1
Article publicat en Perssinoto, L. & Ramón, B. (eds.), Wittgenstein and Plato.
Connections, Comparisions and Contrasts. Edit: Plagave Macmillan, London 2013,
pàgs: 183-195.
------------------------------------------------------------------------------------------------------
PLATO, WITTGENSTEIN AND THE ORIGINS OF LANGUAGE
Antoni Defez
Universitat de Girona
Abstract: In this paper my intention is to analyse the theory of language in Plato’s
Cratylus and compare it to the second Wittgenstein’s philosophy of language. First, I
pay attention to Plato’s conception of meaning and his solution to the problem of the
origins of language. These questions are put in relation to Plato’s ontological and
epistemological views: i.e., essentialism, the theory of reminiscence and the sophistic
thesis of the impossibility of speaking with falsehood. Afterwards, I display, as a rival
view to it, the anti-intellectualist and anti-essentialist approach of Wittgenstein, which
allows us to see language as a social creation that arises from the natural reactions of
human beings, a solution that does not generate the difficulties in Plato’s position.
The question of the origins of language seems to be a problem that sooner or
later our scientific knowledge will be able to solve. However, it might be that this is
not the case because, unlike other similar issues, perhaps the question of the origins of
language has no solution: in fact, whatever hypothesis we construct may never be
verified. Now, we do not need to worry about this, and we need not follow the
example of the Société Linguistique de Paris, which in 1866 decided to remove this
question from its meetings because it was obscurely metaphysical and insoluble given
the state of empirical research. No, there is no reason for it. First, philosophy largely
feeds on problems without solutions. And secondly, if we take this drastic step we
would miss what is really important from a philosophical standpoint.
2
It is true that this decision was influenced by the fact that during the two
previous centuries many speculative hypotheses on this subject had followed one
another –let us think, for example, of the theories of E. Condillac, J.J. Rousseau or
J.G. Herder1
. Yet, this prohibition was insensitive to the fact that, beyond the problem
of its empirical solution, what is really important in a matter like this is just the
speculative stuff, that is, the conceptual and philosophical assumptions that inevitably
come into play when we deal with it. For instance: the way we understand what kind
of things human beings are, what language and linguistic meaning are, how learning
of words happens, what the connection between linguistic behaviour and nonlinguistic
behaviour is, or what the relationship between language and thought is, and
so on.
Well then, driven by speculative interests such as these, in this paper we will
analyse what Plato and the mature Wittgenstein thought about the origins of language.
We will not try to show that Wittgenstein was right and Plato was wrong. It would be
out of place and would certainly be anachronistic and unfair. We shall simply show
that Wittgenstein’s approach, at least for some of us, is a way of looking at the
question of the origins of language that is much more fruitful than Plato’s strategy,
which however has been the most widely accepted throughout the history of Western
thought both in philosophy and in the field of linguistics.
*
1
For a treatment of the hypotheses of these authors, see Defez, A., “Llenguatge i Pensament in
Rousseau” in Comprendre. Revista catalana de filosofia, nº. V, 2003 / 2, Universitat Ramon Llull,
Barcelona, 2003, pp. 181-195. You can consult this work in www.defezweb.net
3
As is known, Plato paid attention to language in the Cratylus, a dialogue that
he wrote when he was about forty years old, and once the Academy had been
created2
. This is a work in which, discussing the problem of correctness of words, the
theory of Ideas is announced: Socrates does not yet assert its separate existence and
archetypal character, but only says that he dreams sometimes of the existence of
essences (439c and ff). Moreover, he also sets the ontological, epistemological and
logical-semantic requirements that prepare the subsequent appearance of the Ideas in
Symposium, Phaedo and The Republic. In particular: the immutability of essences is
established as a necessary condition for knowledge and language; the thesis of
linguistic immanence and the subjectivism and relativism of the Sophists are rejected.
Lastly, Plato asserts that language is not a secure means of knowledge and that
knowledge should not be the task of the linguist, who is lost in fanciful etymologies,
but of the philosopher as long as he possesses the art of asking and answering. In
short, we can see the Cratylus as Plato’s reflections on the language that his posterior
-for the moment, imagined- theory of Ideas demands3
.
