rule out the
possibility that Bentham succeeds, in some particular cases, to put paraphrasis to a
genuinely clarificatory use. Let’s recall, for example, Bentham’s discussion of political
and natural rights: we may think that in that case paraphrasis does help to avoid certain
confusions, even though, at the same time, it is also used by Bentham for defending his
philosophical assumptions about the meanings of words. The two issues are only
externally connected to one another: it seems possible to keep some of the clarificatory
employments of paraphrasis that Bentham points out, while at the same rejecting his
problematic assumptions and thus avoiding to use paraphrasis as an accommodating
device.
It is not my purpose, here, to try to establish whether and to what extend Bentham
actually succeeds in giving a clarificatory use of paraphrasis. (I don’t think the issue can
be determined in any general way, but only through a piecemeal examination of the
various cases in which Bentham appeals to this technique.) My goal has been to lay out
the ambiguity of the motivations that guide Bentham’s appeal to the doctrine of fictitious
entities and to the device of paraphrasis. I now wish to submit that this ambiguity is also a
feature of the use of contextual definitions in the work of some more recent authors in the
analytic tradition.45 Russell is perhaps the clearest example: he says that his theory of
descriptions, which gives a contextual definition of denoting phrases, should be tested
“by its capacity for dealing with puzzles”;46 and we can think that his theory is actually
helpful for dealing with some particular puzzles and confusions; and yet, the theory can
also be regarded as functional to some of Russell’s most problematic philosophical
assumptions. The grounds for this sort of suspicion become more substantive, of course,
when we see that for Russell even the fact that we can talk about tables and chairs
45
I don’t want to claim, however, that anytime an author makes use of contextual definitions there are good
grounds for adopting an unsympathetic interpretation of her motivations. We may think, for example, that
there is no problematic assumption behind Quine’s virtual theory of classes.
46
Russell, Logic and Knowledge, p. 47
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
20
become a “puzzle” that can be “solved” by his theory of descriptions—which is supposed
to show how we can manage to talk indirectly about objects of the external world by
talking directly about our sense data.47
I conclude therefore this part of the paper by agreeing with Quine that there are
very good reasons for attributing to Bentham the recognition of “contextual definitions”:
he used essentially the same technical device that more recent authors and Quine himself
have called by this name, and he used it for similar, and similarly ambiguous,
philosophical motivations. But does this show that Bentham also anticipated Frege’s
context principle?
V
According to Quine, as we saw earlier, the recognition of contextual definitions is
a sign of the recognition of the “semantic primacy of sentences”. Moreover, he thinks that
it is precisely this “semantic primacy” that is celebrated by Frege. Therefore, from the
fact that Bentham acknowledged and employed contextual definitions it follows that
Bentham also anticipated Frege’s view about the relationship between the meanings of
words and the meanings of sentences. The soundness of this reasoning crucially depends
on what we mean by “semantic primacy of sentences”. I want to argue that, in order to
grasp the specificity and the philosophical attractiveness of the view that is encapsulated
in Frege’s dictum (i.e. the dictum that words do not really have a meaning in isolation,
but only in the context of significant propositions), we need to distinguish various
understandings of the “semantic primacy of sentences” and to contrast Frege’s
understanding of it with Bentham’s, rather than aligning the two with one another. First
of all, however, we need to see why Bentham’s use of contextual definitions might even
47
Russell, “On Denoting”, p. 56. Russell’s philosophical assumptions about meaning are in fact very
similar to Bentham’s: he thinks that a word can only be meaningful by naming an entity with which we are
immediately acquainted. Such a similarity, however, should not bring us to think that the ambiguity of the
motivations behind the employment of contextual definitions is a malady that spread exclusively among
empiricist authors. Early Wittgenstein is a good example of non-empiricist author who runs into similar
difficulties: he conceives of “analysis” and of the employment of a perspicuous notation as tools for
overcoming philosophical confusions; and yet, as some commentators have argued, he later came to
recognize that his earlier employment of these “clarificatory tools” was informed by unwitting
philosophical preconceptions (not by empiricist preconceptions, but by preconceptions, say, about the
determinateness of logical space, or about the physiognomy of the activity of following a rule). For a
discussion of these aspects of the Tractatus and of later Wittgenstein’s criticism of it, see Diamond, “Crisscross Philosophy”, and Conant, “Mild Mono-Wittgensteinianism”.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
21
appear to present some assonance with Frege’s context principle and to express some
idea about the semantic primacy of sentences.
