-L-A-N-G-U-A-G-E-
by David Mertz
An Honors Thesis
University of Colorado, Boulder.
Fall, 1987
by David Mertz
_*Table of Contents_*
*0.0 Introduction. *.......................................... 1
*1.0 Saussure.*............................................... 1
*1.0.1 Terminology.*.......................................... 2
*1.0.2 Methodology.*.......................................... 3
*1.0.3 The negative definition of the sign.*.................. 4
*1.0.3.1 Phonetic definition.*................................ 4
*1.0.3.2 Semantic definition.*................................ 5
*1.0.4 The objectivity of the sign.*.......................... 5
*1.0.4.1 The continuity of new generations*................... 7
*1.0.4.2 The multiplicity of signs and complexity of _langue_* 7
*1.0.4.3 Collective inertia or the double-blind predicament*.. 7
*1.0.5 Idealism and Teleology.*............................... 7
*1.1 Criticisms of Saussure.*................................. 8
*1.2 Volosinov.*.............................................. 9
*1.2.1 The subject.*.......................................... 9
*1.2.2 Saussure's Idealism.*.................................. 10
*1.2.3 Volosinov's Semioticity. *............................. 11
*1.2.3.1 Saussure versus Volosinov.*.......................... 12
*1.2.3.2 The absence of langue.*.............................. 12
*1.2.3.3 Economic determinism.*............................... 13
*1.3 Lacan.*.................................................. 13
*1.3.1 The mirror stage.*..................................... 14
*1.3.2 The Other and the Phallus.*............................ 15
*1.3.3 The autonomy of the signifier.*........................ 15
*1.4 Kristeva - Semiotics and Semiology.*..................... 16
*2.0 Sentence-meaning and utterance-meaning.*................. 18
*2.1 Semantic theories.*...................................... 18
*2.1.1 Extensional theories.*................................. 20
*2.1.1.1 Possible-World semantics.*........................... 20
*2.1.1.2 States-of-the-world semantics.*...................... 21
*2.1.1.3 Verification semantics.*............................. 21
*2.1.2 Intensional theories: Chomsky.*....................... 22
*2.1.2.1 Recursive syntax.*................................... 22
*2.1.2.2 Syntactic structures.*............................... 23
*2.1.2.3 Synonymy and paraphrase.*............................ 23
*2.1.1.4 Entailment.*......................................... 25
*2.1.1.5 Relations of degree.*................................ 26
*2.1.1.6 Presupposition.*..................................... 27
*2.1.1.7 Category agreement.*................................. 29
*2.2 Grice.*.................................................. 32
*2.2.1 Conversational maxims.*................................ 32
*2.2.2 Intentions.*........................................... 33
*2.3 Austin.*................................................. 34
*2.3.1 Infelicities.*......................................... 35
*2.3.2*....................................................... 36
*2.3.3 Performatives: implicit and explicit.*................ 37
*2.3.4 Constative utterances.*................................ 37
*2.3.5 Truth and falsity.*.................................... 39
*2.3.6 Conclusions.*.......................................... 40
*2.4 Davidson.*............................................... 41
*2.4.1 Prior theory and passing theory.*...................... 42
*2.4.2 Malapropisms.*......................................... 42
*2.4.3 Language?*............................................. 43
*2.4.4 Language revisited.*................................... 44
*3.0 Conclusion.*............................................. 45
*3.0.1 Some "facts".*......................................... 45
*3.0.2 Some claims.*.......................................... 46
*3.1 Once again: language?*.................................. 47
*3.2 A re-examination.*....................................... 48
*3.2.1 Chomsky revisited.*.................................... 48
*3.2.2 Stereotypes.*.......................................... 49
*3.2.3 Semantics, grammar and pragmatics.*.................... 50
*3.3 Conventions and intentions.*............................. 50
*3.3.1 Conventions.*.......................................... 51
*3.3.2 Intentions.*........................................... 51
*3.4 Lotman.*................................................. 52
*3.4.1 Sameness and difference.*.............................. 52
*3.4.2 Code and content.*..................................... 53
*3.4.3 Diachrony.*............................................ 53
*3.4.4 Language concluded.*................................... 54
*Bibliography and key*........................................ 55
"no entity without identity"
- Quine
*0.0 Introduction. *
A perennial problem of twentieth century philosophy, linguistics,
semiotics, and psychology (as well as other fields) has been the
distinction between language ('langue') and speech ('parole'). This
distinction is very similar, if not identical, to the distinction
between semantics and pragmatics. Or still closer, we may distinguish
between the entire phonetics-phonemics-morphology-syntagmatics-syntax-
semantics axis, and pragmatics. A distinction of this sort arises,
not only with respect to language and speech, but to any semiotic
system. It is the difference between an abstract system of
oppositional sign elements, and the application of this system by
actual concrete beings, notably persons. There are, of course,
semiotic systems other than those operated (and operated on) by
persons; particularly those studied by zoosemiotics, cybernetics,
information theory, and computer science. In the latter three of the
fields which we have mentioned there is an obvious, and justified,
program/implementation split, corresponding to the langue/parole
distinction. In all of these fields the object studied is an
artifact, and one created for specific purposes. Hence, whenever
there is a divergence between program and its implementation, our
highest level classification of this event must be "(machine) error".
While this "error" classification is most appropriate for
artificial systems, we should be wary of applying it too quickly to
natural systems: the emotive and communicative systems of animals,
natural (physical) systems when viewed as informational, and most
importantly, the objects of the human sciences. With the general
academic denouncement of normative ('prescriptive') grammars, in favor
of descriptive ones, linguistics can seem to have freed itself of an
improper reliance on the "error" classification. However, linguistics
- even descriptive - has with philosophical collusion maintained an
essentially normative and teleological[1] conception of persons qua
speaking beings. This tendency, though far from universal, arises
over and over again in many guises, and in many intellectual
traditions. In this paper we shall try to examine several of the
normative systems which would have one favor language over speaking,
and semantics over pragmatics. As we have just indicated, our focus
shall be primarily on natural languages, and the various fields which
concern themselves with natural languages. In the end we shall try to
present a program for study of natural "languages", for philosophers
and others, which is fully materialist and optimally nominalist; which
favors speech over language, and pragmatics over the semantics axis.
*1.0 Saussure.*
The rift between language and speech in modern linguistics
originates with Ferdinand de Saussure. This is not to say that no
such rift exists prior to Saussure. It does exist, in linguistics
with Wilhelm von Humboldt; in philosophy with almost everyone,
starting with Plato. But it is Saussure who first brought the fully
developed conception of a language/speech split to scientific
linguistics. Saussure has also had an unquestionably great influence
on structuralist thinking in all of the human sciences. It is largely
because of this influence that these sciences tend towards an idealist
teleology whenever they examine persons qua speaking beings. Hence we
shall start our examination with an analysis of Saussure's
conceptions, which are still enormously compelling and useful today -
even to those trying to pursue a materialist or pragmaticist[2]
program.
The idealism and teleology which we attribute to Saussure are
tendencies which he explicitly denies, at least in the form we
attribute them. Hence, we shall have to try to draw out some
consequences of Saussure's thinking which are not made explicit. This
is largely a matter of making clear the ontological status of langue,
langage, and parole; at the same time we must have a critical eye
towards Saussure's claims about the ontological (and epistemic) status
of his entities. However, we shall first try to explain Saussure's
system of thought in a neutral manner.
*1.0.1 Terminology.*
To embark on an exposition of Saussure's thought, we must first
explain the sense of three terms: 'langage', 'langue', and 'parole'.
The latter two of these, in particular, are crucial to Saussure's
thought. The most general of these terms, 'langage', is meant to
embrace all elements of human speaking. These elements include the
physiological construction and reception of sounds by persons' vocal
and aural apparati; the physical facts about transmission of sounds or
other media through which language can be carried; and the
psychological and sociological mechanisms through which understanding
can occur. It is the last of these categories which includes
'langue', i.e. 'language'. Saussure's first remarks about langue are
essentially to this effect. He says,
"[Langue] is both a social product of the faculty of speech and a
collection of necessary conventions that have been adopted by a
social body to permit individuals to exercise that faculty
(p.9)."
This merely points to the social nature of langue, we have not yet
provided a rigorous definition of it. Let us now try to do so.
Langue, as Saussure conceives it, is the arbitrary[3] matching up of
"sound-images" with "concepts".
Although, langue is socially determined, it is essentially a
psychological entity. Saussure says, "If we could embrace the sum of
word-images stored in the minds of all individuals, we could identify
the social bond that constitutes language [i.e. langue] (p.13)." We
are to understand that 'sound-image', as Saussure uses it, stands for
an exclusively psychological object, not a physical sound. Saussure
presupposes some sort of representational theory of mind, in order
that sound-images are conceived as some actual entities in the brain.
This is not, however, really crucial. If we preferred to give some
functional explanation of the mind as a system of virtual
representation, this could easily be accommodated to the Saussurian
conception of language.[4] For Saussure, the primary fact about
langue is its arbitrariness. That is, no logical or natural
connection exists between the particular sound-images and concepts in
correspondence. These correspondences, although contained in the mind
of each speaker, are determined by social convention.
Parole is the "executive side (p.13)" of langage, that is, the
individual and accidental in human speech. Parole may be divided into
two parts:
"(1) the combination by which the speaker uses the language code
for expressing his [sic] own thought; and (2) the psychophysical
mechanism that allows him to exteriorize those combinations
(p.14)."
By 'combinations' Saussure means the links between certain
sound-images and concepts which define a language [langue]. Saussure
is only concerned here with the semantic links in language. We shall
not try to extend this to the syntactic combinations in language,
although these should probably play a closely analogous role to those
played by semantic links. We can then see already that Saussure's
distinctions suppose a certain object for linguistics: namely langue.
This is because even individual speech [parole] does not exist, by
definition, without language [langue].
*1.0.2 Methodology.*
Saussure is quite clear about his methodological preference for
language over speech. He says, "[F]rom the very outset we must put
both feet on the ground of language and use language as the norm of
all other manifestations of speech [langage] (p.9)" Hence for
Saussure, there is no possibility of studying the phonological
production of persons, as informational messages, except through the
study of the language [langue] to which they correspond. It is, of
course, still possible for the physiologist to study the purely
acoustic production of sound, but this study is not one embraced by
semiology, but rather by physics. In later sections of this paper we
shall claim it is possible to study this very subject prohibited by
Saussure. For convenience, we shall hence continue to use 'parole' in
the sense defined by Saussure; however we shall use the term 'speech',
or 'individual speech' to refer the physiological/acoustical aspect of
langage. This is contrary to the convention of the translator of
Saussure's book, who uses 'individual speech' as a synonym for the
French 'parole'.
Let us turn to an examination of Saussure's reasons for focussing
the attention of linguistics on langue, and not parole. His primary
reason for this is that parole is an entirely heterogeneous affair, it
follows the whims and peculiarities of each individual speaker. This
mass is far too unsystematic to formulate any laws about. Langue, by
contrast, is an homogeneous structure, which is composed of links
between signifiers (sound-images) and signifieds (concepts). All the
objects within langue are of exactly the same sort: they are
associations between two different kinds of psychological objects.
*1.0.3 The negative definition of the sign.*
There is another, even more important, sense in which langue is
an homogeneous structure. It is in this respect, which we shall now
explain, that we must call langue a structure, rather than an object.
We shall call this 'the negative definition of the signifier and
signified'. According to Saussure,
"[C]oncepts [and sound-images] are purely differential and
defined not by their positive content but negatively by their
relations with the other terms of the system. Their most precise
characteristic is in being what the others are not (p.117)."
Or again,
"[I]n language there are only differences without positive terms.
Whether we take the signified or the signifier, language has
neither ideas nor sounds that existed before the linguistic
system, but only conceptual and phonic differences that have
issued from the system (p.120)."
This may easily be illustrated, both in the case of the signifier and
in that of the signified.
*1.0.3.1 Phonetic definition.*
We may see the negative definition of the signifier by an
examination of an English phone, e.g. /k/. What distinguishes /k/
from the other English phones /m/, /a/, /t/, /p/, /g/, etc. is not any
positive trait of the sound itself, but merely those features in which
it differs from the other sounds. Hence, /k/ is distinguished from
/m/, /a/, etc. by not being a continuant. /k/ is distinguished from
/t/, /p/, etc. by its articulation being behind the velar ridge. It
is distinguished from /g/ only by its lack of voicing. However, /k/
has no positive quality of its own; any sound which is non-continuant,
unvoiced, and articulated behind the velar ridge is an English /k/,
whether it be implosive or explosive, aspirated or non-aspirated,
palatalized or non-palatalized, or having a palatal-velar, palatal, or
glottal articulation. We know that these possibilities do not produce
a common sound by the fact that they are distinguished in some
languages.
*1.0.3.2 Semantic definition.*
The same situation can be shown on the signified side of the sign
(where a 'sign' is defined as a signifier/signified link). For
example, Sanskrit distinguishes three degrees of number, where English
only distinguishes two. Conceptually, the Sanskrit plural does not
have the same value as the English plural. However, this cannot be
because of some positive value of the Sanskrit concept ['plural'][5],
because every group of objects, real or imagined, which would be
classified in Sanskrit to be plural, would be similarly classified in
English. Rather, the difference in conceptual value of the plurals of
the two languages comes solely through the difference in the
signifieds to which they are opposed. In Sanskrit the plural
classification has two numeric oppositions: ['singular'] and
['dual']; whereas in English there is only the opposition
['singular']. The example we have given is one of an opposition of
grammatical morphemes; however, it is clear enough that the same
applies to lexical oppositions. An example of this might be the
English words 'sensual' and 'sensuous', or 'continual' and
'continuous', (or rather, the signifieds of these words), which once
had a specific conceptual opposition, but have generally lost it.
Hence English is left without the means of distinguishing two
concepts, because the concepts themselves were a product of an
arbitrary division of thought.
*1.0.4 The objectivity of the sign.*
Having sketched some of Saussure's thinking, it is now possible
to bring out the aspect of his schema which is really the important
one for the purposes of this paper. We shall call this aspect the
'objectivity of the sign'. Since the sign is a psychological entity,
the objectivity of the sign must be, in essence, a phenomenological
claim. That is: it is claimed that each sign, and the whole
structure of relations of signs, is present to each speaker as an
unchangeable, impervious whole. In Saussure's words, "[Langue] is . .
. outside the individual who can never create nor modify it by
himself; it exists only by virtue of a sort of contract signed by the
members of a community (p.14)." This contract[6] is, in a phrase of
Lacan, "always already given." However, langue is not only outside
the will of the individual, it is also outside the will of the
collective. Saussure says this slightly later, "[Langue] always
eludes the individual or social will (p.17)." We see this in the
various social attempts, such as the French Academy's, to regulate
language; these never entirely succeed.
The objectivity of the sign stems, for Saussure, from its very
arbitrariness. Even where there is a partial iconic[7] value to a
sign, this is dominated by element of arbitrariness once it joins the
system of langue. This is clear in the case of onomatopoeia. Even
where a word originates because of its natural connection with the
thing it stands for, once the word exists it is subject to the same
mutations as any other word in the language. It may seem paradoxical
that a sign can be both arbitrary and conventional, and yet impervious
to change by either an individual or the collective will. Saussure
acknowledges this,
"The signifier, though to all appearances freely chosen with
respect to the idea that it represents, is fixed, not free, with
respect to the linguistic community that uses it. The masses
have no voice in the matter, and the signifier chosen by language
[langue] could be replaced by no other. This fact, which seems
to embody a contradiction, might be called colloquially, 'the
stacked deck.' We say to language: 'Choose!' but we add: 'It
must be this sign and not other.' No individual, even if he
[sic] willed it, could modify in any way at all the choice that
has been made; and what is more, the community itself cannot
control so much as a single word; it is bound to the existing
language (p.71)."
This seeming paradox may be reconciled in several ways, as its
peculiarity is really a conditioned reaction, not a product of the
logical structure of the two facts about langue. Some of the facts we
can point out to quell our reaction to the two "facts" about langue,
are i) the continuity of new generations of speakers; ii) the
multiplicity of signs and over-complexity of the system; iii) what
Saussure calls the 'collective inertia towards innovation, which we
shall call the 'double-blind predicament of speakers'. We shall make
a few remarks about each of these above points, but it should be noted
that these remarks, in keeping with the nature of the task at hand,
shall be heuristic rather than rigorous.
*1.0.4.1 The continuity of new generations*
The continuity of new generations. Since human generations are
not, even approximately, divided into distinct generational groups,
each new speaker is faced by an overwhelming majority of established
speakers. This is the same as the fact that every language, at every
time in history (and prehistory), is already an historical object.