The focus of this work -the dialogue between Socrates, Hermogenes and
Cratylus- is the correctness of words, or rather names, because Plato thinks about
words as if they were names. And the main question is whether there is an accurate
and right description of things depending on their nature, or whether we have to
consider the names as a result of the use and conventions, that is, of human
2
For instance, see Plato, Cratylus, Parmenides, Greater Hippias, Lesser Hippias, Harvard University
Press, Cambridge, 1996.
3
For a more comprehensive analysis of the Cratylus and a discussion of the relevant literature, see
Defez, A., “Llenguatge i Coneixement en el Cràtil de Plato, “ in Enrahonar. Quaderns de filosofia nº.
28, Universitat Autònoma de Barcelona, Barcelona, 1997, pp. 123-143. Also in www.defezweb.net
4
agreement. Put in another way: do things require a particular type of names or would
all names be casual and arbitrary. The problem, as we see, is not only semantic, but
ontological as well, since what is at stake is whether there is also a permanent and
stable nature of things. And here we have four possibilities:
(i) Essences do not exist, and names are entirely arbitrary and conventional.
(ii) Essences do not exist, but names are right and accurate.
(iii) Essences exist, but names are entirely arbitrary and conventional.
(iv) Essences exist and names are right and accurate.
At first Hermogenes seems to defend (i) because he proposes the conventional
nature of names and an ontological relativism in the way of Protagoras. In turn, (iii)
might have been argued by Democritus –a character who does not appear in the
dialogue- in saying that the objects were random combinations of atoms, and to
defend a linguistic conventionalism. On the other hand, Cratylus proposes (ii): in
ontology he is a disciple of Heraclitus and, however, advocates for the natural
correctness of names -in his opinion, there would be a right and accurate way of
naming common to all men. Finally, who defends (iv)? It’s hard to say. This option
seems to be in line with an onomatopoeic interpretation of language: words express
with sounds the essence of things. Obviously, it is not Plato’s position: he does not
defend it in The Seventh Letter, where he accepts the linguistic conventionalism;
moreover in the Cratylus he openly criticises the method of etymologies.
As we see, a name can be accurate and right either by nature or convention.
Thus, for Cratylus, there is an exact description of objects that is identical both for
5
Greeks and Barbarians because, despite the disparity in conventions, the nature of
objects would be the same for all: Heraclitus’s permanent movement. By contrast,
according to Hermogenes, the names would be accurate in a conventional or arbitrary
way because there is nothing in names or in objects that force both of them to go
together: simply, when a word is the name of something, it is its name. Now, whether
by nature or convention, what lies behind the opinions of Hermogenes and Cratylus is
the logical-semantic theory of the impossibility of speaking falsely: because
conventions are the ultimate criterion for correctness; or because by natural
correctness all names would be true and there are not false names -names that are not
names- nor false combinations of names. And this is no an accident: the thesis of the
impossibility of speaking falsely, and the subsequent linguistic immanence, is what
accompanies both Hermogenes’s relativism and the defence of Heraclitus’s ontology
and the method of the etymologies made by Cratylus.
In turn, Plato’s strategy is not to deny the idea of correctness because it would
destroy the true-false nature of language, but to criticize the naive interpretations of
correctness and their harmful consequences. Moreover, it is possible to discover in the
Cratylus an interpretation of the correctness of words that seems to try a synthesis
between conventionalism and naturalism. In fact, Socrates appeals to the mental
representations of things we have, and so words, even whether they are a result of
social conventions, would designate things through what we think of them, and these
affections of the soul would have the natural correctness sought (434e-435b).
Therefore, anticipating what Aristotle wrote later in the Peri Hermeneias (16a 3-8),
Plato seems to have in mind the idea that words designate things in an immediate and
conventional way, while mental affections do it instantly with a natural correctness.
6
But let us turn now to the criticism of the naive interpretations of correctness,
and begin by the answer of Socrates to Hermogenes. As we have said above, the issue
is not only logical-semantic, it is not enough to show that arbitrary conventionalism
contradicts the existence of a true-false speech (385a-387d), but also an ontological
and epistemological one. It is necessary to show that relativism and subjectivism do
not work either: if we can speak truly or falsely is because things have a permanent
essence that is independent of our language (386d-e and 439b-440c). Moreover, men
are neither the measure of things, nor is it certain that anything we say is always true.