One source of the impression of an assonance here is the way in which Bentham
contrasts paraphrastic expositions of names of fictitious entities with other kinds of
expositions of the meanings of words, in particular ostensive definitions (what Bentham’s
calls “expositions by representation”)48 and definitions per genus et differentiam.49 When
the meaning of a word is a real entity within our field of perception, we can explain it by
pointing to the object that it names (John = That!); and when the word that needs to be
explained is a common noun with a superior genus, we can proceed by indicating the
superior genus and the distinguishing mark of the species that we want to single out (man
= rational animal). Bentham contrasts these procedures with the explanation of the
meanings of names of fictitious entities that do not have any superior genus. In such cases
we cannot indicate the meaning of the word to be explained, but we need to indicate, by
paraphrasis, the meanings of whole sentences of which the word is a part. According to
Bentham, for example, we cannot give an explanation of the meaning of the word
“obligation” of the form: “Obligation = …”; here we can only move at the level of the
meanings of complete sentences. So it can seem that whereas certain words have meaning
in isolation, other words have meaning only in the context of complete propositions,
which become the primary vehicles of meaning.
This way of characterizing contextual definition by using a strongly Fregean
vocabulary is explicit in Russell, and this might be one of the factors that led Quine to
align contextual definitions with Frege’s context principle. Consider the following
remarks by Russell:
[Denoting phrases] are not assumed to have any meaning in isolation, but a
meaning is assigned to every propositions in which they occur.50
48
Essay on Logic, Bowring VIII, p. 243: “If all words were significative of real entities, and if these were
all objects which might at all time be brought within the reach of perception both of the learner and the
teacher, exposition would be easy and consist in the pointing to the object in question, and pronouncing at
the same time the word which it is wished to attach to it as its name.”
49
Ibid., p. 245.
50
Russell, “On Denoting”, Logic and Knowledge, p. 42.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
22
[A] denoting phrase is essentially part of a sentence, and does not, like most
single words, have any significance on its own account.51
[Incomplete symbols] are things that have absolutely no meaning whatsoever in
isolation but merely acquire a meaning in context. ‘Scott’ taken as a name has a
meaning all by itself. It stands for a certain person, and there it is. But ‘the author
of Waverly’ is not a name, and does not all by itself mean anything at all […].
[Incomplete symbols] are aggregations that only have a meaning in use and do not
have any meaning in themselves.52
Incomplete symbols, of which denoting phrases are a species, are for Russell expressions
that can only be defined contextually. He claims that they these symbols don’t have any
meaning in isolation, but only in context. This way of speaking strongly recalls the
phrasing of Frege’s dictum. But I suggest that we should be cautious to infer an actual
correspondence in meaning from this correspondence in words.
A first difference that we should notice between Frege’s context principle and
contextual definitions is the scope of their application. For both Bentham and Russell, it
is only a restricted class of words that require a contextual definition—i.e., paraphrasable
names of fictitious entities and incomplete symbols respectively. These words are
contrasted with terms that are not contextually defined—names of real entities in
Bentham’s case, and names of “entities with which we have immediate acquaintance” in
Russell’s case.53 On the other hand, Frege gives no indication that his dictum is meant to
apply only to a restricted class of sub-sentential expressions.54
But couldn’t we see Frege’s dictum simply as a generalization of the view that
Bentham and Russell held about contextually defined terms? The specificity of Frege’s
view, according to this suggestion, is that he takes contextually defined expressions as a
general model for all sub-propositional expressions. But a textual problem should warn
51
Ibid., p. 51.
Russell, “The Philosophy of Logical Atomism”, in Logic and Knowledge, p. 253.
53
Russell, “On Denoting”, in Logic and Knowledge, p. 56.
54
Frege divides the meanings of sub-propositional expressions into two main logical categories, “concepts”
and “objects”; the fact that he talks of object as “saturated” or “self-subsisting”, whereas he talks he of
concepts as “unsaturated” and “incomplete”, might suggest that he thinks that object-words, unlike
concept-words, have a meaning in isolation and don’t therefore really comply with the context principle.