*1.0.4.2 The multiplicity of signs and complexity of _langue_***
The multiplicity of signs and complexity of langue. This point
is certainly not any sort of "in-principle" reason for going along
with the simultaneous objectivity and arbitrariness of the sign.
However, the general point here is that since elements of langue are
relatively motivated (e.g. regularity in declension and combination of
lexical and grammatical morphemes), any change in a particular sign
would involve the reworking of a large part of the rest of the
language. This task generally exceeds the limits of a particular
speaker, or even collective.
*1.0.4.3 Collective inertia or the double-blind predicament*
Collective inertia or the double-blind predicament. We find this
to be the strongest point of the three. Saussure puts it merely that
no speaker can change a sign unilaterally, and still achieve
communication with her fellows. The coordination problems (as well as
power-structure problems) prohibit simultaneous (i. e. all speakers at
once) change in a sign, within a collective. We prefer to think of
this problem in decision-theoretic terms. Assuming that communication
is a desideratum, we may not change our semantic connections without
hopelessly threatening the achievement of this goal. This is because
a given speaker cannot know which new value is given to a sign by
another, except through the communicative mechanism which depends on
the system to which this sign in question belongs. Especially where
we are concerned with the possible change of many signs, it is the
most prudent communication-maximization policy to leave the system as
it is aboriginally agreed upon. The consequence of this is that even
where change does occur in language, it must occur item by item, and
slowly enough so as to assure that all speakers have the means to
become aware of the change.
*1.0.5 Idealism and Teleology.*
Now that we have presented Saussure's "objectivity of the sign",
we shall consider our earlier claims about the nature of langue and
parole. We claimed that langue involves us in both idealism and
teleology. Saussure explicitly denies the former of these: "Language
[langue] is concrete, no less so than speaking; and this is a help in
our study of it. Linguistic signs, though basically psychological,
are not abstractions (p.15)." If Saussure were to say that langue is
an abstraction we could stop worrying about its ontological status.
However, since he does not, we must remain concerned. We shall not
examine this here; both because we shall eventually go against
Saussure's own representation of his project, and because the
discussion of langue's ontology shall prove useful in discussing the
thought of several other thinkers.
The teleology in Saussure's langue, unlike the idealism, can be
read directly from his text. We see several suggestions of this
teleology in the introductory sections of his book. Saussure says,
"language [langue] is not complete in any speaker; it exists perfectly
only within a collectivity (p.14)." Or again,
"Among all the individuals that are linked together by speech,
some sort of average will be set up: all will reproduce - not
exactly of course, but approximately - the same signs united with
the same concepts (p.13)."
These suggest that some transcendent langue exists, outside of and
prior to each speaker, and each speaker tends towards this
transcendent structure. The notion of tendency which we use, however,
is not unproblematic. We might approximate it by saying that, for
Saussure, some causal force acts to bring each speaker's speech closer
to the specifications set by langue; as against whatever tendencies
there are for speakers to speak differently. Certainly there is not
the conceptual association of the transcendent langue with the Good
and the True, which is traditional to teleological thinking, but we
still must recognize this as teleological.
In the fourth part, _Geographical Linguistics_, this teleological
thinking is expounded on in more length. Chapter IV as a whole is
concerned with this, so we shall not bother to reproduce it here. We
shall merely say, in brief, how this teleology is explained. Saussure
says that if language, as originally learnt were spoken by each person
throughout her life, then the varieties of speech would be infinitely
diverse. However, through the effect of verbal intercourse, speakers
drop or change those items which differ from those of their fellow
speakers. Further, speakers will not be encouraged in their use of an
innovative item. This general family of observations may be thought
of as a product of a decision-theoretic maximization of communication
by speakers, in much the same way as we presented the collective
inertia due to the double-blind predicament of speakers. Here,
however, the emphasis is not on conservatism but on the
goal-directedness of the community of speakers. Every speaker tries
to use items in such a way as to maximize communication; as a result,
Saussure claims, the speakers tend toward a common language [langue].
This is teleology in a respectable scientific outfit, but it is
unquestionably a teleological theory.
*1.1 Criticisms of Saussure.*
We should now like to move on to some of the criticisms which
have been made of Saussure, and his langue/parole distinction, by
later structuralist writers. In the process of doing this we hope
both to illustrate the idealist presuppositions of Saussure's thinking
and to deny the objectivity of the sign, at least in the sense in
which Saussure conceives it. We shall focus on three writers, each
with her own rather sharply different goal, who level similar
criticism of Saussure's schema. These writers are V. N. Volosinov,
Jacques Lacan, and Julia Kristeva. There are two aspects present in
each of them which speak against the langue/parole distinction as cast
by Saussure. The first of these aspects is the conception of the
psyche. For Saussure it was possible to establish, in a objective
way, what signifier/signified connections were psychologically made by
a particular speaker. This implies a conception of the psyche as a
closed system. Neither Volosinov, Lacan, nor Kristeva are comfortable
with the possibility or coherency of establishing these connections,
precisely because of their conceptions of the human psyche. Each of
these authors tend to drop the subjective consciousness from the
picture, altogether. The second aspect of these authors' thought
which speaks against Saussure's schema is the manner in which
signifier/signified links are present to speakers. Where Saussure saw
these links as being objectively presented to the subjective
consciousness, Volosinov, Lacan, and Kristeva focus on the
presentation of speech in its concrete actuality, to the detriment of
langue in its systematicity.
*1.2 Volosinov.*
Let us first turn to an examination of Volosinov, who is of the
three probably the most accessible. We should like to first discuss
Volosinov's rejection of the Saussurian relation of a speaker to
langue. This point concerns a speaker's own relation to her speech,
and shall keep us closer to the framework of Saussure than the former
aspect above. This former aspect may be said to concern the
interpretive relation of a hearer to another speaker. The conclusions
we shall draw from an examination of this former aspect shall be more
radical, and by the its nature involve us in more complexity. Hence,
we shall defer that discussion momentarily.
*1.2.1 The subject.*
Volosinov admits, perhaps with more generosity than he need have,
that it is at least coherent to speak of the manner in which langue is
presented to the subjective consciousness. However, even this
admission is tempered by an immediate rejection of granting langue an
objective ontological status. This brings Volosinov into immediate
conflict with Saussure, at least under one reading of the latter.
However, this conflict is best left until later for its resolution.
This is because even if Saussure is read as casting langue as fully
objective, it is consequently also true that Saussure sees langue as
being presented objectively. This is merely the rather banal fact
that Saussure does not believe langue to be objective, yet unknowable
- which would be a rather awkward belief. Volosinov states this
admission:
"If we claim that language as a system of incontestable and
immutable norms exists objectively, we commit a gross error. But
if we claim that language, with respect to the individual
consciousness, is a system of norms, that such is the mode of
existence of language for each member of any given language
community, then what we are expressing in these terms is a
completely objective relationship (p.66)."
He goes on, however:
"Whether the fact itself is correctly constituted, whether
language actually does appear only as a fixed and inert system of
norms to the speaker's consciousness - that is another question
(p.67)"
The conclusion, when Volosinov asks, "Does language really exist for
the speaker's subjective consciousness as an objective system of
incontestable, normatively identical forms (p.67)?", will be that it
does not.
As pointed to above, Volosinov believes that rather than towards
language as system, "the speaker's focus of attention is brought about
in line with the particular, concrete utterance he [sic] is making
(p.67)." For Volosinov, this must always be the case; because "the
meaning of a word is determined entirely by its context (p.79)."
Independently of some particular situation of use, there is no
underlying meaning to a given signifier. It is only by reifying some
finite number of uses of a signifier that dictionaries or descriptive
grammars can claim to give _the _meaning of a given word. Volosinov
claims that these descriptive grammars, which retain the term langue,
are only a product of linguistics' historical root in philology.
Where the object under study is a finite corpus in a dead language, it
is certainly possible (and useful even) to try to isolate what is
common to all uses of a given word, or given construction. However,
when we turn to trying to describe all the possible uses (i.e. the
"core meaning") of a word in a living language, we lose what is
essential to the concrete ideological and socioeconomic environment in
which the word occurs, and which gives it its meaning.
Volosinov makes some remarks about the actual presentation of
language to us. Volosinov agrees that words seem to have particular
semantic values in particular situations, but these "words are always
filled with content and meaning drawn from behavior or ideology
(p.70)." Rather than utterances having "only one linguistic
criterion: correct versus incorrect (p.54)",
"the criterion of linguistic correctness is submerged by a purely
ideological criterion: an utterance's correctness is eclipsed by
its truthfulness or falsity, its poeticalness or banality, . . .
[whether it is] good or bad, important or unimportant, pleasant
or unpleasant, and so on (p.70)."
This hints at the program we shall try to pursue in later sections of
allowing grammaticality judgements to play a part in actual speech,
but a part that is no different in kind from that played by judgements
of truthfulness, vulgarity, clarity/unclarity, and the like. For
example, ungrammaticality may be met with censure, but so may
insincerity. It shall be in these comparisons that it shall
eventually prove most poignant to turn back to the teleological nature
of langue.
*1.2.2 Saussure's Idealism.*
The second part, here, of Volosinov's critique of Saussure's
langue/parole distinction regards Saussure's conception of the human
psyche. Before moving on to Volosinov's conception of the psyche, it
shall be useful to push on Saussure's conception, particularly as
regards the ontological status of langue. Saussure implicitly relies
on a Rationalist conception of the psyche as a rational system, which
is radically separated from the world. This generally goes
hand-in-hand with an ontological dualism, in which mind is one kind of
substance, and matter is quite another.
We need not be ontological dualists to adopt a Rationalist
conception of the psyche, however. An alternative is to be material
monists, but give a functional explanation of mind which demands a
sharp distal/proximal split. The precise sense of this split depends,
of course, on the particular functional explanation we give. However,
in general we may explain it as follows. The functioning of the mind
might involve a strict categorization of all cognition into two
categories: those which we call distal and proximal. The distal part
is that which is regularly consequent upon some external stimulation;
the proximal is the "internal" portion of cognition, that which only
stands in relation to other proximal cognition. If Saussure were to
rely on an outright ontological dualism, it would follow immediately
that langue is an ideal, not a material object. This is by Saussure's
very own definition of langue (my p.4): "the sum of word-images
stored in the minds of all individuals." According to ontological
dualism, anything stored in the substance of the mind is an ideal
object.
There is more difficulty in pushing Saussure's conception of
langue if he is read as a functionalist (of mind). However, even
under this reading Saussure cannot maintain his conception of langue.
The two important things to keep in mind are Saussure's claim that
"Language [langue] is concrete . . . [and] linguistic signs . . . are
not an abstraction (my p.11)."; and the objectivity of the Sign. The
objectivity of the Sign demands that, under a functional picture, a
signifier/signified link is presented as a distal object. This is
not, of course, to claim that langue must be a distal object, but
merely that the psyche has the same organization relative to a
signifier/signified link as it does relative to a distal object. Once
we add in that signs are not abstractions, but concrete objects, we
are, however, forced to call signifier/signified links actual distal
objects. We may retreat slightly from this by claiming, not that
signs are realized outside the speaker, but merely that the mind is a
modular structure containing a part which contains representations of
signifier/signified links. Saussure's claim of the concreteness of
signs does, at this point, demand a strictly representational system.
However, now we are faced with Saussure's statement, "[Langue] exists
perfectly only within a collectivity (my p.11)." When we take this
into account is impossible to say any longer that each speaker merely
stands in relation to a modular representation of her own idiolect.
For Saussure, the idiolect is always an imperfect copy of the langue.
However, this excludes the location of the objective langue from a
modular portion of the mind. It is also clear, and Saussure says this
himself, that signifier/signified links are psychological, not
physical objects. Hence the only possible "place" left for langue is,
presumably, some ideal realm. Perhaps the easiest way to resolve this
is to make the move which Volosinov points out; namely, allow that
langue is subjectively presented as objective, but deny it real
objectivity. That is, as Volosinov says,
"Language as a stable system of normatively identical forms is
merely a scientific abstraction, productive only in connection
with certain particular practical and theoretical goals (p.98)."
One of these goals may be the eminently practical one of
communication, but the objective langue remains an abstraction.
*1.2.3 Volosinov's Semioticity. *
Let us now turn to Volosinov's conception of the human psyche.
For Volosinov, the psyche does not have the radical separation from
the world which it does in the Rationalist conception. Rather, "The
individual consciousness is a socio-ideological fact (p.12), [which]
can only arise and become a viable fact in the material embodiment of
signs (p.11)." We are to understand here that "the sign may not be
divorced from concrete social forms of intercourse (p.2)." Hence
where Volosinov speaks of the 'sign' he means, in Saussure's closest
approximation, parole. More neutrally, we may equate Volosinov's use
of the word 'sign' with 'utterance', where the latter is thought of as
a purely physical event. The collection of all the habitual forms of
utterance within a particular group Volosinov calls 'ideology', or
more specifically, 'behavioral ideology'. Keeping this in mind, we
are to understand that "there is no basic division between the psyche
and ideology; the difference is one of degree only (p.33)." And
further, "[T]he mutual delimitation of the psyche and ideology can be
solved on the unitary territory of the ideological sign which embraces
both (p.39)." The picture which is being drawn here is of the human
psyche occupying, not some pre-given role in the organism homo
sapiens, but a particular section of the ideological system of a
society.
Hence as the ideological realm of signs is constitutive of the
human psyche, there is no part of the mind which can properly be
called pre-linguistic, or at any rate, no part which is pre-semiotic.
This is true especially of experience, which to the Rationalists was
the essence of the subjective mind. Volosinov says, for example,
"The experiential, expressible element and its outward
objectification are created, as we know, out of one and the same
material. After all, there is no such thing as experience
outside of embodiment in signs. . . . It is not experience that
organizes expression, but the other way around - expression
organizes experience (p.85). . thought and experience,
[consciousness] has already constituted a social event on a small
scale and is not an inner act on the part of the individual
(p.90)."
This complete semioticity of consciousness shall prove very
threatening to Saussure's objectivity of the Sign.
*1.2.3.1 Saussure versus Volosinov.*
For Saussure, the sound-image and concept must exist in the mind
of a speaker, prior to being joined together. It is admittedly true
that, for Saussure, neither the mass of phonic representations, nor
that of concepts is differentiated prior to being divided by language.
This is clearly a step away from a Rationalist picture of clear and
distinct ideas, which are differentiated by their individual natures.
As we have pointed to above, Saussure's concepts, and his
sound-images, exist purely through their differentiation from one
another. Nonetheless, there is a pre-given medium of thought in which
these concepts and sound-images are placed; and this medium is
necessarily pre-semiotic. With Volosinov, even this pre-semiotic
medium is taken away. Volosinov's most startling, though subtle,
phrasing of this is: "A sign can only be illuminated with the help of
another sign (p.36)." Here we are to read sign in the sense in which
Saussure proposes it, namely as a signifier/signified link; at least
this is the interpretation we shall push.
*1.2.3.2 The absence of langue.*
Let us examine some of the consequences of this view of
Volosinov's. It first of all insists that a sign is not illuminated
by an always already given psychological link between signifier and
signified. The illumination of a given sign call it 'a', can only
come through the production or utilization of a second sign, whose
signifier is the signified of 'a', or whose signified is the signifier
of 'a', or which signifies 'a' itself. These signifiers of signifiers
find no stopping point in the subjective consciousness, and hence no
already given langue. Rather, as later Structuralists shall claim,
the "unending chain of signifiers" must find its ground in something
which is beyond signification itself. For Lacan, for example, this
something is the drive; for Althusser, it is the societal totality,
and ultimately the means of production.
*1.2.3.3 Economic determinism.*
The latter view is probably close to Volosinov's belief.
Volosinov really only points at this economic determinism, but it is
nonetheless present. He says, for example, "Production relations and
the sociopolitical order shaped by those relations determine the full
range of verbal contacts between people (p.19)"; or again,
"In order for any item, from whatever domain of reality it
may come, to enter the social purview of the group and elicit
ideological semiotic reaction, it must be associated with the
vital socioeconomic prerequisites of the particular group's
existence; it must somehow, even if only obliquely, make contact
with the bases of the groups material life (p.22)."
A few other comments to this effect are also in his writing. The
importance of this is not to try to find the exact basis of this
"something" which grounds the signifiers in Volosinov. This exact
basis is never given. The importance of these quotes, immediately
above, is to show that Volosinov realizes the need for some thing to
ground the ideological/semiotic system of signs. In a relatively
natural extension of Marx, Volosinov identifies semioticity with
ideology, and hence grounds it in the relations and mode of
production.