On the contrary, far from the linguistic immanence, things exist in themselves and in
accordance with their essence, and we can tell truths or falsehoods about them. But
what explains this true-false nature of language? To answer this question, Plato
analyses the act of naming and the figure of a legislator in the origins of language
(387d-391a).
Naming, as whatever activity in general, would have its own essence: things
have to be designated according to their nature, so we cannot make words mean what
we want. Names designate objects according to their essences -that is, names serve to
highlight and describe things that exist by themselves and according to their nature.
Language, therefore, does not create these distinctions, but these distinctions exist
before language, and language only records prior ontological divisions. But how
might names do it? In fact, we have only the names that social use has given us. Now,
names might have another origin, though we possess them as social heritance. In this
sense, Socrates calls on the work of a legislator, an architect of names that in the
beginning would have established them starting from sounds and syllables in order
that they conform to the essence of objects (388e-390b).
7
This resort to this ideal and mythical theory of language seems to obey a
double purpose: to subordinate philology (linguistics) to dialectic and, in the second
place, to subordinate language and knowledge to ontology. The idea is the following:
whatever the origins of names are, names have some natural correctness because they
are tools of ontological training. Now, this original correctness would not be strong
enough and we cannot leave ontological research in the hands of etymologists; this
research corresponds to dialectics, the philosophers (390c). In fact, if we interpreted
the hypothesis of the legislator literally, the task of the etymologist could be in order:
maybe the reconstruction of origins carried out by the study of etymologies might
exhibit the correctness of names that the evolution of language had disguised. So, it is
not a surprise to see that Socrates recognizes that there is some truth in the theory of
natural correctness, and yet afterwards, when he debates with Cratylus, tries to
demonstrate the useless role of the hypothesis of the legislator understood as a
rational explanation of the origins and correctness of names.
For its part, the criticism to Cratylus is already implicit in the rejection of the
etymologies as a means of access to the essences of things: this is a useless approach
and, besides, is open to the risk of talking for the sake of talking, typical of sophists,
young people and poets. Moreover, we have to take into account that, for Cratylus,
etymologies were not the same thing as for us -historical reconstruction of words from
their previous significance. No, in that time etymologies surely had a transcendent
value: to uncover the meaning that words would have had originally when ex
hypothesi language and reality fit each other, that is to say, to reveal the true meanings
that had been disguised in their use by spurious elements. In this way, Cratylus, from
8
his linguistic immanence, thought that etymologies would confirm Heraclitus’s panta
rei (402a and ss, 411b-c, 436e and 437d).
On the contrary, Socrates uses the method of etymologies ironically: he cannot
recognize the huge and wonderful display of wisdom that apparently this procedure
involves on its own, and he considers it necessary to ask the help of some god to
explain it (386c-e, 401e and 428c-d). But not only that: Socrates also tries to show the
implausibility of this practice and the theory of the legislator that sometimes
accompanies it (421c-427d). For this purpose, he distinguishes between primitive and
derived names, being the explanation of the correctness of the first, which is
necessary because the primitive names would be the perfect candidates for imitating
the essence of things through the onomatopoeic value of sounds. Socrates, sometimes
comically, discusses the imitative value of 14 letters of the 24 that made up the Greek
language.
Now, the true criticism of the naive conception of the natural correctness
consists of two stages. First, Socrates discusses the concept of likeness, which would
be the ideal candidate to explain the imitation, and his conclusion is that likeness is
always incomplete and imperfect because if it were not so, the imitation -words in this
case- should only be a duplicate of the thing imitated. And this need for a difference
in the representation, according to Socrates, destroyed the supposed natural
correctness of words based on likeness or imitation (432d). In this sense the response
of Cratylus is significant: the legislator is infallible -if you want, that the necessary
difference in the representation only affects the accidental features of things
9
represented. In fact, Cratylus concludes, for this reason there are not incorrect names
and it is impossible to speak falsely.