But such a suggestion is explicitly rejected by Frege (Foundations of Arithmetic, p. 72): “The selfsubsistence which I am claiming for number [which Frege considers to be an object] is not to be taken to
mean that a number word signifies something when removed from the context of the proposition, but only
to preclude the use of such words as predicates or attributes, which appreciably alters their meaning.”
52
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
23
us that we are not on the right track. Frege, in fact, is quite opposed to contextually
defined expressions. In the following passage, for example, he considers and rejects the
proposal to take classes as contextually defined terms, so that the class-talk would be a
mere way of speaking, a mere simulation of predicate logic, as in Quine’s virtual
theory:55
[Suppose we] regard class names as sham proper names which would thus not
have a Bedeutung. They had to be regarded as part of signs that had a Bedeutung
only as wholes. Now of course one may think it advantageous for some end to
form different signs that partly resemble one another, without thereby making
them into complex signs. The simplicity of the signs requires only that the parts
that may be distinguished within it should have no separate Bedeutung. On this
view, then, even what we usually regard as a number sign would not really be a
sign at all, but only an inseparable part of a sign. A definition of the sign ‘2’
would be impossible; instead we should have to define many signs, which would
contain ‘2’ as an inseparable part, but could not be regarded as logically
compounded of ‘2’ and another part. It would thus be illicit to replace such an
inseparable part by a letter; for as regard the content of the whole sign, there
would be no complexity.56
The problem with contextually defined terms, according to Frege, is that they are not
genuine semantic units, and that the sentences of which they are parts are not
semantically or logically complex. Take for example these two sentences: a ∈ {x: Fx}
and b ∈ {x: Fx}. Frege argues that if we see classes as contextually defined terms, we
have to regard these two sentences as logically simple: each of them, as a whole, has a
meaning (the two sentences are equivalent, respectively, to Fa and Fb), but they don’t
mean what they mean in virtue of what their syntactical parts mean. The two sentences
are, of course, graphically and syntactically complex, but not semantically complex.57
The sign {x: Fx}, for example, appears in both sentences, and presents itself,
syntactically, as a unit: it seems to name a particular object, a class. But, according to the
55
This way of interpreting the theory of classes is also discussed (and rejected) by Frege in his
correspondence with Russell (Philosophical and Mathematical Correspondence, pp. 158-161).
56
Frege, Basic Laws of Arithmetic, Vol II, Appendix, in The Frege Reader, p. 282.
57
Earlier in Basic Laws of Arithmetic, Frege clarifies the specificity of semantic complexity: “Any symbol
or word can indeed be regarded as consisting of parts; but we do not deny its simplicity unless, given the
general rules of grammar, or of the symbolism, the Bedeutung of the whole would follow from the
Bedeutungen of the parts, and these parts occur also in other combinations and are treated as independent
signs with a Bedeutung of their own” (The Frege Reader, p. 269).
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
24
view that Frege here is criticizing, this impression is misleading: the sign {x: Fx} serves
only as a notational device that allows us to translate whole sentences containing it into a
more perspicuous form; it has “no separate Bedeutung” and is not really not “a sign at all,
but only an inseparable art of a sign”.
Now we can think that Frege here is exaggerating and anyway misrepresenting
the situation to the extent that he is claiming that sentences containing contextually
defined expressions are completely devoid of semantic articulation. He claims that we
would need to define one by one each of the sentences containing a contextually defined
expression, and this seems wrong, because the aim of contextual definitions is to afford a
systematic way of getting rid of contextually defined expressions:58 we don’t need, for
example, two separate rules of translation for rewriting in a perspicuous form the
sentences a ∈ {x: Fx} and b ∈ {x: Fx}; we just need to apply the same rule twice. But
this inaccuracy on Frege’s part should not distract us from the main criticism that he is
leveling against contextually defined terms: the point is that these terms obscure (even
though not completely) the logical structure of the contents expressed by the sentences in
which they occur. The greater the number of contextually defined terms a language
contains, the more un-perspiscuous the language becomes. And it does seem accurate to
say that if we take contextually defined terms as a model for all sub-sentential
expressions, then what we obtained is a system of communication that contains signs for
entire propositions, but devoid of any semantic articulation. In such a system, the
graphical or phonetic articulation of each sentence-sign gives us no clue about its content;
each sentences-sign, here, does need to be translated one by one. For Frege, it would be a
mistake to regard such a system of signs as an accurate model of language. He argues
explicitly and repeatedly that “language in the proper sense would be impossible”59 if the
parts of the sentence did not correspond to the parts of the thought it expresses, “so that
the structure of a sentence serves as an image of the structure of the thought”.60
58
This point is explicit in Bentham and Russell, and explicitly emphasized by Quine (see Quine, Theories
and Things, p. 68).