The exact ground of the sign in Volosinov is unimportant to
seeing how Volosinov's conception of the human psyche is incompatible
with Saussure's notion of a subjectively objective langue. This
incompatibility follows very immediately from the explanations we have
already given. It is quite simply a matter that, for Volosinov, there
does not exist any psychological medium in which to represent, prior
to semioticity, sound-images and concepts and in which to join the
two. The consequences of this might, at first glance, seem to be that
there is simply no way in which speakers can speak. However, as we
shall show later, the actual consequence is short of this, though
still somewhat radical. The actual result of Volosinov's conception
of the human psyche, and the ideology to which it belongs, is that the
signified of the sign drops out, leaving only signifiers connected to
one another. This elimination of the signified is largely common to
all the structuralist thinking which seeks to eliminate the human
subject from its ontology.
*1.3 Lacan.*
With the elimination of the signified, we may tie in the thinking
of the psychoanalyst, Jacques Lacan. It shall not be possible to give
a complete, or even a cursory, summary of Lacan's thought, as he is
perhaps the most enigmatic thinker of the twentieth century. All that
we shall attempt in these confines is some sketches of the parts of
Lacan which are relevant to our rejection of Saussure's langue/parole
distinction. Before we begin, however, it should be acknowledged that
the use to which we shall draw on aspects of Lacan's thought shall be
contrary to his thinking as a whole. In particular, Lacan accepts the
existence of a somewhat Saussurian langue, though not without some
slight modifications. Our effort shall not be to show the actual
train of Lacan's thought, which includes the langue which we shall
reject; rather, we shall merely try to add some aspects of Lacan's
thought to the framework we started in the above discussion of
Volosinov.
We started our rejection of the langue/parole distinction with a
mention of two lines which could lead to these rejection. These were,
briefly, firstly, the nature of the psyche; and secondly, the actual
nature of language in the psyche of a speaker. It is only the first
of these with which we shall concern ourselves, as regards Lacan.
Clearly, with a rejection of the closed, unified nature of the psyche
comes a rejection of the sort of objective presentation of langue to
this consciousness which we found in Saussure. However, our focus
shall be to merely point out Lacan's conception of the psyche; we
shall leave the consequences dangling.
*1.3.1 The mirror stage.*
The chief concept which we shall clarify from Lacan is that of
the 'mirror stage'. Lacan takes this term from the phenomenon in
which an infant presented with its own reflection will be able to
perform acts requiring more coordination that the infant is able to
perform at will. With Lacan, this phenomenon is generalized to
explain the original creation of the symbolic realm, and with it the
subject. This symbolic realm is a term-of-art of Lacan's, which
stands for what we might call the semiological; it is that which
concerns the exchange of signifier for signifier. Lacan describes the
mirror stage:
"We have only to understand the mirror stage as an
identification, in the full sense that analysis gives to the
term: namely, the transformation that takes place in the subject
when he assumes an image (Ecr., p.2)."
Jacqueline Rose makes explicit the linguistic nature of this image,
"For Lacan the subject is constituted through language - the
mirror image represents the moment when the subject is located in
an order outside itself to which it will henceforth refer. The
subject is the subject of speech (Fem., p.31)."
This image which appears in language creates a division within the
speaking subject: on the one hand, the subject retains her aboriginal
drive (the imaginary realm); on the other, the subject internalizes
the contradictory system of signifiers (the symbolic realm). Rose
continues,
"[T]he subject can only operate within language by constantly
repeating that moment of fundamental and irreducible division.
The subject is therefore constituted in language as this division
or splitting (Fem., p.31)."
Hence "for Lacan, men and women are only ever in language. Men and
women are signifiers bound to the common usage of language (Fem.,
p.49)." Lacan puts this denial of the autonomous subject in an
eloquent metaphor,
"A certificate tells me that I was born. I repudiate this
certificate: I am not a poet, but a poem. A poem that is being
written, even if it looks like a subject (FF, p.viii)."
*1.3.2 The Other and the phallus.*
We may give a thumbnail sketch of how this denial of the
autonomous subject, and split in what subject is left, comes about.
We need to bring to the fore two terms for this sketch. These are the
'Other' and the 'phallus'. The Other is the primary object outside
the predeveloped speaking subject. We may think of this primary
object as the mother, or at least call it by the term 'mother'. The
phallus stands beyond the Other, and is, according to the imaginary
order, the object of the Other's desire. Lacan explains these terms
and their relation to the subject:
"If the phallus is a signifier then it is in the place of the
Other that the subject gains access to it. But in that the
signifier is only there veiled and as the ratio of the Other's
desire, so it is this desire of the Other as such which the
subject has to recognize.then the child wishes to be the phallus
so as to satisfy this desire (Fem., p.83)"
Rose echoes this relation, perhaps more comprehensibly,
"[T]he child's desire for the mother does not refer to her but
beyond her, to an object, the phallus, whose status is first
imaginary and then symbolic (Fem., p.38)."
However, "[T]he status of the phallus is a fraud (Fem., p.40)."
There is no object which stands in the place reserved for the phallus.
It is when the pre-subject finds the absence of an actual phallus that
signification begins. A shift to the symbolic occurs, and the absent
phallus is made present, though not as the object it claimed to be,
but rather as a signifier. In our societies, it is the penis which
plays this role of signifier; however, in keeping with Saussure's
arbitrariness of the sign, any signifier could work equally well. In
fact, all signs do work just the way the penis fills in for the absent
phallus: the signifier is necessary only with the absence of the
signified. The phallus does not, however, stand on the same level as
all signifiers. As Lacan says, "The phallus is the privileged
signifier of that mark where the share of the logos is wedded to the
advent of desire (Fem., p.82)." That is, the phallus is the signifier
which bridges the imaginary and the symbolic, and which originates
signification.
*1.3.3 The autonomy of the signifier.*
With the advent of the signifier in the psyche comes "the
incessant sliding of the signified under the signifier (Ecr., p.154)."
That is, the signified drops out of the determination of the psyche
when the phallus is seen to be missing. This process is not confined
to the occurrence of the privileged signifier of the phallus. With
the production of every signifier, a split is recreated in the human
psyche, in which the signifier is inserted into the place of the
absent signified. We spoke of the deletion of the signified at the
end of the discussion of Volosinov; it is here, with Lacan, that we
can explain this deletion. Between the signifier and the signified is
always a play of presence and absence, with the present terms (the
signifiers) forming, as a consequence, an autonomous network. Lacan
says,
"[W]e cling to the illusion that the signifier answers to the
function of representing the signified, or better, that the
signifier has to answer for its existence in the name of any
signification whatever (Ecr., p.150)."
But in fact, "The signifier alone guarantees the theoretical coherence
of the whole as a whole (Ecr., p.126)." Where the whole in question
here is both the system of language and the psyche which reflects it.
With this autonomy of the signifier we arrive at the conclusion
we pointed to with Volosinov; namely, that "the meaning of each
linguistic unit can only be established by reference to another (Fem.,
p.32)." That is, there is no subject who may master a sign, and fix
the signifier/signified links in her mind. For Lacan, as for
Volosinov, "the truth of the subject, even when he [sic] in the
position of master, does not reside in himself, but, as analysis
shows, in an object that is, of its nature, concealed (FF, p.5)." The
truth of the subject may be taken to mean the signification of the
subject; and the concealed object is in the indefinite signifier links
which always exceed the subject. These links must lie in the whole
practice of the community of speaking beings, that is, in ideology.
*1.4 Kristeva - Semiotics and Semiology.*
At this point it may seem as if all we have really done is
substitute the term 'ideology' for the rejected 'langue'. We do not
believe this to be so, and this can be shown in the difference in the
programs of Kristeva's semiotics, and Saussure's semiology. Kristeva
herself does not reject the place of semiology (whose key item is
langue), she merely suggests her semiotics as a supplemental study.
We shall claim not only that semiotics can edge out semiology, but
that it must. Kristeva is well aware of Lacan, and generally works
within the framework set out by him. That is, for both Lacan and
Kristeva, speech exists only at the place of division between the
imaginary and the symbolic (for Kristeva, between the semiotic and the
symbolic). Kristeva puts it,
"[S]emiotics, by studying language as a discourse enunciated by a
speaking subject, grasps its fundamentally heterogeneous nature .
. . [it] is at once system and transgression (negativity), a
product of both the 'drive-governed basis of sound production'
and the social space in which enunciation takes place (Read,
p.25)."
Kristeva's semiotics operates in that space which Saussurian
linguistics refuses to; namely, the individual 'drive-governed basis
of sound production'. Kristeva makes this clear,
"[S]tructural linguistics could not become a linguistics of
speech or discourse; it lacked a grammar, for in order to move
from sign to sentence [our utterance] the subject had to be
acknowledged and no longer kept vacant (Des., p.128)."
It is the place of the subject which Kristeva wishes to establish for
linguistics. However, it is not a Cartesian subject which shall
occupy this place; rather, it is a split subject which speaks only
through the internalization of the radical alterity of the symbolic
realm. However, this subject is only has one foot in the symbolic;
the other is still in the imaginary, the pre-linguistic or
trans-linguistic drive. It is this second realm, which Kristeva
labels the semiotic. She says, "The semiotic is linked to the
pre-Oedipal primary process (Read, p.12)." Since this semiotics
studies just the opposite side of the equation of the speaking subject
as does Saussure's semiology, it demand a new methodology, a new
logic. Kristeva says of this, "[S]emiotic logic [is] of the sociality
in which the (speaking, historical) subject is embedded (Read, p.25)."
That is, it is the logic of exactly that which Saussure thought
unimportant and accidental to language.
For Kristeva, it is only where semiotics and semiology come
together that speech occurs, and these only come together within a
divided subject. She says,
"[P]ractice is taken as meaning the acceptance of a symbolic law
together with the transgression of that law for the purpose of
renovating it. . . . [W]e can speak of practice wherever there is
a transgression of systematicity, i.e., a transgression of the
unity proper to the transcendental ego. The subject of the
practice cannot be the transcendental subject, who lacks the
shift, the split in logical unity brought about by language which
separates out, within the signifying body, the symbolic order
from the workings of the libido (this last revealing itself by
the semiotic disposition) (Read, p.29)."
However, we are not quite to equate semiotics with the working of the
drive itself, for this is the domain of psychology. Rather, semiotics
studies the processes of the drive as it is specifically shaped by
linguistic practice. For Kristeva, semiotics is "talking about
something other than language - a practice for which any particular
language is the margin (Des., p.25)."
Semiotics thus turns towards each network of signs, not insofar
as it forms a system, but rather insofar as it creates and constrains
a practice. Kristeva says,
"Semiotics must not be allowed to be a mere application to
signifying practices of the linguistic model - or any other
model, for that matter. Its raison d'etre, if it is to have one,
must consist in its identifying the systematic constraints within
each signifying practice (Read, p.26)."
We should note here that this rejection of semiology for the sake of
semiotics, is nearly identical to the rejection of semantics for the
sake of pragmatics which we shall discuss below. What is crucial to
Kristeva's project is that she recognizes the heterogeneous nature of
actual speech, which the formal structure of langue can not capture.
She does not go so far as to out-and-out reject langue, as we shall,
but she certainly does point towards the limits of its explanatory
power. Kristeva sees that there is something deeply individual about
speech, which may nonetheless be subject to systematic constraint.
This systematic constraint is a very different sort of affair from the
closed rules of Saussure's structural linguistics. Where the latter
can decide every sentence as grammatical or ungrammatical, and as
having some given meaning, the former may only exercise pressure, not
legislate. The former is langue, the latter ideology.
*2.0 Sentence-meaning and utterance-meaning.*
We shall now turn to an issue which stands to analytic philosophy
as the langue/parole division does to continental philosophy. This is
the possibility of a semantic theory of language. As above, we shall
largely reject such a theory in favor of pragmatic explanations of
language use. However, in analytic philosophy there is no figure
analogous to Volosinov, in his thoroughgoing rejection of the abstract
system of langue. Rather, we are faced with thinkers who try, at
most, to limit the area in which semantics can explain the meaning or
force of an utterance; that is, who limit the element of
sentence-meaning in a particular utterance-meaning. In analytic
philosophy no one really thinks that utterance-meaning contains no
sentence-meaning, nor does anyone believe that sentence-meaning can be
the whole of utterance-meaning. Rather, analytic philosophers tend to
take stands along a continuum of the supposed degree of
sentence-meaning in utterance-meaning, or perhaps on a continuum of
the occasions where sentence-meaning is sufficient.
Nonetheless, it is possible to separate these philosophers into
camps over whether they believe that a semantic theory can be expanded
indefinitely to encompass more and more phenomena. Some believe that
this is the case; others that,
"[N]onlinguistic beliefs, intentions of the speaker, and other
factors enter into the interpretation of utterances in so
intimate - and perhaps so fluctuating and indefinite - a fashion
that it is hopeless and misguided to attempt to represent
independently the 'purely grammatical' component of meaning (SGG,
p.67)."
The author of this above quote is Noam Chomsky, whom I shall take for
a fairly typical proponent of the first camp; namely, those who think
that a semantic explanation is adequate to explain the meanings of
speech. However, Chomsky more than most sees his position as an
entirely empirical one. We see this, for example, in his
acknowledgement that it might yet turn out that the assertion which he
makes above is true. We choose Chomsky for explication because, in
addition, he has does some of the most concrete and incontrovertible
work in syntax. Although we shall eventually disagree with the status
granted to syntacticality, we shall agree that humans must have the
sort of syntax generator proposed by Chomsky.
In response to the camp which sees semantics as largely adequate
to interpreting utterances we shall discuss three authors. The first
is Grice, who suggests that an understanding of intentions is
necessary for understanding the force and meaning of an utterance.
The second is Austin, who shows a whole class of utterances in which
the semantic "meaning" of an utterance, if such exists at all, cannot
suffice to explain the force of the actual issuance. Lastly we shall
discuss a recent article by Davidson which directly denies that there
is anything which a semantic-theory can be a theory of. In the last
author it is worth noting that the author is primarily reacting to his
work. We find it curious that, within analytic philosophy, the most
powerful opponents of semantic theories started out as the most
powerful proponents of the same.
*2.1 Semantic theories.*
In general, we might divide semantic theories into two sorts:
intensional and extensional. This difference corresponds, roughly, to
Frege's distinction between _sense_ and _reference_. Intensional
semantics tries to explain meaning in terms of the relations of
concepts to one another. Any theory of truth which is likely to be
attached to an intensional theory of meaning shall be in terms of the
logical relations of the terms and predicates. This theory will
include the truths of first-order logic, as well as a few others. For
example, two predicates, P^1^ and Q^1^ might be conceptually linked in
a manner such that (A)x[P^1^(x) implies Q^1^(x)]. Then if an
expression ({?} implies P^1^) is true (i.e. a logical truth), ({?}
implies Q^1^) is true. This ability to speak of conceptual links
commits the intensional semanticist to an essentialist ontology, if
the words/concepts are to correspond to the world in his theory. That
is, if as above, two predicates P^1^ and Q^1^ have some logical
connection and these two predicates are true of some particular
objects in the world, then this must be because the objects picked out
by each predicate have some common essence or property.
Extensional semantics tries explain meaning in terms of
reference. Each term stands for some set of individuals in the world;
and each predicate is true of some set of individuals. This is the
traditional nominalist position, and hence the extensional semanticist
is not ontologically committed to properties or essences. The
extensional theory of truth is in many ways similar to the intensional
one. It also starts with the tautologies of first-order logic.
However, conceptual connections are ruled out. In their stead, we may
have predicates (or terms) related in terms of their extensions in the
world. For example, it may be that {?}x[x{?} {y| P^1^(y)} implies
x{?} {y| Q^1^(y)}]. _This ___use of the set notation does not commit
our ontology to anything besides individuals; in particular, we need
not believe that such things as sets exist. Quine points out in _Set
Theory and Its Logic_ that we may "simplify" an expression of the sort
'x{?} {y| P^1^(y)}' to the expression 'P^1^(x)'. Our use of the
set-theoretic notation may be thought of as only a fanciful way of
talking about the predicates themselves. Of course, this is not to
say that _all_ use of set-theoretic notation is reducible to talk
about predicates, but only the piece that concerns us here. Using the
"simplification" suggested by Quine, we may reduce the extensional
statement above to the form, '{?}x[P^1^(x) implies Q^1^(x)]'. One
immediately notices that this statement is identical in form to the
intensionalist statement of the previous paragraph. However, the
extensionalist's statement is not a statement about the concepts
linked to P^1^ and Q^1^, but rather a statement about how words map to
objects in the world.