But this answer, and we are already in the second moment of the critique, is
precisely the loophole that Socrates will not leave to Cratylus: granting that imitation
happens through a complete and perfect likeness, yet it might be that the legislator
had acted incorrectly making an improper use of names, that is, attributing mistaken
names to objects. In other words: Socrates does not accept the infallible divine nature
of the legislator, but he humanizes him: why might the legislator not have committed
systematic errors? Thus, the hypothesis of a legislator, which could have some value
as a myth that goes where our rational capacities are not able to arrive, becomes
useless when it is used philosophically as an explanation of the origins and
correctness of the names.
And this is not the worst difficulty. The action of a legislator -infallible if you
want- might not explain the correctness of names because he should already know in
advance the nature of things, and this knowledge is only possible, according to
Cratylus, throughout the very names (438a-e). Put in another way: to create an
onomatopoeic language, it would be necessary to understand this very language
previously, since to decide whether a name has a specific meaning it is necessary to
know what kind of things this word might designate. In other words: to know if a
name reflects the essence of a particular type of object with natural correctness, we
should previously know what this essence is. And it is illuminating that this argument,
which certainly brings to mind Wittgenstein’s critique of the ostensive definition, is
10
precisely the only one that Cratylus accepts as reasonable, an argument that Cratylus
can only answer with his ad hoc resort to an infallible divine force (431e and 436c).
To sum up, Cratylus was not right to say that to know how to use words is the
same as knowing the essence of objects (435d). No, to know things we have to go to
things themselves, to their essences (439b). And this is precisely what Plato tries to do
in interpreting the natural correctness and conventionalism in a non naive way. If
something in the word does not represent anything in the thing and, nevertheless,
contributes in some sense to the meaning, then it corresponds to the social use of
language, the conventions of the polis -not subjective and arbitrary decisions as
Hermogenes thinks (434e-435c). As we saw above, according to Plato, words would
be linked conventionally to what we think, so that conventions make us recall the
thought -the affection of the soul- that directly represents the essence of things. Put in
another way: the name-object relationship is mediated by the affections of the soul of
the speakers, that is, the mental states we have when we use words. Now, how should
we interpret the concept of affections of the soul and correctness that Plato seems to
give them?
In the case of Aristotle the semiotic triangle is made up of words, affections of
the soul (the sensible and intelligible forms of the objects captured by the soul) and
the objects themselves or any of their qualities. In particular, the relationship "wordaffection
of the soul" would be conventional, and the relationship "affection of the
soul-object" a causal one. The ontology and epistemology of Aristotle permits the
conception of the affections of the soul as effects of the action of objects upon the
soul, and words as effects of the affections of the soul, although physically built in a
11
conventional manner. To sum up: the sensitive forms and the intelligible forms are
both present in the objects and they would be the formal cause of the affections of the
soul; in turn, these affections would be the formal causes of words, and the linguistic
conventions the material causes of words. Now, what can we find in the Cratylus?
Well, it is obvious that we cannot find in the Cratylus a philosophy of
language such as we understand it in the present, but simply, as we have noted before,
the reflections on the language that the Ideas -for the moment only dreamed- demand.
However, without forcing Plato’s thought too much, it is possible to say the
following. For the case of individuals, names would designate objects through mental
representation of the speaker and the appropriate social conventions. In turn, the case
of universal and abstract words should not have to be different. For Plato -the Plato of
the theory of Ideas- the affections of the soul would be caused not by the sensible and
intelligible forms present in the objects, as Aristotle would say, but by the Ideas that
the soul knew before its current life in the body: the traces that the Ideas would have
left upon it.
Now if this might be the solution of Plato’s late philosophy, it is not entirely
clear what his position in the Cratylus is. As we said in the beginning, in this dialogue
Plato still does not recognize a separate existence to the essences, and besides the
theory of reminiscence and the immortality of the soul do not play any role. In fact,
the Cratylus would be a tentative work: Plato would still be developing his theory of
Ideas and the epistemology and the metaphysics that this theory needs. In this sense, it
is not surprising that Plato does not make it clear in this dialogue how to understand
the relationship of natural exactness -non naive, non onomatopoeic- between words
12
and essences: essences are only a requirement and they have an existence only
dreamed of.