59
Frege, “Letter to Jourdain, Jan 1914”, in The Frege Reader, p. 320
60
Frege, “Compound Thoughts”, in Logical Investigations, pp. 55-56. See also, “Notes for Ludwig
Darmstaedter”, in Posthumous Writings, p. 255; “Logic in Mathematics”, in Posthumous Writings, pp. 225,
243; “Letter to Peano, 29.9.1896”, in Philosophical and Mathematical Correspondence, p. 115.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
25
But Frege’s criticism of contextually defined expressions seems to lead us into a
philosophical and exegetical impasse. On the one hand, Frege says, in his Grundlagen,
that words have no meaning in isolation but only in the context of the propositions; and
on the other hand, in Grundgesetze and other later writings, he says that words must have
a “meaning on their own account” and that they must be “separable signs”; he even says
that their meaning must be “independent of the other parts of the proposition”.61 Some
commentators have taken the philosophical impasse to be genuine, and dealt with the
exegetical one by adopting a developmental approach: later Frege gave up the context
principle that he had previously endorsed. I wish to submit, however, that this
interpretation is more grounded on the apparent absence of an alternative than on
persuasive textual evidence.62 In order to see that there is alternative here—i.e., in order
to see that the philosophical impasse here is only apparent—we need to notice that
phrases such as “Words have meaning in isolation” and “Words have meaning only in the
context of a proposition” can mean, each of them, two different things, in accordance
with following diagram:
1) “Words have meaning in isolation”
=
The meanings they have are in no way conceptually
dependent on the meanings of the sentences of
which they are parts.
2) “Words have meaning in isolation”
=
Words have their own meaning, they are semantic
units.
3) “Words have meaning only in
propositional context”
There is a conceptual dependence between the
=
meanings of words and the meanings of the
sentences of which they are parts.
4) “Words have meaning only in
propositional context”
Words are not semantic units; they do not have their
=
own meaning; only the sentences of which they are
parts really have a meaning.
Frege can maintain, at the same time, both (2) and (3). He can maintain, against (4), that
proposition are articulated into semantic units—units that give their own semantic
61
Frege, “Letter to Peano, 29.9.1896”, Philosophical and Mathematical Correspondence, p. 115.
Indeed, the developmental approach is saddled by several passages of later Frege’s that seems to confirm
his commitment to his “early contextualism”. See for example “Notes to Ludwig Darmstaedter, 1919”, in
Posthumous Writings, p. 253.
62
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
26
contribution to the meanings of the sentences of which they are parts. Such parts are
separable signs, independent from the rest of the sentence, in the sense that each of them
can occur in other sentences. At the same time, Frege can maintain, against (1), that these
semantic units are conceptually dependent on the wholes of which they are parts: subsentential elements have meaning, are genuine semantic units, only in the context of
complete significant propositions. Take a word out of its propositions context, and it will
loose its semantic properties. The notion of “part” that Frege is invoking here is a notion
according to which a part cannot be what it is except as a part of an appropriate whole.63
The specificity of the view of propositional meaning that is encapsulated in
Frege’s context principle lies precisely in a joint endorsement of (2) and (3): words have
meaning only in the context of the proposition, but they have their own meaning. Given
this basic commitment, Frege can surely maintain that (4) holds good for some terms.
Frege does not need to deny, and does not deny, that our language comprises contextually
defined terms, so that we sometimes communicates our thoughts by means of expressions
that lack a complete semantic articulation. What he does deny, however, is that we should
take these expressions as a general model of propositional meaning.64 Moreover, he
thinks that even though contextually defined expressions are part of our ordinary
language, they should not appear in a perspicuous notation specifically designed to
facilitate the recognition of valid inferences.
The diagram given above can also help to contrast Frege’s view with Russell’s
and, finally, with Bentham’s.