Chomsky, who we mentioned above, is a proponent of intensional
semantic explanation. This sort of explanation, roughly, originates
for analytic philosophy with Russell, and perhaps Frege. Russell's
theory of definite descriptions, for example, is an intensional
explanation of the meanings of names. Extensional explanation of
semantic meanings owes much of its origin to Tarski's redundancy
theory of truth. David Kaplan, Quine, Kripke, Davidson, and
verificationists are some examples of theorists of extensional
meaning. Kaplan and Kripke wish to spell this out in terms of
possible worlds; Quine, and to an extent, Davidson, wishes for an
extensional theory of meaning because his behaviorism rules out
intensional explanation. We shall start the explication of semantic
theories in analytic philosophy with a rough sketch of extensional
theories. Next, we shall discuss in more detail Chomsky's intensional
approach to semantics. No effort shall be made to explicate other
intensional semantic theories than Chomsky's, as they shall all share
the aspects which we shall criticize. Before we discuss these
theories we should make the caveat that extensional and intensional
theories are not mutually exclusive. Each could be used to explain
some particular class of expressions, but only between the two would
all expressions be explained.
*2.1.1 Extensional theories.*
Extensional theories of meaning are, roughly and as a rule,
theories of truth. Some retreat has been made from this in the
proposal to shift the concern from truth to warranted assertability.
This might be, for example, a move which various of the verification
oriented logical positivists wish to make. However, this move does
not really affect extensional theories in a way which shall interest
us. In general, extensional theories of meaning explain meanings by
states of the world outside of the mind of the speaker of an
utterance. This is so both of the "meanings" of singular terms and of
the "meanings" of propositions. In the case of singular terms, our
concern in extensional theories is generally with reference. Hence we
encounter such terms-of-art as Donnellan's "reference by false
description", Kripke's "rigid designators", and Kaplan's "direct
reference"; all of which are used to point out that often our
intentions are irrelevant to the reference of our words. When we turn
to propositions, the extensional theories direct our attention to the
circumstances in which a sentence, or utterance, is true; or perhaps
to the ways in which it may be verified. For Quine, the meaning of a
sentence might be the collection of circumstances in which a native
speaker would utter it.
All of these cases have the general pattern of explaining
meaning, in the vocabulary of set theory, as an injection of sentences
into sets of things other than sentences. These "_things_" other than
sentences are states-of-the-world, possible worlds, or perhaps events
of the right sorts. Let us examine each of these, briefly.
*2.1.1.1 Possible-World semantics*
_Possible world semantics_ is an extensional semantic theory
which maps sentences into the worlds in which they are true. This is
a change in the traditional nominalist positions in extensional
semantics, in that the individual of concern are not the ordinary
objects like tables and people, but are rather the possible worlds in
which these things may be. David Kaplan, in several essays[8],
develops a detailed set-theoretic model of what this amounts to.
Kripke, more influentially, uses this framework in an intuitive way to
make various arguments within analytic philosophy of language. The
general idea is that given any sentence, or perhaps a sentence
together with a situation of utterance, there corresponds a set of
possible worlds in which the sentence is true. The "meaning" of the
sentence is, then, none other than this collection of worlds. The
terms and predicates, as in the normal case, pick out sets of objects;
however, these sets indexed for each possible world. That is, if we
let {p/n/}= P be the set of possible worlds; and {o/m/}= O be the set
of collections of objects; then the set of "objects" picked out by a
term is actually a complete function from P into O containing pairs of
the form , for p/n/ {?} P and o/m/ {?} O. Whether all the
objects which are the elements of elements of O are real, or whether
some of them are possible objects, is a version of the problem of
haecceitism[9]. Either way, a sentence determines a set of possible
worlds as follows. The subject(s) (which are terms) in a sentence
pick out, in each possible world, some particular collection of
objects o/m/. Likewise, the predicates pick out, in each possible
world, some collection o/k/. The set of worlds picked out by a
sentence is the set of worlds in which, for the appropriate o/m/,
o/k/, o/k/ is a subset of o/m/.
This can be extended, fairly easily, to include non-indicative
sentences. For example, the "meaning" of an imperative sentence can
be equated with the set of possible worlds in which the command is
correctly carried out, together with an imperative flag. If we accept
Quine's assertion that all occasional sentences can be replaced by
equivalent standing sentences, then we can do so as a first stage
transformation on every actual utterance in order to eliminate
utterance situation from the domain of the "meaning" injection.
*2.1.1.2 States-of-the-world semantics***
_"States-of-the-world"_ semantics is similar to possible world
semantics, but perhaps more restrained in its ontology. The "meaning"
of an indicative sentence is seen to be exactly the state-of-the-world
it purports. For example, the stock example, 'snow is white' is
evaluated as follows. First we determine the reference of the term
'snow'; namely, snow. Then we evaluate the predicate by the
appropriate property. We might manage to do this without making our
ontology too replete by defining a property as the "right" kind of
resemblance to a paradigm case. In the nominalist explanation, we
dispense with properties altogether and speak of sets of things which
fulfill a predicate. At any rate, the "meaning" of 'snow is white' is
cast as the stuff, snow, having the property (or fulfilling the
predicate) of "being white". In Fregean terms, we may say the
unfilled predicate together with the term make a true proposition, or
rather the proposition "True". The idea here is that the set of
things referred to by the term is a subset of the set of things
fulfilling the predicate. The reference of every sentence as a whole
is one of the two propositional values, True and False, but each
sentence purports a subset relation between different collections of
objects. Here there is no question of haecceitism, as there is only
the real world for objects to exist in.
*2.1.1.3 Verification semantics.*
The events which might found an extensional semantics are
verifications. The meaning of an indicative sentence on this account
is considered to be the claim that certain results have followed, or
would follow, the "right" sort of experimental procedure. This is not
merely a translation from a sentence to an equivalent, which might be
part of an intensional semantics. Rather, we cast the verificationist
semantics in a fully extensional way. Namely, we are given a partial
function from pairs of events into sentences. This is a slight
modification of our suggested injection of sentences into other
_things_, in two ways. Since we have reversed the injection, it is
now possible for sentences to correspond to multiple pairs of events.
Also, since we have only set up a partial function, not every pair of
events need be considered to explain the meaning of some sentence. It
should be noted, in addition, that our function is into, not onto;
hence the verificationist does not claim to necessarily give a meaning
of every sentence. We may weaken this mapping slightly more to a
partial relation between pairs of events and sentences. This would
allow for sentences to be true synonyms. In plainer language, what is
being done is cashing the "meaning" of (at least some) indicative
sentences as one or more pairs of events with consequent events. The
former events are thought of as the setting up of experimental
conditions, the latter as the expected, or anyway purported, results.
For example, one pair of events which may be considered the "meaning"
(or part of it) of 'snow is white' might be: 1) Taking a snow sample
and putting it in a properly working spectrometer; 2) The spectrometer
reading equal amounts of all bands of visual light.
*2.1.2 Intensional theories: Chomsky.*
Having sketched a few possible positions in extensional
semantics, we shall turn to intensional semantics; in particular, that
of Noam Chomsky. The basic notion with which intensional semantics
concerns itself is that of synonymy. Further ideas which we might
hope to explain in an intensional semantics are entailment,
presupposition, relations of degree, and categorical agreement. All
of these are dealt with by Chomsky, particularly in his concern for
the relation between syntax and semantics. Before showing Chomsky's
specific treatment of some issues, however, we should make a thumbnail
sketch of his lines of thought.
*2.1.2.1 Recursive syntax.*
Chomsky takes it as axiomatic that human language, by virtue of
the fact that it can be learned, must be generated in a recursive
fashion. The sentences in a language are presumed infinite in number,
and the ability of humans to store information is presumed finite;
hence the motivation for recursive rules. This recursive generation
of natural language will be computationally equivalent to the abstract
systems of Turing, Church, Kleene, et al. Given this axiom, Chomsky
goes on to give some specific proposals about the sort of system which
is involved in this recursive generation. Chomsky's system sees
semantics as being consequent to syntax, or perhaps independent of it
- but certainly he does not believe syntax to depend on semantics. He
says as much,
"Investigation of proposals [that grammar, i.e. syntax, relies on
semantic notions], however, invariably seems to lead to the
conclusion that only a purely formal basis can provide a firm and
productive foundation for the construction of grammatical theory
(SS, p.95)."
This formal basis is the sorts of recursive procedures noted above.
We may then assume that, for Chomsky, semantics is, so to speak,
epiphenomenal to the generation of syntactic sentences, or other
strings. The alternative - namely, that semantics is wholly dependent
on utterance-context - is left empirically open, but not pursued.
The actual theory of syntax-based semantics, proposed by Chomsky,
has undergone numerous modifications. The latest version which I
shall try to explicate is that of the mid-1970's; though no essential
revisions have followed this. We may think of a "Chomsky-machine",
which should include persons as instantiations, as a syntactic machine
to generate ordered quadruples. These quadruples consist in: a
phonetic representation; a surface structure; a deep structure; and a
semantic representation. We may call these elements, 'P', 's', 'd',
and 'S' respectively. The Chomsky machine may be realized
algorithmically in an indefinite number of ways. Hence, although the
modelling Chomsky uses does not have the semantic representation, S,
as its computational starting point, some equivalent procedure might.
This might best correspond with our intuition that meaning should
precede grammar in the mind of a speaker.
*2.1.2.2 Syntactic structures.*
Chomsky's modelling proceeds through a series of syntactic
structures Z= (P/1/, ..., P/i/, ..., P/n/), where the following are
true of the sequence: P/i/ is the deep structure d; P/n/ is the
surface structure s; for some h< i, lexical insertion occurs between
the structures P/h/ and P/h+1/. In the earlier Chomsky it was thought
that the semantic structure S could be determined solely by the deep
structure P/i/= d. However, later Chomsky was led to believe that S
must depend on some transformation(s) P/j/, for i< j< n+1. The
particular concern which motivated this is some differences in the
presuppositions of active and passive sentences[10]; where these have
been presumed to differ only in surface structure. Following the
computation of the sequence Z, a transformation is made on P/n/= s to
produce the phonetic representation P. This, in a general way,
explains the generation of the quadruples in question. Let us now
turn to how this modelling can account for the semantic notions we
mentioned above.
*2.1.2.3 Synonymy and paraphrase.*
The first of the notions to be explained is synonymy, and under
this heading we should also be concerned with paraphrase. Chomsky
originally thought that synonyms would be exactly those sentences, or
other strings, which shared the same deep structure. This was later
modified due to the difference in presupposition of some active and
passive sentences, which are still presumed to share the same deep
structure. Nonetheless, we may still recognize that identity of deep
structure makes two sentences close paraphrases of each other. Of
course, strict synonymy is a matter of identity of semantic
representations; but since no concrete proposals have been properly
elaborated for the system of semantic representation, this is a fact
with little bite. The bite of Chomsky's own explanation comes from
the relatively complete proposals for generation of the sequence Z
which have been made. For the being, we may hence assume that
identity of deep structure provides us with an adequate notion of
synonymy.
Prior, in the sequence Z, to the determination of deep structure,
lexical insertion was performed. It is at this stage that we expect
to find the most of interest to semantic notions, as it is somehow the
lexicon which seems to determine meaning, over and above grammar.
Chomsky assumes here that our stock of lexical items is innate. He
says this,
"First, it is important to determine the universal,
language-independent constraints on semantic features - in
traditional terms, the system of possible concepts. The very
notion of 'lexical entry' presupposes some sort of fixed,
universal vocabulary in terms of which these objects are
characterized . . . It is surely our ignorance of the relevant
psychological and physiological facts that makes possible the
widely-held belief that there is little or no _a priori_
structure to the system of 'attainable concepts' (ATS, p.161)."
It is still allowable that some lexical entries may be compound forms,
but the elements of which they are formed must nonetheless be innate.
Given this innatist bias, Chomsky is freed from the need to explain
the properties of the lexical entries themselves, except where these
entries are compound forms.
Let us look at several manners in which particular synonyms and
paraphrases may be explained by Chomsky's modelling. In the case of
the four sentences,
(1) The vehicle is green;
(2) The car is green;
(3) The automobile is green;
(4) The truck is green;
we can pick out several different degrees of synonymy and paraphrase.
Strict synonymy seems to hold between (2) and (3). We may then assume
that the phonetic representations 'car' and 'automobile' are in free
variation. If so, the deep structures of (2) and (3) are identical,
and no transformation is made on S after the stage P/i/ in Z.
The relations between (1) and (2), (1) and (3), and (1) and (4)
seem to be of the same sort. That is, (2), (3), and (4) are more
specific in the information conveyed than is (1). We might plausibly
explain this by assuming that the lexical items 'car', 'automobile',
and 'truck' are compound lexical entries containing the primitive
semantic notion, [vehicle]. Following the level P/h+1/ of lexical
insertion, we may postulate that the presence of a compound lexical
entry forces certain sorts of transformations. For example, in the
structure which produces (1) we may assume P/h/ is of the form,
'Article-Noun-Copula-Adjective'.
Lexical insertion gives us the form,
'The-Vehicle-Is-Green'.
This is essentially the form of the surface structure also, and merely
needs be transformed into a phonetic representation. However, the
formation of (2), (3) or (4) is more complicated. Given the same
syntactic form of P/h/, we may postulate that the lexical insertion of
the form,
'The-Vehicle, for Passengers, Four-wheeled-Is-Green'
occurs. This will, perhaps, then undergo another transformation
resulting in,
[The-Vehicle-Is-Green-[It(same thing)-Is-for Passengers, It(same
thing)-Is-Four-Wheeled]]
as the semantic representation. For (4), similarly, we might have the
semantic representation,
[The-Vehicle-Is-Green-[It(same thing)-Is-for Materials, It(same
thing)-Is-Four-Wheeled]].
These semantic representations are not actually proposed by Chomsky;
they are our own. However, we believe these semantic representations
to be enough in the spirit of Chomsky to demonstrate his approach to
semantic relations. Assuming these semantic representations, we can
see that the relation between (1) and (2), (1) and (3), and (1) and
(4) are as hoped for: (2), (3), and (4) each contain the assertion of
the content of (1), plus some additional information.
Similarly, we get the intuitively correct semantic connection
between (2) and (4), or (3) and (4). There is the same basic type of
assertion in each case, but with difference in the details of the
compositional subject. The intuition we have that the words 'car' and
'truck' have a semantic connection is born out by the common primitive
concepts [vehicle] and [four-Wheeled].
*2.1.1.4 Entailment.*
Similar sense can be made of the other semantic notions we have
mentioned above. For example, relations of degree and entailment can
both be easily accounted for by the idea of compositional lexemes. We
probably feel inclined to say that (5i) entails (6).
(5) (i) John murdered the boy.
(6) John killed the boy.
This can be explained by the assumption that the lexeme [murder] is
compositional. Perhaps it is composed of the lexemes
[act-intentionally] and [kill]. This might be in conjunction with the
fact that [kill] is itself composed of [cause-to-die] and
[play-active-role]. Hence, the p/h+1/ of (5i) might be,
'John-Cause-to-die, Play-active-role, Act-intentionally--The-Boy'.
After the same sort of transformation suggested above we get the
semantic representation,
(5) (ii) [John-Cause-to-die--The-Boy-[John-Play-active-ro
le- [John-Act-intentionally]]].
Notice that we have a strict ordering of the compositional elements of
the lexeme [murder]. Hence we may add (7),
(7) John caused the boy to die;
and have the entailment of (6), (7) by (5i), and (7) by (6) - but not
(5i) by (7), (5i) by (6), or (6) by (7).
*2.1.1.5 Relations of degree.*
Relations of degree are also explained by compositional lexemes.
Consider,
(8) Jane was happy; and
(9) Jane was ecstatic.
We feel that (9) is a stronger form of (8). If we assume that the
P/h+1/ of (9) is,
'Jane-Is--Happy, Extreme-degree',
we might have the semantic representation,
[Jane-Is--Happy-[Jane(does same in)-Extreme-degree]].
This expresses the relation of degree, as well as the entailment from
(9) to (8). In general, we might see the marker, [X(does same
in)-Extreme-degree], wherever a relation of degree exists between
sentences. Perhaps this marker could contain some index to the actual
degree. Since as far as we know every natural language contains only
a finite number of degrees of any attribute, this might well be
plausible.
*2.1.1.6 Presupposition.*
Let us now turn to how the modelling proposed by Chomsky can
handle presupposition. Much of Chomsky's explanation of
presupposition relies on focus, which he assumes to exist only in the
surface structure. For example, consider the sentence,
(10) It was an ex-convict with a red SHIRT that he was warned to look
out for.
Under normal intonation, the emphasis of the sentence falls on
'shirt'. If so, the focus may taken to be any of the following:
(11) (i) shirt;
(ii) a red shirt;
(iii) with a red shirt;
(iv) an ex-convict with a red shirt.
The sentence (10) presupposes, in any case, that something fulfills
the sentence by taking the position of the focus. We may see this by
examining natural denials of (10). Corresponding to each of the
possible foci in (11) we might deny (10) by saying,
(12) (i) No, he was warned to look out for an ex-convict with a red
TIE.
(ii) No, he was warned to look out for an ex-convict with a
CARNATION.