*
We have said before that Plato’s criticism of the hypothesis of a legislator of
language brings to mind Wittgenstein’s criticism of the ostensive definition. However,
the affinities have to be properly understood. First, we should have in mind that
Plato’s approach to linguistic meaning is precisely what is behind the view that the
origins of language, both individually and in the human species, have to be
understood according to the ostensive definition. In the second place, we should not
forget that Plato’s rebuttal of the hypothesis of a legislator is not the rejection of the
myth itself, but the dismantling of the use of this myth with an explanatory intention.
For Plato, there are issues beyond human capabilities, and this would be the place for
myths. Now, myths cannot claim to be rational explanations –in this case, reason
would destroy them as easily as Socrates does-, but they must remain as myths, and as
a reminder of our limitations. In fact, is the theory of reminiscence less mythical than
the hypothesis of a legislator of names?
What is more, it would be possible to rise to the theory of reminiscence the
same criticism that Plato uses against the hypothesis of a legislator of names. In
effect, why would the soul not make systematic errors in the contemplation of Ideas?
And how could the soul distinguish, identify and understand the Ideas -their
meanings- without the help of language and the activities carried out by the souls
from her embodied existence? Let us, for the moment, leave aside this problem and
see how Plato’s criticism of the hypothesis of the legislator belongs to a philosophical
13
scenario very different to the scenario of Wittgenstein’s criticism of the ostensive
definition.
First, we have the paradigm of names and so the assimilation of all semantic
function to the nomen-nominatum relationship in which the meaning of a name is its
reference, the designated object. Two: an incipient ideacionism to the extent that
words would refer to things through a mental intermediary, which fits things with
natural correctness, but conventionally attached to words. Moreover, we have
metaphysical realism that is what semantic realism needs. Indeed, to assert that the
basic function of language is to describe reality and that the basic function of words is
to name its components we need to assume that reality is already segmented and
structured in entities -Ideas, in the case of Plato- and that they exist by themselves and
are what they are regardless of language. Only in this way can we say that words
become names of these entities and that the language describes reality.
Lastly, these theses also presuppose the idea that exists or makes sense to
think about the existence of one knowledge of reality that is prior and independent
(previous in time and in a logical sense) to the language –for instance, the soul
contemplating the essences or Ideas-, because only in this way would it be possible to
distinguish, identify and understand the components of reality and then give them a
name according to the uses and conventions of the polis. This epistemological thesis
is clearly an intellectualism and it is an inseparable companion of metaphysical
realism.
Well, once these assumptions are accepted –semantic realism, ideacionism,
metaphysical realism and intellectualism-, what one would expect to happen in fact, in
14
the Cratylus, does not happen: a defence of the ostensive definition, and an initial
baptism of components of reality. However, in this context we should remember that
Plato seems to mean that there is not a rational explanation of the origins of language,
and that we cannot say anything about it other than through myths. Obviously, we
might also ask whether it is necessary to speak about the origins of language in the
way of platonic presuppositions. Why should we imagine the origins as an initial
baptism or a set of ostensive definitions, although we later say that it is only possible
to construct myths: in fact, would it not be possible to imagine the origins of language
in another way.
Wittgenstein does not accept Plato’s assumptions, nor does he defend any
linguistic idealism, despite what has sometimes been said4
. In his opinion, reality
exists and is structured in entities and facts, but it only makes sense to say so from our
natural ways of acting (sometimes mere reactions, but often spontaneous, symbolic,
intentional, creative actions) and, of course, from the ways of speaking that have
grown, and still grow, from this natural behaviour; interwoven with it, continuing it,
replacing it, enriching it, etc. Therefore, it would be meaningless to go, as
metaphysical realism proposes, beyond this "in such a way human beings act and
speak”, because only these ways of acting and speaking -not any intellectual
contemplation- is what makes it possible to distinguish, point up, identify, and make
meaningful the entities that we say constitute reality. So, it would be meaningless
both to speak about reality itself –the concept of reality itself is a senseless- and to
4
See Wittgenstein, L., Philosophical Investigations, Basil Blackwell, Oxford, 1958; The Blue and
Brown Books, Basil Blackwell, Oxford, 1964; and On Certainty, Basil Blackwell, Oxford, 1969.
15
claim that there is a non-human way of knowing reality itself. There is nothing to say
or think about5
.