Russell has only room for (4) and (1). He can see how there can be contextually
defined terms, i.e. terms that are dependent on their propositional context in the sense that
they don’t event amount to genuine semantic units. By means of contextual definitions,
we can paraphrase these terms away and obtain sentences that contain only genuine
semantic components. But for Russell these components have a meaning of their own,
63
See “Compound Thoughts”, pp. 55-56, where Frege warns us about the dangers of speaking of the
“parts” of a thought—a way of speaking that could misleadingly suggest that a thought is a mere
agglomeration of atomistically conceived components. See also “Logic”, in Posthumous Writings, p. 145,
where Frege compares the parts of a thought to the parts of a living organism.
64
See ‘Letter to Jourdain, Jan 1914”, in The Frege Reader, p. 320, where Frege considers the possibility of
adopting “the convention that certain signs were to express certain thought, like railway signals (‘The track
is clear’)”; if all our expressions, however, were similarly devoid of semantic articulation, “language in the
proper sense would be impossible”.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
27
independent of the context of the proposition, in a much stronger sense than what Frege is
willing to allow. Russell general semantic and metaphysical system is completely
atomistic.65 The meanings of genuine sub-sentential units are “entities with which we
have immediate acquaintance”; and we can be acquainted with each of these entities
independently of its mode of combination (or lack of combination) with other entities.
For Russell, words that are not contextually defined definitely have a meaning outside of
the context of the proposition; there is no conceptual dependence between the meanings
of these words and the meanings of the sentences in which they may occur.66 According
to Russell, therefore, (4) holds good of contextually defined expressions and (1) holds
good of all the expressions that are not contextually defined.
Bentham, on the other hand, ends up with a view that only leaves room for (4). At
first glance, it might seem that Bentham’s position is basically the same as Russell’s: (4)
applies to names of fictitious entities, and (1) applies to names of real entities. But the
extended application of the doctrine of fictitious entities marks a significant point of
divergence between Bentham and Russell. For Russell, it is essential to any significant
proposition that it can be rewritten in a form that contains only real propositional
components, i.e. words that name entities with which we are immediately acquainted. For
Bentham, on the other hand, as long as we remain within the confines of articulate
language, we cannot get rid of fictions. When we try to break a significant proposition
into semantic parts, we obtain, on the one hand, names of real entities, which have a
meaning of their own; but we also obtain names of “qualities”, that is expressions that, in
spite of grammatical appearances, do not genuinely have a meaning. As suggested by
Bentham’s ambiguous fascination for “interjections” and the “language of brutes”,
propositions could be expressed in a completely perspicuous way only by means of
sounds or signs that are devoid even of grammatical articulation: in that way, we would
not be led to attribute to sentences a semantic articulation that is not (and cannot be)
really there. Bentham, we might say, starts his analysis of meaning with a view about
sub-propositional meaning, according to which the only words that genuinely have a
meaning are names of real entities; but since he is unable to develop a coherent account
65
See P. Hylton, Russell, Idealism and the Emergence of Analytic Philosophy, pp. 105-275.
As is well known, this atomistic conception of sub-propositional semantic components led Russell right
into the problem of the unity of the proposition.
66
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
28
of the relationship between propositional and sub-propositional meaning (names of real
entities are never enough for expressing the content of an entire proposition), he ends up
with a view of the meanings of entire propositions as completely un-articulated wholes,
thus jettisoning—ultimately—the very idea of sub-propositional semantic components.
In conclusion, therefore, I agree with Quine about the fact that the use of
contextual definitions is linked to the recognition of a sort of “semantic primacy” of
sentences; and I think that Bentham, by claiming that every possible proposition contains
some names of fictitious entities, is led to attribute to sentences precisely this sort of
semantic primacy. But I have argued, contra Quine, that this is not the semantic primacy
that is at issue in Frege’s context principle. It is a primacy that—unlike the one
acknowledged by Frege—does not leave room for the specific semantic role of subsentential expressions.