(iii) No, he was warned to look out for an ex-convict wearing
DUNGAREES.
(iv) No, he was warned to look out for an AUTOMOBILE salesman.
In each case, the denial, like the assertion, (10), holds for some
value in the place of the focus. We may see clearly that it is really
the focus playing this role by looking at natural shortenings of (12).
These would be,
(13) (i) No, a red TIE.
(ii) No, with a CARNATION.
(iii) No, an ex-convict wearing DUNGAREES.
(iv) No, for an AUTOMOBILE salesman.
In each case, what is given in the denials, (13), is only the segment
prior to the assumed focus and a new choice for the focus. The fact
that these short forms of denials are easily and naturally understood
suggests that we assume that denial must only be of the proper object
or word for the focus of an assertion. The reiteration of the segment
prior to the new phrase merely serves to demonstrate the assumed
focus.
We may observe a similar sort of explanation of presuppositions
of sentences which do not have the sort of nested surface structure of
(10)[11]. Sentences which contain a nesting only in the semantic
representation have similar presuppositions. Recall our analysis for
the semantic representation of (5i); namely, (5ii). If we continue to
assume this is a correct analysis, we can easily explain the
presuppositions of (5i). Using the same method as above, let us
examine natural denials of (5i). These are,
(14) (i) No, John killed the boy, unintentionally.
(ii) No, John caused the boy to die, through inaction.
(iii) No, John did not cause the boy to die.
(iv) No such thing happened (i.e. the boy is alive).
The first three of these deny a particular element of the
compositional lexeme [murder]. The elements of the lexeme must be
denied in a specific order, so that denial of a less nested element
entails denial of a more nested one. Hence, any denial of (5i)
specifies a particular level of nesting at which the denial occurs,
while presupposing the other elements of the lexeme, which are less
nested, to hold. We should note that (14iv) is something of an
anomaly here. It denies a presupposition of (5i) which is not
expressed in our semantic representation (5ii). We shall not explore
how to handle this presupposition, except to suggest that the lexeme
[cause-to-die] might not, itself, be primitive. Rather, insertion of
this lexeme may cause the semantic representation to take the form,
'(5ii)'-[The-Boy-Is-Dead].
*2.1.1.7 Category agreement.*
We shall now show how Chomsky's modelling deals with categorical
agreement. Chomsky points out that the examples such as (15) are
anomalous compared with (16).
(15) (i) the boy may frighten sincerity;
(ii) sincerity may admire the boy;
(iii) John amazed the injustice of that decision;
(iv) the boy elapsed;
(v) the boy was abundant;
(vi) the harvest was clever to agree;
(vii) John is owning a house.
(16) (i) sincerity may frighten the boy;
(ii) the boy may admire sincerity;
(iii) the injustice of that decision amazed John;
(iv) a week elapsed;
(v) the harvest was abundant;
(vi) the boy was clever to agree;
(vii) John owns a house.[12]
However, the anomalies of (15) are not the same as purely syntactic
anomalies such as (17).
(17) (i) sincerity frighten may boy the;
(ii) boy the frighten may sincerity;
(iii) injustice the of that John amazed decision;
(iv) elapsed week a.
Chomsky suggests that this difference may be explained by
distinguishing between the syntactic anomalies of the sort in (17) and
_semantic_ anomalies of the sort in (15). This difference may be
spelled out as follows. The sentences (or strings, rather) in (17)
are already ruled out before the stage P/h+1/ of lexical insertion.
For example, the P/h/ of (17i) would be of the form,
'Noun-Verb-Auxiliary Verb-Noun-Article',
which cannot be generated by any Z for English. However, the
anomalies of (15) are presumed to occur only subsequent to the stage
of lexical insertion. So while the P/h/ of (15i),
'Article-Noun-Auxiliary Verb-Verb-Noun',
is perfectly allowable, something is wrong with the particular choice
of lexical insertion.
We shall find that the particular problem that arises subsequent
to the lexical insertion in the sequence for the generation of the
examples in (15) is lack of category agreement. Chomsky lists some of
the categories with which we shall be concerned with,
"The N _boy_ is a Count Noun (as distinct from the Mass Noun
_butter_ and the Abstract Noun _sincerity_) and a Common Noun (as
distinct from the Proper Noun _John_ and the Pronoun _it_); it
is, furthermore, an Animate Noun (as distinct from _book_) and a
Human Noun (as distinct from _bee_); _frighten_ is a Transitive
Verb (as distinct from _occur_), and one that does not freely
permit Object deletion (as distinct from _read_, _eat_); it takes
Progressive Aspect freely (as distinct from _know, own_); it
allows Abstract Subjects (as distinct from _eat, admire_) and
Human Objects (as distinct from _read, wear_) (ATS, p.75)."
Probably, the easiest way to handle category agreement is to assume
that the categories mentioned by Chomsky, and perhaps a few others,
are primitive semantic items which are attached to every lexeme.
These categories, like other compositional elements of lexemes, are
attached in a particular order. Hence it might be that every noun is
either a Count Noun, Mass Noun, or Abstract Noun; amongst Count Nouns
there are Animate Nouns and Inanimate Nouns; amongst Animate Nouns the
feature Human Noun is either present of absent. For example, the
lexeme [boy] might have the compositional form [male,
child-[Count-[Animate-[Human]]]]. The lexeme frighten might have the
compositional form, [frighten-[Animate-object]]. The lexeme
[sincerity] might be expanded to [sincerity-[Abstract]]. After
compositional lexemes are inserted into a syntactic structure, we, as
above, make certain transformations to arrive at the semantic
representation. Hence if we assume the P/h/ form of (15i) to be,
(18) 'Article-Noun-Auxiliary Verb-Verb-Noun';
this will give us the P/h+1/ form, after lexical insertion, of,
(19) 'The-Male, Child, [Count [Animate ...]]-May-Frighten,
[Animate-object]-Sincerity, [Abstract]'.
Performing the same type of transformation as above we might get a
semantic representation of,
(20) [The-Male-May-Frighten-Sincerity-[It(subject)-Is-Child-[It(subjec
t)-Is-Count-[It(subject)-Is-Animate ...]]],
[It(object)-Is-Animate], [It(object)-Is-Abstract]].
This might not be entirely clear, at first glance. We shall make a
few remarks on the procedure used to produce it. The least nested
portion of the semantic representation is merely the main
compositional element of each lexeme in (19). The rest of the
semantic representation is nested one level. This consists of the
secondary elements of each lexeme, arranged in parallel fashion. That
is, everything separated by commas within the same square brackets
should be considered independent of order. We have arranged these
secondary elements of each lexeme in the order in which the lexemes
occur in the un-nested portion of the semantic representation. Within
each secondary element of a particular lexeme, there may occur items
with a greater degree of nesting. The anomaly of (15i) arises from
the fact that both (21i) and (21ii) occur within the same square
brackets in (20).
(21) (i) [It(object)-Is-Animate];
(ii) [It(object)-Is-Abstract].
(21i) and (21ii) arise as secondary elements of different lexemes, but
are nonetheless present on the same level within the semantic
representation (20). If a similar semantic representation for (16i)
were worked out we would start with the same P/h/ form (18); however,
the semantic representation would not contain contradictory elements
(21i) and (21ii). Rather the semantic representation of (16i) would
merely contain the feature (21i), repeated twice on the same level
(or, if you like, a redundancy elimination transformation might occur
prior to the semantic representation S).
We have not explained every sort of semantic analysis which can
be performed with Chomsky's modelling. All we hope to have done is to
provide a feel for how Chomsky can deal with some semantic
explanation. This should be sufficient to understand our eventual
rejection of semantic programs and the weakening of them performed by
the authors to be discussed below.
*2.2 Grice.*
Among the several notions that Paul Grice raises in his
examination, in numerous essays, of communication and conversation is
one that shall particularly interest us. This notion is
_implicature_. Implicature is unlike the classical logical notion of
implication. Where we are said to _imply_ something if it follows as
a necessary consequence of something we have said, we _implicate_
something which was already assumed before we started talking. We may
not be aware of what is implied by what we say, but we are almost
certainly already confident that what we implicate is so. Implicature
may be thought of as the assumptions or beliefs that must exist for a
given utterance to really make sense.
An example of implicature would help here. Grice gives us this
answer to an inquiry about some person, C: "Oh [he is] quite well, I
think; he likes his colleagues, and he hasn't been to prison yet
(LC)." Such an answer is most peculiar if we do not have some
relevant piece of background knowledge about C; or if we do not, at
least, know something about the speaker's beliefs about some relevant
topic. The answer would be less peculiar if we know, for example,
that C's colleagues are 'really very unpleasant and treacherous
people'; or that, at any rate, the speaker believes this. The
question arises of what facts or assumptions get implicated, and in
what circumstances. Grice gives us a fairly general account of this.
This account is in Grice's _conversational maxims_.
*2.2.1 Conversational maxims.*
Grice's conversational maxims are "rules of conversation" which
are both prescriptive and descriptive. They are prescriptive in the
sense that we are expected to obey them, and generally subject to
censure if we violate them without adequate reason. The maxims are
descriptive in the plain sense that people do obey them most of the
time, at least according to Grice. We shall not individually examine
all the maxims, but we may as well present them all. The maxims are
of four sorts, each divided into sub-maxims. They are as follows.
Quantity,
(1) Make your contribution as informative as is required (for the
current purposes of the exchange).
(2) Do not make your contribution more informative than is required.
Quality - 'Try to make your contribution one that is true.'
(1) Do not say what you believe to be false.
(2) Do not say that for which you lack adequate evidence.
Relation - 'Be relevant.'
Manner - 'Be perspicuous.'
(1) Avoid obscurity of expression.
(2) Avoid ambiguity.
(3) Be brief (avoid unnecessary prolixity).
(4) Be orderly. (LC)
Grice does not claim that these are necessarily exhaustive.
*2.2.2 Intentions.*
Our purpose in presenting these maxims is to show how
attributions of intentions figure into our interpretation of the force
and meaning of utterances. The particular intentions which we often
attribute are those in the conversational maxims; we generally assume
that a speaker intends to fulfill the maxims. For example, the
following pair is unexceptionable. Question: 'My car is out of gas,
where may I get some?' Answer: 'There is a gas station around the
corner.' The answerer may be said to implicate a number of things:
that the gas station is open; that it has gas; etc. If, however,
these are not true and this is known to the answerer we should say
that she is, if not lying, being deceptive or uncooperative. This is
because the answerer is presumed to be obeying the conversational
maxims. She has not, let us say, told a literal untruth, but she may
be violating some other maxims than those of quality. Chiefly, in
this case, the answerer is violating the maxim of relation. What she
has said is in no way relevant to the conversation at hand if she
knows the gas station to be closed.
It is only understandable that the utterance 'there is a gas
station around the corner' may be taken as meaning 'you may get gas at
the gas station around the corner' if we assume the answerer to intend
to obey the conversational maxims. The hearer of the utterance must
go through a process, however unconscious, of figuring out what
relevance this utterance has to the conversation, and what
conversational goals, intentions and beliefs must exist in the
speaker. The same process is also involved where the conversational
maxims are not explicitly involved. Other maxims involve aesthetic,
social, or moral restraints on our speech. For example, the maxim 'be
polite' may lead to implicature in much the same manner as do the
conversational maxims we listed. It is clear that we reconstruct
intentions on all levels, and not merely those of broad rules of
conversation. When someone slurs a word, or substitutes the "wrong"
word, or uses ambiguous or unclear syntax, we make hypotheses about
what it is they are intending to say. Some of this will come out in
our discussion of both Austin and Davidson, below.
*2.3 Austin.*
Let us examine how J. L. Austin can be seen as speaking against a
semantic, or at least a purely semantic, theory of language. The
field of philosophy called speech-act theory, of which Austin is
perhaps the progenitor, has as its main goal the explanation of that
in language and speech which cannot be captured under the rubric of a
semantic theory. That is: the meaning or force of speech which cannot
be explained by sentence-meaning. The speech-act theorists believe
this portion of the meaning of speech to be a major, if not
overwhelming, part. Austin leads us to the meaning in speech which
cannot be explained through sentence-meaning, in his _How to Do Things
with Words_, through a discussion of _explicit performatives_[13].
Austin gives some examples of these explicit performatives.
"I do (sc. take this woman to be my lawful wedded wife)' - as
uttered in the course of the marriage ceremony.
'I name this ship the _Queen Elizabeth_' - as uttered when
smashing the bottle against the stem.
'I give and bequeath my watch to my brother' - as occurring in a
will.
'I bet you sixpence it will rain tomorrow.' (p.5)"
What Austin notices about these examples, and what we should notice,
is at least twofold. The first point is that these performatives are
not assertions about facts, despite their superficial structure. We
shall go into this point below. The second issue to notice is that
the force[14], and indeed the meaning, of these utterances depends a
great deal upon the situation in which they are uttered. This is not
the simple form of context dependence which exists, for example, with
deictic elements. Rather, it is a situational dependence which
involves intricate social conventions, intentions of speakers, and the
interpretation of listeners. This situational dependence leads us to
read Austin as demanding a properly pragmatic explanation of language.
*2.3.1 Infelicities.*
Austin points out that explicit performatives, as well as other
performatives (we shall see later how broad a class these others are),
are subject to what he calls _infelicities_. We shall not detail all
of these and Austin's descriptions and comments about these. Rather
we shall let it suffice to mention the three broad types of
infelicities. These are:
(1) there may not exist a conventional procedure to perform an
action, or the procedure may not be used properly;
(2) if a procedure exists it may not be carried out correctly or
completely;
(3) the action may not be accompanied by the appropriate
intentions or emotions, or may not be followed by expected
subsequent actions.
In all of these infelicities the normal action associated with the
utterance is not carried out, and in addition, the utterance does not
carry its normal force and meaning.
Austin points out that this is especially true of infelicities of
the first and second sort. We may illustrate these in one example to
clarify this. In the case mentioned above of the marriage ceremony,
infelicities of either of the first two kinds void the action, and
cause the utterance to carry a different force and meaning. An
example of the first type of infelicity here, might occur during an
effort to wed a chimpanzee; of the second kind during a ceremony
performed by an imposter of a judge. In neither case has the act of
marriage been performed; the act which was, in fact, performed depends
on a great many factors not given by our description. What is of
greater interest here is that what was, in fact, _said_ differs in
these infelicities from what was said in the normal case. Hence we
may normally _mean_, 'I do take this woman as my wife', when making
the same utterance; with the force of marrying her. However, in the
infelicitous cases this is not so. In our first case the meaning of
the actual utterance may be of the sort, 'I consider the institution
of marriage a mere mockery to start with'; and either have the force
of criticizing the institution of marriage or of persuading the
spectators of one's lunacy. In our second case, the meaning of the
actual utterance may be, 'I should like to be married to this woman';
and carry the force of expressing interest in this marriage.
One could question the since in which the utterance, 'I do take
this woman as my wife', could have the meanings ascribed above - or
indeed, any meaning other than 'I do take this woman as my wife.' The
sense in which this may happen is quite straightforward. If asked,
'what is the meaning/force of that utterance of yours', we often
answer with the meanings and forces described above (or these sorts).
Austin's program is largely to argue that there is utterance-meaning
which is not captured by sentence-meaning, so it should not surprise
us to see (in Austin) this notion of the literal meaning of utterances
depending on the context of utterance. Of course, everyone would
acknowledge that in some situations even literal meaning is context
dependent. Some obvious cases of this are homonyms and ambiguous
sentence structures. The speech-act theorist embraces a much wider
range of utterances which have context dependent literal meaning.
Infelicities of the third sort may also change the force and
meaning of an utterance, though perhaps not quite so broadly (or not
always so broadly). Hence when we utter, 'I warn you not to walk near
that bull', although it is penned and of no danger, we may be said to
lack an emotion or intention appropriate to the normal issuance of
this remark. The possible meaning and force which Austin would likely
attribute to this case are as follows: the meaning is the same as in
the normal case, namely 'I warn you . . .'; however, the force of the
utterance may be thought of as something like trying to cause
unnecessary fear in you. We may propose a reading of this utterance
which changes not only the force of the normal case, but the meaning
also. Hence perhaps, the meaning of the utterance may be, 'I make a
joke about the danger of the bull'; and its force the making of a
joke.
*2.3.2*
The examples and descriptions we have given are largely
anecdotal, as they shall have to remain. However, we are now in the
position to discuss how Austin's general approach to speech-act theory
detracts from, if not undermines, the semantic projects which we have
discussed above. For the case of extensional semantics, we have
already pointed out some limits in the extent of explanation. Namely,
performatives do not have truth values. Neither do performatives have
verifications, conditions of warranted assertability, or corresponding
possible worlds. In the last of these we gave a possible extensional
explanation of a particular non-constative type of utterance; namely,
commands or imperatives. We said that command could be explained by
reference to the set of possible worlds in which they are carried out.