As Wittgenstein’s criticism of private languages shows, only inside our ways
of acting and speaking are we able to distinguish, point up, identify, and make
meaningful the entities that we say constitute reality. And so, only inside this reality,
that is always a human reality, and it does not make sense to think of another, the
words come at times, but not always or primarily, have a reference value. In short,
what Wittgenstein seems to mean -and to say this is a bit risky because it has the
appearance of a theory and Wittgenstein denied that philosophy had the role of
building hypotheses, theories or explanations- is that language is part of our natural
and social history –the natural and social history of the animals we are-, and that in
this process nothing resembles an initial baptism, as the platonic view suggests, in
which things are in front of us waiting to be labelled with words.
Rather, what we find is the oldest geological strata: deeper, more primitive
language games, in which words above all have primarily an expressive function. In
fact, in our languages such expressive uses still survive in a fossilized way; and not so
fossilized, because they are still in use. Moreover, the learning of language -as it is
known, Wittgenstein thought that the description of how people learn to use a word
5
I have dealt these aspects of Wittgenstein's philosophy in: “Realism without empiricism”, in Anales
del Seminario de Metafísica, Universidad Complutense, Madrid, 1994, pp. 13-26; “Racionalitat,
llenguatges privats i ontologia” in Taula n º 29-30, Universitat de les Illes Balears, 1998, pp. 65-74;
“Realismo esencialista y nominalismo realista. Acerca del conocimiento del mundo”, in Pensamiento
Vol. 54, nº. 210, Universidad Pontificia de Comillas, Madrid, 1998, pp. 417-442, and “¿De qué sujeto
habla el segundo Wittgenstein?” in Ariso, J.M. (ed.), El yo amenazado. Ensayos sobre Wittgenstein y el
sinsentido, Biblioteca Nueva, Madrid, 2010, pp. 111-128. Also in www.defezweb.net
16
was of vital importance to understand their actual uses- shows how the referential and
descriptive functions of language in many cases develop from its expressive features.
Now, the expressive function of words would not be the end of story because
language creates fields of meaning -worlds- within which human existence takes
place. And not only how culture performs this function, but in the most basic and
primary sense that humans beings do things with words (situations, relationships,
intensional realities), because the primary function of language would not be to name
and describe reality -we have seen that naming and describing only happen inside
language and through language. In this sense, we have to notice that to speak of
language in the case of Wittgenstein might be misleading: it would be better to speak
of linguistic action and symbolic activities.
And this detail is of paramount importance to understand the philosophical
distance between Plato –Platonism, if you wish- and the later Wittgenstein. Language
is action, a self-constitutive activity that generates its own rules -it takes care of itself
and does not require anything external to justify its regulations-, and whose origins
should not be understood throughout intellectual acts of recognition of entities or rules
–the entities of a mythical initial baptism, the rules that speakers should understand if
they are to be able to use words. As we have seen, language is a self-contained
activity developed from the common natural and spontaneous way that human beings
act.
In the beginning we said that our intention was not to demonstrate that one of
the two authors about which we were going to speak –Wittgenstein- was right about
the other –Plato-, but simply to show that Wittgenstein’s approach was much more
17
fruitful. Likewise we can now add that this approach also frees us of philosophical
responsibilities in relation to the origins of language since, from Wittgenstein’s
position, this question seems to fade or dissolve. First, because we can leave off
seeing this problem as something needed for a philosophical explanation. And in the
second place, because we can remove from our philosophical agenda questions such
as the following: how did an intellectual being –man- begin to talk in a referential
and true-false way of a reality, a reality that exists in front of him and apart from him
and is ontologically organized by itself?
Wittgenstein thinks about the matter otherwise, and seems to give us these two
pieces of advice. On the one hand, that we should contemplate language as an
evolution and development of the natural and spontaneous action of human beings;
and on the other, that philosophy should only make a very general description –not
hypotheses, not explanations- of this evolution and development. Better yet:
descriptive observations of the uses and the learning of words that permit us to
imagine transitions between linguistic and non-linguistic actions, and also between
linguistic and non-linguistic primitive actions. And in this way we arrive at a very
well-known place: before the current eggs and chickens, there existed even more
primitive chickens and eggs.