VI
The conclusion I reached about the sort of semantic primacy that Bentham grants to
sentences is corroborated, I think, by some passages that are not immediately connected
with his discussion of names of fictitious entities. Interestingly enough, commentators
have quoted some of these passages as further evidence of a supposed line of continuity
with Frege’s context principle. Let’s consider, for example, the following passage, which
has been quoted and discussed by Peter Hacker:
But by anything less than the entire proposition, i.e. the import of an entire
proposition, no communication can have place. In language, therefore, the integer
to be looked for is the entire proposition—that which Logicians means by the
term logical proposition. Of this integer, no one part of speech, not even that
which is most significant, is anything more than a fragment; and, in this respect,
in the many-worded appellative, part of speech, the word part is instructive. By it,
an intimation to look out for the integer, of whish it is a part, may be considered
as conveyed. A word is to a proposition what a letter is to a word.67
According to Hacker, this passage expresses Bentham’s commitment to a form of the
context principle “closer to the later Wittgenstein than to Frege’s (not altogether happy)
contention that a word has a meaning only in the context of a sentence”. Such a form of
67
Chrestomathia, Bowring VIII, p. 188.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
29
the context principle, Hacker continues, “rightly stresses that the sentence is, as
Wittgenstein was later to maintain, the minimal move in a language game”.68 I suspect
that Hacker is here attributing to Bentham a weaker notion of semantic primacy of
sentences than the one that is acknowledged by Frege: a notion that allows words to have
meaning outside of propositional contexts. We can indeed maintain that sentences are the
“minimal move in the language game”, in the sense that only my means of complete
sentences we actually say something, and still leave completely open the possibility that
we can mean something by means of sub-sentential elements even though they do not
occur as parts of meaningful propositions. If this is the view that Bentham is endorsing,
then it should be contrasted rather than aligned with the view endorsed by Frege, since
Frege’s dictum, after all, says precisely that words do not have really a meaning outside
of significant propositional contexts. My sense, however, is that the passage from
Bentham quoted above expresses a view that differs from Frege’s in exactly the opposite
way: Bentham is endorsing a stronger notion of the semantic primacy of sentences than
Frege’s—a notion that rules out of the picture the very idea of sub-sentential semantic
units. I think that Frege could agree with each of the sentences of the passage, except the
last one. He could agree, in particular, that sub-sentential semantic units are essentially
parts of complete significant propositions. But the last sentence suggests that the
vocabulary of “part”, “fragment” and “integer” is here used in a different register than in
Frege’s works. For Frege, what is typically true of a word is exactly that it is not “to a
proposition what a letter is to a word”. A word, typically, does not mean what it does in
virtue of what its letters means; letters, unlike words, are not semantic units. When
Bentham and Frege talk about “parts” and “fragments” of significant propositions, they
are talking about two different kinds of “parts” and “fragments”: Frege is talking about
semantic parts, whereas Bentham is only talking about graphic or phonetic parts.
The commitment to a semantic primacy of sentences that rules out of the picture
their semantic articulation seems also to emerge from this other passage of Bentham’s,
where he contrasts his own view with an atomistic understanding of the compositionality
of language that he attributes to Aristotelian logicians:
68
Hacker, “The Rise of Twentieth Century Analytic Philosophy”, p. 67.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
30
The subjects [that Aristotelian Logic] began with were terms, i.e. words of a
certain description, and beginning with the consideration of these terms, it went
on to the consideration of propositions in the character of compounds capable of
being composed out of these elements.
[…] These terms are accordingly spoken of as possessing of themselves an
original and independent signification, as having existence before anything of the
nature of a proposition came to be in existence;—as if finding these terms
endowed, each of them, somewhat or other, with a signification of its own, at a
subsequent period some ingenious person took them to hand, and form them into
propositions.