While it may well be that commands are a particular sort of
performative, this possible worlds extensional explanation clearly
will not work for performatives in general. To take out well used
example of uttering, 'I do' during a marriage ceremony, we may say
that this utterance corresponds to the possible worlds in which we do,
in fact, take this woman as our wife; but a cursory examination of the
situation will reveal that these worlds are coextensive with the set
of worlds in which we say 'I do'. To be a bit more particular, we may
say that the former are coextensive with the worlds in which our
utterance of 'I do' is, in Austin's term, "happy". Either way, this
explanation is trivial, if not absurd.
However, the problems which speech-act theory presents to
extensional semantics run deeper that this. Extensional semantics
does not even stand on firm ground as regards constative utterances.
For Austin, constative utterances are only particular cases of
performatives. Or more conservatively, since we do distinguish the
two, both constative and performative utterances are kinds of
illocutionary acts - in neither case can we restrict our concern, in
evaluating either force or meaning, to the locutionary act alone[15].
Before we try to show the deep similarity between constative and
performative utterances, it would be well to discuss a few of the
varieties of performative utterances.
*2.3.3 Performatives: implicit and explicit.*
The first fact we should notice here is that performatives come
in at least two varieties: explicit and implicit. Not all
performatives have an explicit version, though presumably all have one
or several implicit versions. For example, although such an explicit
performative as, 'I promise I shall do X' has an implicit equivalent
(at least as uttered in some situations) of 'I shall do X'; such
implicit performatives as the insult 'You are scum!' have no explicit
version. Austin points our correctly that such an explicit variation
as, 'I insult you by calling you scum!', it has a meaning at all, has
only the meaning of a constative in the habitual present.
Implicit performatives have a sort of ambiguity not found in
explicit performatives. That is, implicit performatives may be
equivalent to any one of a number of explicit performatives. Hence,
when we say, 'That bull is dangerous', we may be warning, stating,
asserting, venturing, joking, promising, expressing condolences, or
any of a number of other things. Explicit performatives were not
subject to quite this kind of ambiguity. While explicit performatives
may be used with other meanings and forces than those normally
associated with them, these other meanings, and especially forces, are
not _conventionally_ carried by these utterances. In our above
outlandish, though by no means inconceivable, example of saying 'I do'
to a chimpanzee in order tho criticize marriage, the actual meaning
and force carried by the utterance are non-conventionally carried. We
should probably not make too much of this distinction, as it is not at
all clear what conventions exist and which do not. However, it does
at least seem true that the ambiguity described in implicit
performatives is an ambiguity between several explicit versions; also
some number of conventionally defined procedure which lack an explicit
performative to perform may enter in the ambiguity.
*2.3.4 Constative utterances.*
Austin points out in Lecture XI, that constative utterances
really share all the properties or performative ones, including the
explicit/implicit division. Austin summarizes his earlier discussion
of performatives.
(1) [T]he performative should be doing something as opposed to
just saying something; and
(2) the performative is happy or unhappy as opposed to true or
false. [i.e. it may either be free from infelicities or suffer
from them] (p.132)
Constative utterances also have just these properties ascribed to
performatives. To show this we need first point out that constatives
have the same implicit/explicit division which performatives were
shown to have. Some explicit versions of constative utterances are,
I state that P.
I assert that P.
I believe that P (in one sense).
and perhaps,
I argue that P.
I suggest that P.
I hold/find that P.
and others.
The first of these is generally taken as a paradigm case of
constative utterance. We need not argue that some or all of the other
examples are real constatives, as long as it is accepted that stating
(explicitly) is constative. Austin has us consider this
unexceptionable remark:
"In saying that it was raining, I was not betting or arguing or
warning: I was simply stating it as a fact (p.133)."
Austin continues with some similar examples.
"In saying it was leading to unemployment, I was not warning or
protesting: I was simply stating the facts (p.133)."
Or again, we are to compare
I state that he did not do it.
with
I argue that he did not do it,
I suggest that he did not do it,
I bet that he did not do it, etc.
All of these simple, and unexceptional, sentences put stating on
exactly the same level are the various illocutionary act with which it
is compared. In the latter comparison which we are asked to make, we
are to notice that any of the sentences can equally well make explicit
an utterance of, 'He did not do it.' We should not then suppose that
the implicit constative, 'He did not do it' is any less on par with
various performatives than the explicit with the explicit constative,
'I state that he did not do it.' We may see this be turning the issue
on its head somewhat. We evaluate the latter explicit constative, and
indeed such utterances as 'I think he did not do it', in just the same
manner as we do the implicit constative; that is, by checking its
truth, not by checking the action performed or the intentions of the
speaker. Conversely, the implicit constative is no less performative
than its explicit counterparts - whether these counterparts be
so-called constative, or performative in our clearest sense.
We further notice the performative nature of constatives through
an examination of infelicities which constatives suffer from. Having
shown that some conventional action is carried out by making
constative utterances (e.g. stating), if we can show that constatives
suffer from all of our infelicities then we shall have shown them to
be essentially a particular branch of performatives. The third sort
of infelicity most obviously applies to constatives. Just as when
promising we are expected to intend to perform the right subsequent
actions, when stating we are expected to believe what we state and to
intend to convey the right fact.
The first two sorts of infelicities, however, also apply to
constatives. Austin gives a number of examples of cases where there
is no convention to make a constative utterance, or where the
convention is not carried out properly. One of these is in the
problem raised by Strawson of presupposing. One sort of
presupposition we can make is existential; however, there is no
convention for imputing properties or traits to non-beings (or at
least not a general convention - we do talk of unicorns and rational
roots of 2, in some cases). We may explain the impropriety of saying
'The present King of France is bald' by pointing out that there is no
convention to impute traits to non-beings. Much more generally, we
are not always the right person, or in the right place, to state
anything. Austin says,
"Just as we often say, for example, 'You cannot order me', in the
sense 'You have not the right to order me', which is equivalent
to saying that your are not in the appropriate position to do so:
so often there are things you cannot state - have no right to
state - are not in a position to state. You _cannot_ now state
how many people there are in the next room; if you say 'There are
fifty people in the next room', I can only regard you as guessing
or conjecturing (p.137)."
Or again,
"Just as sometimes we cannot appoint but only confirm an
appointment already made, so sometimes we cannot state but only
confirm a statement already made (p.137)."
These point to cases where constative acts can and cannot be carried
out.
*2.3.5 Truth and falsity.*
One additional example will, we hope, suffice to show the breadth
of the conventionality of constatives. This example concerns, in some
way, the truth or falsity of constatives. Consider the statement,
'France is hexagonal.' Austin says of this, that one
"can see what you mean by saying that it is true for certain
intents and purposes. It is good enough for a top-ranking
general, perhaps, but not for a geographer. 'Naturally it is
pretty rough', we should say, 'and pretty good as a pretty rough
statement'. But then someone says: 'But is it true of is it
false? I don't mind whether it is rough or not; of course it's
rough, but it has to be true or false - it's a statement isn't
it?' How can one answer this question, whether it is true or
false that France is hexagonal? It is just rough, and that is the
right and final answer to the question (p.142)."
The point here is that the appropriateness of uttering, 'France is
hexagonal' is entirely conventional and dependent on the context.
This statement might be perfectly allowable in a grade school text,
but a bad joke among geographers. This sort of utterance, in fact
constatives in general, may suffer from infelicities of the second (or
first?) kind anytime that the utterance does not conform to the
guidelines for statements relevant to the situation of utterance. The
relevant aspects of the utterance situation may not, however, always
be obvious. Where we believe a certain sort of statement to be
conventionally allowable, we can be quite wrong. This goes a little
way toward explaining our earlier comment about the situational
dependence of utterances upon social conventions, intentions of
speakers, and interpretations of listeners.
*2.3.6 Conclusions.*
Having sketched Austin's program, let us make explicit how this
program speaks against semantic theories, both extensional and
intensional. Semantic theories, of either sort, insist that meanings
are attached to sentences. Austin, to the contrary, points out that
meanings are utterance specific, and dependent on the situation of
utterance. Since, as we have said, extensional theories are generally
theories of truth, Austin's program will be most poignant to these in
its dealings with the notion of truth. We should first notice that
every statement is firstly an act, and only derivatively true or
false. Already here we shall have to admit that any semantic theory
misses the force of an utterance, even if it manages to capture the
meaning. If our theory of meaning is a theory of truth, then Austin's
point, that even some constative utterances can only be 'pretty good'
and not true, even further limits the explanatory range of our
semantic theory. Secondly, we should notice that meaning (including
literal truth or falsity) is not the sort of thing that tags onto
sentences, but the sort of thing that tags onto utterances. Further,
an utterance's meaning is carried by virtue of social conventions
about utterance types. These utterance types correspond to the forces
which may conventionally be carried by an utterance. If the same
sentence is taken as or meant as one utterance type it can have one
meaning - if taken as a different type, it has a different meaning.
It turns out that, in a deep sense, any theory of meaning must be a
theory of social conventions; and hence a pragmatic theory.
The same criticism is made of intensional theories. These
theories also believe it possible to explain meaning without regard
for social conventions[16]. We shall make some brief remarks here
about how the specific topics explicated in our discussion of Chomsky
demand pragmatic explanation. The first of these topics was synonymy
and paraphrase. Just as we earlier said that many constatives are not
really true or false but only 'pretty good' in some contexts, synonyms
tend to be not true or false but only 'pretty good' in the right
context. For most or all synonyms, the interchangibility of the pair
is limited to certain contexts, both semantic and pragmatic. An
example of synonyms where this is obvious is the two words,
'gentlemen' and 'guys'. Although, presumably, gentlemen and guys are
the same people, we cannot always (or even usually) interchange the
words in an actual utterance. When we open a formal address we might
utter 'Gentlemen!' - and although these same men could well be called
'guys', we cannot open the address by saying 'Guys!' The converse
might be true of trying to draw the attention of a group of friends in
an informal situation.
We shall do no more than hint at it, but we believe that the
differentiation of relations of degree tends to involve the situation
of utterance. Hence two words said to differ in degree will be
sometimes interchangeable, and other times not. Similarly, entailment
is only conventionally carried. That is, even assuming that lexemes
are compositional, as discussed, a denial of an assertion involving a
compositional lexeme may turn out to be a denial of any one or several
of the compositional elements - and which is a pragmatic matter. We
have already mentioned presupposition briefly above, and shall not
bother with any remarks on it here.
About category agreement we shall suggest that the
unacceptability of the "semantic" anomalies mentioned can equally be
explained as pragmatic anomalies. That is, just as some assertions
were previously shown to lack a conventional procedure, there are no
conventional illocutions associated with these "semantic" anomalies.
This remark does not in any way explain the particular anomalies, but
it hopefully does point out that there is no necessity that the
anomalies be explained in semantic terms.
*2.4 Davidson.*
Donald Davidson ends his essay, "A Nice Derangement of Epitaphs",
with
"[T]here is no such thing as a language. not if a language is
anything like what many philosophers and linguists have supposed.
There is therefore no such thing to be learned, mastered, or born
with. We must give up the idea of a clearly defined shared
structure which language-users acquire and then apply to cases
(p.174)."
We may begin here. In particular, what many philosophers and
linguists have supposed of language is, at least, the following three
things. What Davidson here calls '_first meaning_' we might call
'linguistic meaning' or 'sentence-meaning'.
"(1) First meaning is systematic. A competent speaker or
interpreter is able to interpret utterances, his [sic] own or
those of others, on the basis of the semantic properties of the
parts, or words, in the utterance, and the structure of the
utterance. For this to be possible, there must be systematic
relations between the meanings of utterances.
(2) First meanings are shared. For speaker and interpreter to
communicate successfully and regularly, they must share a method
of interpretation of the sort described in (1).
(3) First meanings are governed by learned conventions or
regularities. The systematic knowledge or competence of the
speaker is learned in advance of occasions of interpretation and
is conventional in character (p.161)."
If this reminds us of a figure from much earlier in this essay, namely
Saussure, it probably should.
*2.4.1 Prior theory and passing theory.*
Davidson makes a distinction between _prior theory_ and _passing
theory_, which has some connection with the distinction between
sentence-meaning and utterance-meaning. The connection is not,
perhaps, as direct as we would like, but we shall try to find it
nonetheless. Davidson explains his distinction,
"For the hearer, the prior theory expresses how he is prepared in
advance to interpret an utterance of the speaker, while the
passing theory is how he _does_ interpret the utterance. For the
speaker, the prior theory is what he _believes_ the interpreter's
prior theory to be, while his passing theory is the theory he
_intends_ the interpreter to use (p.168)."
A first try at establishing the connection between prior/passing
theory and sentence/utterance-meaning might be to claim that the prior
theory applied to an utterance gives us the sentence-meaning; while
the passing theory applied to the utterance gives us the
utterance-meaning. We shall see that this is not quite right, but it
captures just enough of the two distinctions to proceed with the claim
in mind.
*2.4.2 Malapropisms.*
Our next point in order ought to be the motivation for Davidson's
distinction between prior and passing theories, and so it shall be.
Davidson begins his essay with a discussion of what may be an
interesting topic, malapropisms. The very interesting, and peculiar,
thing about malapropisms is that we can most often understand them.
According to standard linguistic theories (semantic theories), this
fact is inexplicable. If our theory is as in (1) - (3) and our
semantic rules give the meaning 'a nice derangement of epitaphs' to
Mrs. Malaprop's utterance of 'a nice derangement of epitaphs'; how do
we understand her as meaning 'a nice arrangement of epithets'?[17]
This is where the distinction between prior theory and passing theory
comes in. Prior to conversing with Mrs. Malaprop we are prepared, and
inclined, to take 'a nice derangement of epitaphs' as just that.
However, once we have conversed with her, and it is clear that (as we
say) "she cannot possibly mean that!", we decide we would be better to
interpret the good woman as meaning 'a nice arrangement of epithets.'
This is all the more reinforced if an arrangement of epithets is a
matter we believe Mrs. Malaprop to have an interest in discussing.
Our latter interpretation is what Davidson calls a passing theory.
The former interpretation is what Davidson calls our prior
theory, but this is not identical to the theory which determines
sentence-meaning. This is because as we manage to make stereotypes
about individuals or about groups, our prior theory about how these
people are likely to talk changes. Hence, our prior theory in
entering a conversation with one person may be quite a different
matter than our prior theory regarding a second person.
Sentence-meaning is something which floats somewhere around our
collection of prior theories. Namely, it is the interpretation we
would give (according to our prior theory) to an utterance by someone
we have minimal knowledge (and more crucially, stereotypes) about.
This is a broad step away from the kind of rigid systematicity of
semantic rules which we see in traditional semantic theories. In
fact, it reverses the order of precedence between sentence-meaning and
utterance-meaning. That is: where semantic theories would explain
utterance-meaning as, at most, some sort of derivation of
sentence-meaning; Davidson explains sentence-meaning as a particular
category of utterance-meaning (namely, uninformed interpretation).
*2.4.3 Language?*
Let us turn back to the philosopher's characterizations of
language given at the beginning of our discussion of Davidson. We
claimed that the use of 'first meaning' could be translated
'sentence-meaning'; but now we can point out that this is not quite
right. 'First meaning' must, rather, be taken as the meaning which is
given us by our prior theory. If it were taken as 'sentence-meaning',
the philosopher's characterizations would not explain _any_ of what
goes on in our actual interpretation of utterances, except in the
sharply limited case of uninformed interpretation. That is, both
prior theory and passing theory play some role in interpreting an
utterance, but neither is fixed in the way that sentence-meaning is
presumed to be fixed. But if we turn with this caveat to the three
characterizations we shall find some new problems with these
characterizations. Davidson criticizes the second characterization.
"It is quite clear that in general the prior theory is neither
shared by speaker and interpreter nor is it what we would
normally call a language. For the prior theory has in it all the
features special to the idiolect of the speaker that the
interpreter is in a position to take into account before the
utterance begins. One way to appreciate the difference between
the prior theory and our ordinary idea of a person's language is
to reflect on the fact that an interpreter must be expected to
have quite different prior theories for different speakers
(p.171)."
That is, there is no particular reason to suppose that speaker and
hearer come into an interaction with the same prior theory. Either
communication can be achieved by bringing the passing theories into
line, with or without changing the prior theories correspondingly; or
communication can fail to a great or small degree. However, we should
be careful not to overestimate the degree of accurate theory necessary
for communication. If the speaker and hearer share only a small part
of the same theory (either prior or passing) they are likely to
communicate a great deal - due to redundancy in ordinary speech,
conversational cues, reconstruction of speaker intent, and numerous
other such devices.
We have already shown the unacceptability of characterization
(3), but we shall make this explicit. It was claimed that first
meanings are governed by learned conventions, which are learned in
advance of interpretation. We have pointed out that, in fact, first
meanings are tailored to our stereotypes about individuals and groups
- and not candidates for the semanticist's sentence-meaning. This
does not really answer the question of whether these first meanings
are conventional, because we have no clear sense of what conventions
are and are not. All that we shall be able to do toward settling this
issue is point out a couple properties that conventions might have.
Conventions might need have a certain rather wide circulation - if so,
then characterization (3) is false. Since prior theories are often
differentiated down to the individual, a particular prior theory might
only exist in one person's evaluation of one other; and this is a most
narrow circulation. Conventions might be imposed by social pressure
rather than deliberately chosen - if so, then again characterization
(3) is false. While it is certainly possible that social pressures,
formal or informal, may sometimes cause us to choose one prior theory
over another, there is no reason that we know to suppose that this is
the case as a rule. At very least it is an empirically contingent
matter whether such pressures exist, or whether our prior theory
(which amounts to stereotypes, at the end of the day) is purely an
expression of preferences and cognitive processes.
*2.4.4 Language revisited.*
The denial of characterizations (2) and (3) has certainly put a
crimp on our notion of there being a language. Davidson's arguments
have, hopefully, shown that these two characterizations cannot really
be true of anything, or at least anything which affects how we
actually interpret real utterances. Davidson does say, however,
"Perhaps we can give content to the idea of two people 'having
the same language' by saying that they tend to converge on
passing theories; degree or relative frequency of convergence
would then be a measure of similarity of language (p.173)."
We find there is something to this, and shall discuss this notion in
the final section of this paper. A better way of phrasing this, for
our purposes, might be that two people are able to form the same
stereotype. Hence, where Davidson says,
"If we ask for a cup of coffee, direct a taxi driver, or order a
crate of lemons, we may know so little about our intended
interpreter that we can do no better than to assume he will
interpret our speech along what we take to be standard lines
(p.170);"
he is really thinking of some stereotype pattern which just anyone can
be assumed to possess. "Standard lines" has much the same status as
"language"; both say something (and nearly the same thing) about who
can and will act on which stereotypes. Now we must notice that this
matter of who has which stereotypes is a purely contingent one.
Notice also that these matters are essentially characterization (1).
To the degree there actually are standard lines and languages,
characterization (1) holds. We raise some of these questions again in
section 3, particularly the status of characterization (1).
*3.0 Conclusion.*
Several of the claims we made in our discussion of Davidson were
empirically contingent, and so it would be well to see how "the facts"
line up. The question which is most interesting to us, here, is that
of whether there are languages and/or standard lines of interpretation
- given the meaning of there "being languages" which we had Davidson
give for us. Linguists have been considerably less sanguine about the
existence of language than have philosophers - and as the former
depend on the study of language for their livelihood to a greater
degree than do the latter, we might pay them some attention. If we
stop taking as brute that it is the existence of a single language
which grounds each idiolect, we might well doubt that any language is
actually emergent out of the potpourri of idiolects. We would like to
point to a couple of quotes from linguists which may support
Davidson's assertion that there is no such thing as a language.
*3.0.1 Some "facts".*
In David Decamp's paper, "Introduction: The Study of Pidgin and
Creole Languages", he says,
". . . there is also a socio-economically oriented linguistic
continuum in Jamaica, a continuous spectrum of speech varieties
whose extremes are mutually unintelligible but which also
includes all possible intermediate varieties. At one end of this
continuum is the speech of highly-educated Jamaican leaders, many
of whom claim to be speaking standard British English but who are
actually using what seems to be evolving into a standard Jamaican
English; it is mutually intelligible with but undeniably
different from standard British. At the other extreme is the so
called 'broad creole' or 'broken language', the variety which so
far has recieved the most attention from linguists. Each
Jamaican speaker commands a span on this continuum, the breadth
of the span depending on the breadth of his [sic] social
activities; a labor leader, for example, can command a greater
span of varieties than can a sheltered housewife of the suburban
middle class. A housewife may make a limited adjustment downward
on the continuum in order to communicate with a market woman, and
the market woman may adjust upward when she talks to the
housewife. Each of them may then believe that she is speaking
the other's language, for the myth still persists in Jamaica that
there are only two varieties of language - standard English and
'the dialect' - but the fact is that the housewife's broadest
dialect may be closer to the standard end of the spectrum than is
the market woman's 'standard'."
Lest we suggest that this arrangement of varieties of speech is
strictly socio-economic let us use one other quote. After briefly
rejecting mutual intelligibility as properly slicing up languages
(e.g. the "languages" Norwegian and Swedish are mutually
comprehensible) - and we shall come back to this suggestion - R. A.
Hudson, in his book _Sociolinguistics_, says,
"Varieties may be arranged in a _Dialect Continuum_, a chain of
adjacent varieties in which each pair of adjacent varieties are
mutually intelligible, but pairs taken from opposite ends of the
chain are not. One such continuum is said to strech from
Amsterdam through Germany to Vienna, and another from Paris to
the south of Italy (p.36)."
It is not only recently that these linguistic transitions have been
noticed. No less a figure than Saussure also discusses them. Not
only does Saussure realize that dialect continuums exist[18] but he
gives a precise explanation for them. Saussure talks of what he calls
_innovating waves_ - which are the process by which languages evolve.
He says,
"(1) Evolution takes the form of successive and precise
innovations that include as many partial facts as could be
enumerated, described, and classified according to their nature
(phonetic, lexicological, morphological, morphological,
syntactic, [for our interest, semantic], etc.).
(2)Each innovation embraces a definite and delimited area. There
are two possibilities: either the area of innovation embraces
the whole territory and creates no dialectal differences (the
less usual possibility), or the change affects only a part of the
territory, each dialectal fact having its special zone (the more
common occurence). (p.200)"
The consequence of a history of such innovations is a dialect
continuum - the closer two groups are to one another, the more
dialectal items they are likely to share. The sense of "close" here
shold not be taken in the purely geographic terms which Saussure
emphasizes, but should also be taken to include "closeness" in
socio-economic factors, ethnicity, gender perhaps, and any number of
other things. Two groups which are sufficiently close speak in
mutually intelligible ways; and this is because they share enough
stereotypes about "standard lines" of interpretation to come up with
passable interpretations of each others' meanings.
*3.0.2 Some claims.*
We are, however, getting somewhat ahead of ourselves - as the
quotes we have given are not really any more than anecdotal. What we
are trying to show in our denial of language is that characterization
(1), which we gave in our discussion of Davidson, is false. This
characterization was, in brief, that language is systematic. More
exactly, the characterization was that 'first meaning is systematic'.
However, the claim - when taken as an empirical one - that there are
languages amounts to the claim that there is a particular first
meaning that just any speaker, within some specifiable group, can and
does use. For us to claim that, ala Davidson, there is no such thing
as a language we must be prepared to show that there is no first
meaning which is within the repertoire of every speaker who we think
speaks a given language. To do so is tantamount to showing that
"language-groups" do not divide up speakers in any way which we would
find at all acceptable; and to do this is, in turn, to cast doubt
whether all, or any, speakers are speakers of a "language". In terms
of a more traditional epistemological question, we might ask whether
languages are natural kinds. We ask here a number of different
questions in different terminologies, which are nonetheless somehow
equivalent - and all of which should be open to empirical evidence.
Not all the notions used in the above family of assertions are
entirely clear; hence, our effort will have to be to clarify these
notions at the same time as we argue for the assertions involving
them. We should therefore be prepared to revise the notions at the
same time as we check the assertions. Most of this project shall not
be within the bounds of this paper. Our only goal in the remainder of
this paper is to suggest that the family of assertions above warrant
methodological pursuit, and that some of the concepts involved are
likely to be fruitful.
*3.1 Once again: language?*
In the speech communities sketched in the above quotes, there
seems no natural way to delimit the number of languages spoken.
Turning to the Jamaican case we notice that "standard" Jamaican is
intelligible by "standard" British speakers, the latter by midwest
Americans, and these in turn by some, but not all, African "English"
speakers. We commented above that mutual intelligibility did not
really work as a possible criterion of language membership - but it is
closely tied with shared first meaning and linguistic stereotypes; any
possible sense we might hope to make of "language" will be fairly
closely related to mutual intelligibility. If we limit ourselves to
those speakers in this sort of chain who are traditionally called
"English speakers" (e.g. on your favorite language distribution map),
we have still included a huge proportion of the world's speakers. If
we further realize the banal fact that mutual intelligibility comes in
degrees (often not symmetrically distributed in a conversational
pair), the intelligibility relation forms a richly variegated topology
on the universe of "English" speakers.
It seems reasonable to ask what the "English" language consists
in, given the above remarks on mutual intelligibility. What
linguistic stereotypes do all English speakers share? What first
meaning(s) can every English speaker resort to as "standard lines"?
One presumes that a language must be, in essence, a grammar combined
with a lexicon - and this is consistent with the semanticists we have
discussed: Saussure, and particularly Chomsky. Let us focus on the
lexicon, though we do not think that any principled difference exists
in the grammar (although grammar is more resistant to the kind of
innovating waves described by Saussure, so somewhat greater
homogeneity may exist between "distant" speakers). Unless we make
cultural imperialism a linguistic principle, we cannot assume that,
for example, the O.E.D. has the last word on membership in the lexicon
of "English". Actually, even if we are cultural imperialists, this
proposal will not help us if our effort is descriptive linguistics.
We could claim that the O.E.D. gives the prior theory that one
_should_ use when entering a discourse (with the right sort of person
- namely, one we don't know), but this still says nothing about the
standard lines that speakers _do_ use in such encounters. As a matter
of the facts, different speakers use the same words (i.e. sounds,
phonetic/phonemic strings) to convey different semantic values, and
this is true even of the circumstance where we are talking to persons
about whom we have few or no stereotypes. Further, one speaker will
often choose different prior theories, of different standard lines,
for different occasions of speech encounters; in particular, this is
not only because of the stereotypes we bring to an encounter, but
because of completely extraneous factors such as our mood, our other
recent encounters, our conversational goals, and others. We doubt
that any one semantic value can be singled out as belonging to
"English"; particularly given that nearly any such value will never be
used, nor understood, by some groups of (or individual) "English"
speakers.
*3.2 A re-examination.*
Once we examine the actual links that people make in concrete
situations, between lexical items and semantic values we are compelled
away, even, from Davidson's prior theory to his passing theory. That
is, while we cannot really argue against a prior theory framework on
theoretical or empirical grounds, we think the idea of a prior theory
to be largely superfluous. If we break down the line of analysis of
speech (and interpretation) which we have been developing, we describe
speech (and interpretation) as follows. A person enters a
conversational context with only her linguistic stereotypes (about
groups or individuals), knowledge of conventions, conversational
goals, and hypothoses about other's goals, intentions, beliefs and so
on. She does not gain anything else while in a conversation. Our
goals and our hypothoses about other's goals were discussed during our
discussion of Grice; conventions during the discussion of Austin; and
linguistic stereotypes during the discussion of Davidson. In the
latter, we made a distinction between prior theory and passing theory,
which divided the stereotypes we make into two groups. However, this
now seems like a distinction without a difference. The two groups do
not divide stereotypes into different kinds, but only into those we
have with our mouths open and those we have with our mouths shut.
Hence we will, from this point, not bother with Davidson's
distinction, but only speak of linguistic stereotypes. These
stereotypes are of two general sorts - the same categories which
language was presumed to have - stereotypes about lexical/semantic
(signifier/signified) links, and stereotypes about grammaticality.
*3.2.1 Chomsky revisited.*
It is here that we are able to make some remarks about Chomsky's
notion of syntax, and to a degree about his ideas of semantics.
Chomsky presumes that we have syntax and semantics generators in our
brains. These are recursive machines which are able to generate the
infinite number of sentences which we can produce and understand.
While we cannot really understand or produce an infinite number of
sentences, we can see the sense in Chomsky claiming we can. There is
no clearly defined upperbound on the length of sentences. Any
upperbound we might give - such as, "the longest string of phonemes
utterable in 72 years" - probably gives us more different sentences
than our brains could hold by rote enumeration. Hence, there is
reason to think that we do utilize recursive rules for the generation
of sentences. However, there is no immediate reason to suppose that
each person uses the same recursive rules, or that a single person
sticks to the same set of rules from utterance to utterance. We
shall, in fact, propose that sets of rules are as common as are
linguistic stereotypes. More strongly, we shall claim that our
stereotypes of grammaticality are just simply sets of recursive rules.
When we enter into a discourse situation we bring the set of recursive
rules of grammaticality, which we believe the other to hold, into our
interpretation and construction, just as we bring in our stereotypes
about the other's lexical/semantic links. Chomsky's mistake, on this
picture, is exactly the same as absolutely any other semanticist - he
believes we have just one set of rules for interpreting or creating
utterances.
*3.2.2 Stereotypes.*
Let us turn back to the discussion of stereotypes, perhaps it
will be illuminating. Lexical items (and recursive rules) are
distributed by idiolect, and by register. This latter may be
understood as the speech situation, thought of in broad terms such as
formal/informal, authoritative/subordinate, solidarity/distance, etc.
We might conceivably enumerate the elements of register. Let us call
a particular speaker's idiolect in a particular register at a
particular time the _"utterance-context"_. We prefer this
condensation of speaker and situation into a single variable (though a
multi-dimensional one), because we concur with Volosinov in thinking
the person of a speaker to be no different in kind from the other
situations encompassing utterances. It seems to be empirically true
that lexical/semantic links and recursive rlues of grammar, while
somewhat systematic, are not common to all utterance-contexts (or most
even), even within fairly narrow social groups. A particualar string
of phones/phonemes is used and understood one way by one class, in one
region, and using one register - and differently elsewhere. Class,
region, and register are often dividing lines for different uses of a
sound, but none of these are often sharply divided.
It is usually the case that we are unsure about the semantic
values and recursive rules, to a greater or lesser degree, used by
others in our conversations. Symmetrically, we are often unsure how
our lexicon and syntax will be taken. Often, prior to a conversation
we call to mind stereotypes of the other's likely lexical/semantic
connections, based on her appearance, demeanor, and dress. In all
cases, we draw these stereotypes once engaged in conversation. The
other's accent, lexical choice, and grammar lead us to further
stereotypes about her class, region, register, etc. - which we use to
determine her likely semantic values and intended syntax. We also use
all the above items to develop a Gricean "passing theory" about our
partner's likely appraisal of our own class, region and register.
Insofar as there is such a thing as a "standard" language, both
parties may choose the register in their repertoires which most nearly
conform to the "standard". What makes one set of lexical/semantic
connection (or recursive rule) "standard" in not the ontological
status of languages, it is actual historical facts about invasions,
immigrations, and power-structures in societies.
This ability to form stereotypes largely accounts for the success
and failure of mutual intelligibility. In DeCamp's description of
Jamaica above, this is well illustrated. The housewife and market
woman are able to communicate just insofar as they form reasonably
good stereotypes of one anothers' lexical/semantic connections and
recursive rules (although they may yet believe their stereotypes to be
much better than they actually are). These stereotypes may be
available to whatever degree for a listener. If the author personally
hears a Japanes speaker, he will not make any lexical/semantic
connections; if he hears a Russian or Chicano he will make a few
(though not many); if he hears a Jamaican creole speaker he will have
more correct stereotypes; if he hears an American midwesterner, most
all the stereotypes will fit. A passing thought here is that it might
be easier to define languages by mutual unintelligibility than by
mutual intelligibility. That is, two speakers may be said to "share a
language" if they are both incomprehensible by most all the same
people. This shall not be pursued herein, but neither is it quite a
joke.
*3.2.3 Semantics, grammar and pragmatics.*
We can see that the framework of families of inter-related
semantic values and recursive rules of grammaticality bring us much
closer to a pragmatic theory than to a semantic one. We should also
notice that stereotypes are no more pre-given than are languages.
That is, it is not merely that we know some large number of languages
before entering a discourse situation; and then merely try to choose
the right one. Rather it is that, just as new sentences can be
generated, so can new stereotypes. The Soviet semiotician Yuri
Lotman, in his essay "Problems in the Typology of Texts", nicely
captures something like the picture we are aiming for - albeit his
concern is for the written, not the spoken word.
"Our reasoning is not constructed thus, 'Pragmatics, not
semantics and syntax determines the character of the text,' but
thus: 'A change in the function of a text gives it a new
semantics and a new syntax (p.120)."
Lotman's concern is admittedly somewhat different than ours, and so is
his terminology. We should best understand, for our purposes, 'text'
as 'sentence' or 'sound' (i.e. that which stays fixed); and 'function'
as 'utterance-context'. This is not rewriting the whole intention of
Lotman's assertion, but merely shifting the concern from writing to
speech. We shall show below that Lotman is most sympathetic with our
program.