But the truth is, that in the first place came propositions, and that out of these
propositions, by abstraction and analysis, terms possessed, each of them, of an
independent import, were framed.69
Aristotelian Logic, according to Bentham, fails to recognize any conceptual dependence
between the meanings of sub-propositional elements and the meanings of the
propositions of which they are parts. The paradoxical upshot of this atomistic
understanding of the meanings of words is that the very fact that we use words to say
something appears as a curious accident. The view, according to Bentham, is committed
to the intelligibility of a scenario in which some subjects master the lexicon of a language
without mastering the language itself: they know the meaning of each word, but they
don’t know how to use words for the expressions of complete thoughts. Now, Frege
would surely agree with Bentham’s rejection of this conception of the compositionality of
language. But he would also disagree with the positive view that Bentham proposes as an
alternative. From a Fregean perspective, Bentham over-reacts to the mistakes of the
“Aristotelians” and ends up occupying an opposite but equally unsatisfying position:
recoiling from a view that confers semantic priority to the meanings of words without
recognizing their conceptual dependence on the meanings of sentences, he ends up with a
view that confers semantic priority to the meanings of sentences without recognizing
their conceptual dependence on the meanings of their components. (If we consider the
diagram given above, Bentham embraces (4) by recoiling from (1)). For Bentham it is
just a curious accident that we use articulate sentences; our sentences express exactly the
same contents that, at the beginnings of human society, we used to express by means of
completely inarticulate expressions. Bentham says that “abstraction and analysis” gave to
69
Essay on Language, Bowring VIII, p. 322.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
31
the parts of propositions “an independent import”; but he is not really entitled to this
claim. As we have seen, abstraction and analysis are, in his view, intrinsically falsifying
operations. Such operations inevitably introduce names of fictitious entities, that is terms
that possess only a sham independent import. For Bentham, in spite of grammatical
appearances, the terms into which our propositions are articulated no more have a
meaning of their own than the letters of a word.
References
BENTHAM, J., The Works of Jeremy Bentham, ed. J. Bowring, Edinburgh: W. Tait,
1843.
—— Of Laws in General, ed. H.L.A. Hart, London: The Athlone Press, 1970.
—— A Commentary on the Commentaries and A Fragment on Government, J.H. Burns
and H.L.A. Hart eds, London: The Athlone Press, 1977.
CONANT, J., “Mild Mono-Wittgensteinianism”, in Wittgenstein and the Moral Life:
Essays in Honour of Cora Diamond”, Cambridge (Mass.): The MIT Press, 2007, pp. 31142.
DIAMOND, C., “Criss-Cross Philosophy”, in Wittgenstein at Work: Method in the
Philosophical Investigations, E. Ammereller and E. Fisher (eds.), London: Routledge,
2004, pp. 201-219.
FREGE, G., The Foundations of Arithmetic, trans. J.L. Austin, 2nd edn, Evanston:
Northwestern University Press, 1953.
—— Logical Investigations, ed. P.T. Geach, trans. P.T. Geach and R.H. Stoothoff,
Oxford: Blackwell Publishing, 1977.
—— Posthumous Writings, ed. H. Hermes, F. Kambartel, and F. Kaulbach, trans. P.
Long and R. White, Oxford: Basil Blackwell, 1979.
—— Philosophical and Mathematical Correspondence, ed. Gottfried Gabriel et al., trans.
Hans Kaal, Oxford, 1980.
—— The Frege Reader, ed. M. Beaney, Oxford: Blackwell Publishing, 1997.
HACKER, P.M.S., “The Raise of Twentieth Century Analytic Philosophy”, in The Rise
of Analytic Philosophy, ed. H-J Glock, Oxford: Blackwell, 1997, pp. 243-268.
Silver Bronzo
DRAFT: Context Principle and Contextual Definitions in Bentham
32
HARRISON, R., Bentham, London: Routledge and Kegan Paul, 1983.
HYLTON, P., Russell, Idealism and the Emergence of Analytic Philosophy, Oxford:
Clarendon Press, 1990.
KAPLAN, D., “What is Russell’s Theory of Descriptions?”, in Bertrand Russell: A
Collection of Critical Essays, ed. D.F. Pears, New York: Anchor Books, 1972.
OGDEN, C.K., Bentham’s Theory of Fictions, New York: Harcourt, Brace & Co., 1932.
QUINE, W.V., From a Logical Point of View, Cambridge (Mass.): Harvard University
Press, 1953, pp. 20-46.
—— Set Theory and Its Logic, 2nd edn, Cambridge (Mass.): Harvard University Press,
1969.
—— Ontological Relativity and Other Essays, New York: Columbia University Press,
1969.
—— Theories and Things, Cambridge (Mass.): Harvard University Press, 1981.
—— Philosophy of Logic, 2nd edn, Cambridge (Mass.): Harvard University Press, 1986.
RUSSELL, B., Logic and Knowledge, ed. R.C. Marsh, London: Routledge, 1956.
WISDOM, J., Interpretation and Analysis in Relation Bentham’s Theory of Definition,
London: Kegan Paul, Trench, Trubner & Co., 1931.