*3.3 Conventions and intentions.*
Stereotypes, for all the work they do, are not the whole of our
explanation of speech and interpretation. At least two other sorts of
factors enter in. The first is conventions; the second is intentions.
We have been and remain very uncomfortable about talk of conventions,
as it is enormously less that clear what conventions are and which
conventions exist. However, there is work to be done which cannot be
done with stereotypes alone. This work concerns speech-acts.
Although our knowledge of conventions is not, perhaps, sharply divided
off from the stereotypes we have - the paradigm cases of each play
distinctly different roles. The conventions we are aware of are of
what _acts_ can be performed by saying things. Our stereotypes are
still, as our quote from Lotman brought our, contextualized semantic
theories. Our discussion of Austin should, hopefully, have brought
out semantic theories' inability to deal with speech-acts. When we
say, 'I do take this woman', no possible values we can give to the
individual lexical items, and no set of recursive rules we can use to
explain the composition, can explain the meaning or force of the
utterance, _unless we know that there is a conventional procedure to
become married_, and that this is it. In some sense, our explanation
of speech and interpretation in terms of stereotypes (however much it
departed from traditional semantic theories) is still a semantic
theory. We shall not reject this semantic theory, though we shall
reject any semantic theory which does not demand _radical_
contextualization. However, we do claim that a pragmatic theory of
speech-acts is necessary on top of our "semantic" theory of
stereotypes.
*3.3.1 Conventions.*
Our pragmatic theory, though, must take precedence over our
"semantic" theory. Recall from our discussion of Volosinov,
"[T]he criterion of linguistic correctness is submerged by a
purely ideological criterion: an utterance's correctness is
eclipsed by its truthfulness or falsity, its poeticalness or
banality, . . . [whether it is] good or bad, important or
unimportant, pleasant or unpleasant, and so on."
There are conventions that we say things which are true rather than
false (for the particular context of utterance); which are important
rather than unimportant; which are pleasant rather than unpleasant;
and similarly, which are contextually grammatical rather than
ungrammatical. Each of these convention is, or course, breakable and
often broken. At the same time, we know that if we break these
conventions we are subject to censure of a more or less conventional
nature, which depends largely on more or less enumerable aspects of
the utterance-context (and do not forget that utterance-context
includes the person of the speaker). Since contextual grammaticality
can be included in these categorical conventions without any strain on
the framework, it must be subsumed under our pragmatic theory. We
have already stated that stereotypes, like utterances, must be
recursively generable. When we claim that there are conventions
(pragmatic objects) about which arrays of stereotypes to bring to a
particular utterance-context, it follows in an immediate fashion that
parts of our pragmatic theory must include rules of recursive
generation. Whether we include the actual rules for generation of
sentences in our pragmatic theory, or only include the rules for
generating new patterns of stereotypes is of little or no matter.
*3.3.2 Intentions.*
In addition to stereotypes and conventions, the process of speech
and interpretation involves intentions, in various ways. These are
the Gricean kinds of cases. While Grice's conversational rules are
indeed conventions of a certain sort, saying this does not quite go
far enough toward explaining actual utterances. The thing that is
left out is the interpretation we actually end up with from assuming
that speakers are obeying these conversational rules. So for example,
when we respond to 'Where can I get some petrol?' with 'There is a gas
station around the corner,' we must presume certain intentions to
understand the interchange. The meaning which we should like to
assign to the latter utterance is of the sort, 'You may get petrol
around the corner.' The linguistic stereotypes we have of the
utterance (imagine an actual circumstance) are not likely to give us
this meaning. Neither are there conventions specific enough to direct
us to this meaning. It is conventional to give responses to
questions, and to tell truths rather than falsities, and so on. But
none of these are specific enough to determine the actual meaning
being given to the utterance above, unless we supplement our
understanding by certain assumptions about the intentions of the
speaker. This point was, hopefully brought out in our discussion of
Grice. Hence, we shall let this mention of it suffice herein.
*3.4 Lotman.*
Lotman, whom we have already mentioned in passing, is interested
in the abstract study of information. While speaking is clearly not
entirely a matter of transmitting information, this element does
figure in. Almost as a postscript we wish to consider whether the
consideration of language in information-theoretic terms will lead us
to the same framework we reached by other means. The information
which is of interest to Lotman is not only of the world outside the
communicative act, but information about the communicative act itself.
If we reflect just briefly on the processes of actual speech we will
realized that communication about communicative acts is a large part
of our total communication.[19]
*3.4.1 Sameness and difference.*
When we are concerned with communication about communicative acts
we are, according to Lotman, driven to an explanation in terms of an
opposition between sameness and difference of code. This is a
structural opposition, which should remind us of the negative
definition of signifiers and signifieds which we saw in Saussure.
However, Lotman is not concerned with the individual elements in
communication, but with communicative systems as a whole - what we
would call 'stereotypes' or 'systems of stereotypes'. Lotman makes
these remarks about the play between sameness and difference of code.
"Let us imagine two human individuals who still exchange not
signs but irregular signals that are involuntary symptoms of
their psycho-physiological processes. The common character of
some elementary codes of the 'fear-pleasure' type and the common
character of their surrounding situation, which is deciphered by
this code set, allow them to distinguish a common feature in
mutual signals that they qualify as 'intelligible.' The
possibility arises of the equivalent exchange of identical
meanings."
This is the sameness aspect of code. Lotman continues.
"However, while the mechanism from which the system of semiotic
communication can develop is present here, the content for
exchange within it is theoretically not provided. A participant
in communication is operative for me presicely because he is
'another person,' and the information obtained from him is
valuable precisely insofar as it issues from another person and
does not duplicate what is already known to me. To the extent
that participants in communication are united by a common code,
they are one person. (both p.95)"
This is the difference aspect of code.
*3.4.2 Code and content.*
Before we proceed to Lotman's remarks on the precise opposition
between these, we need to explain parts of the above quotes. Lotman,
following his compatriate Volosinov, does not wish to distinguish code
and content in natural language - at least not in the simplistic way
which analytic semantics would. For Volosinov, as we saw, this
follows from the absolute semioticity of thought. Every thought and
every perception is already a semiotic object when it is presented,
and does not have a pre-semiotic "content". For Lotman it is the
same; any information we have in our brains is not merely passively
realized in an objective code, but is the code itself. Hence, two
persons who are not identical in the information they contain think in
different systems of code.
A problem arises as to how information can be conveyed between
two persons with different systems of code (but yet it is only those
with different systems of code who have any information _to_ convey).
Lotman describes the situations of similarity and difference of code.
"Thus two semiotic situations are inherently given: in one
situation, the mechanism of communication is inherently given,
but the content of communication is theoretically absent; in the
other situation, the content of communication is inherently
given, but the mechanism of communication is theoretically
absent. It is apparent that real speech can be assessed as a
compromise and oscillation between these systems (p.96)"
This oscillation and compromise appears at a synchronic level as an
"inherent structural tension" in the pattern of communication.
However, at a diachronic level the structural tension plays a
functional role in the maintenence of societies. Let us examine the
opposition of sameness and difference at a diachronic level.
*3.4.3 Diachrony.*
Saussure has rightly pointed out that there is a tendency toward
sameness of language. We explained this in terms of the desirability
of communication, and the rational method of maximizing it, and so on.
Our explanation was undoubtedly much too intentional. Persons are not
nearly so rational as our discussion suggested; which is not to say
that there is not a societal movement towards standardization of
language: there is! Saussure should be excused from the blame for
this intentionalism, as it was only our heuristic in explaining the
tendency he observed. However, what Saussure observed was only half
of the picture. There is also a tendency toward individualization of
language. Lotman says this.
"The development of a linguistic system brings about an ever
greater individualization of language. This is linked to the
fact that the personality of both speaker and hearer become
increasingly complex in the course of cultural evolution. There
is seldom growth in the number of semiotic systems with which
each person encodes his [sic] speech behavior. But the number of
possible combinations of linguistic codes eventually exceeds the
number of members in the collective; that is, it becomes
individual. (p.97)"
We should not take too seriously Lotman's implicit claim that
there was some aboriginal society in which codes were not as common as
individuals. Russian linguists and philosophers since Marr have taken
their metaphors about the origins of language much too seriously.
Nonetheless, Lotman suggests how real historical people do become
differentiated in their semiotic systems.
These two opposing tendencies, for codes to differentiate and to
generalize, play a functional role in societies. This we have said.
It would be premature to claim that this functional role actually
explains the tendencies; so we shall offer it only as a suggestion,
not an assertion. The function that similarity of code plays in
society is the obvious one of allowing communication. The function of
differentiation of code is as follows.
"Specialization in the structure of individual codes - the
possibility of a purely personal representation in text of
extralinguistic reality - meets deep needs of the collective as a
whole, since a shortage of information typical of any human
collective can most effectively be compensated for by the
stereoscopic quality, polyglottism, and multi-level character of
specialization. (p.97)"
The specialization mentioned here is analogous to what Hilary Putnam
calls "division of linguistic labor". That is, different people are
allowed the final word on different linguistic matters. The
'stereoscopic quality' Lotman mentions here is merely the greater
effectiveness which a multiplicity of perspectives brings do deciding
an issue - whether it be one of truth, morality, beauty,
appropriateness, or whatever.
*3.4.4 Language concluded.*
The diachronic perspective suggested above allows us one final
chance to conceptualize what language could be, and what it could not
be. A language could be a synchronic slice of a group of humans, in
which the tendency toward commonality of code is predominant. Whether
such slices exist is an empirical matter, as have been the other
suggestions we have made about what languages might be. This account
is in keeping with our conceptions in terms of stereotypes, and common
prior theories, and mutual intelligibility. However, this perspective
allows us to see something which the others have not. If languages
exist in particular places at particular times this is an entirely
accidental fact. Nothing within language (on any of our accounts)
causes languages to reproduce themselves. Only particular social
facts, of a non-linguistic nature, could have the effect of
reproducing language. In particular, for a language to be a lasting
entity there would have to be some outside pressure for the conformity
of linguistic codes. We conjecture that this would have to be the
result of a most peculiar functional demand of a society.
*Bibliography and key*
Saussure, _Course __in __General __Linguistics_, Philosophical
Library, 1959. (all references)
Volosinov, _Marxism __and __the __Philosophy __of __Language, _Harvard
University Press, 1986. (all references)
Lacan, _Feminine __Sexuality, _edited with introductions by J.
Mitchell and J. Rose, W. W. Norton and Co., 1982. (Fem.)
Lacan, _The __Four __Fundamentals __of __Psychoanalysis, _W. W. Norton
and Co., 1977. (FF)
Lacan, _Ecrits: __ __A __Selection, _translated by Alan Sheridan,
Norton, 1982. (Ecr.)
Kristeva, _The __Kristeva __Reader, _Columbia University Press, 1986.
(Read)
Kristeva, _Desire __in __Language, _Columbia University Press, 1980.
(Des.)
Chomsky, _Studies __on __Semantics __in __Generative __Grammar,
_Mouton and Co., 1972. (SGG)
Chomsky, _Aspects __of __the __Theory __of __Syntax, _John Wiley and
Sons limited, 1965. (ATS)
Chomsky, _Syntactic __Structures, _M.I.T. Press, 1957. (SS)
Quine, _Set __Theory __and __Its __Logic, _Belknap Press, revised
edition, 1969. (all references)
Kaplan, David"DThat", Syntax and Semantics v.9, 1978.
"On the Logic of Demonstratives", ?
Manuscript of "Demonstratives", 1977.
Davidson, "A Nice derangement of Epitaphs", in _Philosophical Grounds
__of __Rationality, _edited by Grandy and Wagner, Academic Press.
(all references)
Grice, "Presupposition and Conversational Implicature", in _Radical
Pragmatics, _edited by Peter Cole, Academic Press, 1981. (PCI)
Grice, "Logic and Conversation", in Syntax and Semantics v.3, edited
by Cole and Morgan, Academic Press, 1975. (LC)
Hudson, _Sociolinguistics, _Cambridge University Press, 1980. (all
references)
DeCamp, "Introduction: The Study of Pidgin and Creole Languages", in
_Pidginization __and __Creolization __of __Language, _edited by D. H.
Hymes, Cambridge University Press, 1971. (all references)
Lotman, Yuri. "Primary and Secondary Communication-Modeling Systems";
"Two Models of Communication"; "Problems in the Typology of Texts", in
_Soviet __Semiotics, _ edited and translated by Daniel Lucid, John
Hopkins
University Press, 1977. (all references)
Austin, _How __to __Do __Things __with __Words, _Clarendon Press,
1962. (all references)
1. We use 'teleological' and its cognates in a somewhat nonstandard
sense. Where the normal use might be only to describe systems which
through internal laws tend towards some ideal state or equilibrium, we
use a somewhat broader, though, we believe cohesive sense. We shall
use 'teleological' to describe any system or object which either tends
towards an ideal state, or toward an ideal, and perhaps externally
defined, regularity in behavior. Further, we shall allow that this
tendency might be the consequence, not only of internal laws, but of
the structure in which the object is placed. This applies concretely
to our interest by assuming the object of interest to be persons,
whose behavior may tend toward e.g. the production of syntactical
speech due to their position in the structure of society. We shall
argue that in this sense, persons, in Saussure's schema, are
teleologically directed toward langue. This should also probably be
the case of Chomsky's 'competence', though we shall not specifically
deal with this herein.
2. We use 'pragmaticist' here not in reference to the term which
Pierce invented to distinguish his philosophy from James' pragmatism,
but in the sense found in linguistics. Pragmaticism in linguistics
is, roughly, the position that the meaning and force of utterances are
to be explicated in terms of the pragmatic situation of utterance, not
by a semantic theory of language.
3.'Arbitrary' is the word normally used in talking about Saussure's
thought, and the word that occurs in the English version of his
_Course. __ _However, it is not to be understood in quite the sense
normally associated with it. A better word to use in the place of
'arbitrary' and its cognates might be 'conventional'. That is, when
Saussure writes of the "arbitrary matching up of sound-images with
concepts", he does not mean that it is done at random, or at the whim
of each individual user at each occasion of use. Rather Saussure
merely means that the particular matching that occurs does not follow
any natural (i.e. asocial) necessity. This is not to imply, as we
discuss, that each speaker can do the matching she chooses. We shall
retain the use of 'arbitrary' and its cognates, however, in order to
remain consistent with the English translation of Saussure's work.
4. By a 'system of virtual representation' we mean any sort of
computational system which, although not actually storing emplates, is
able to make comparisons between inputs of different times past. We
think that it must be generally acknowledged that the mind is a
virtually representational system, if not an outright representational
one.
5. I shall use the square brackets around the names of conceptual
values. These values are not the same as either the thing named, or
the word naming (though in Saussure, and elsewhere there is a close
tie between the word itself and the concept). Hence, [snow] is to be
distinguished from both snow (the stuff) and 'snow' (the word).
6. An obvious confusion presents itself at this point in connection
with the word 'contract'. This is nearly the same ambiguity that
arises with many seventeenth and eighteenth century thinkers over the
so-called 'social contract'. One reading of 'contract' would be a
literal one in which some collective actually met and somehow agreed
on particular conventions. The conventions of interest to Saussure
are, of course, the semantic connections in langue. However, this is
by no means intended by Saussure (and perhaps not by 'social contract'
theorists). Rather, Saussure intends to make a purely
phenomenological or psychological claim about the way that language
"appears" to a speaker. That is, language stands in the same relation
to the speaker as does a contract she (or her society) has actually
entered into. However, we might still be unhappy with this phrasing,
as an actual contract, unlike the language, is so-to-speak, "made to
be broken". But here again, we should do well to think of Saussure's
"contract" in much the same way as the "social contract". It is
something which appears to have been inexorably entered into by some
previous generation (with the emphasis still on "appears").
7. The word 'iconic' is here used in a somewhat technical sense.
Pierce originated the tripartite distinction between icon, symbol, and
sign - and these are fairly common terms in semiotics today. Ignoring
the difference between symbol and sign, we may merely contrast
'iconic' with Saussure's term 'arbitrary'. An iconic sign is one in
which some natural fact compels the particular relation which exists
between a signifier and its signified. The most obvious example of
this is an actual icon, which is fashioned in the shape of the figure
which it is meant to represent.
8. Some essays in which we see this are "DThat", "On the Logic of
Demonstratives", and "Demonstratives". See the bibliography for full
information on these.
9. In particular, haecceitists would maintain that all objects are
actual objects; while anti-haecceitists think that some objects are
only potential.
10. The difference in the presupposition of past-perfect active and
passive is over whether the subject (in active version) is presumed
living. The example here is that in 'Einstein has taught me physics',
Einstein is presumed to be alive. With, 'I was taught physics by
Einstein' no such presumption exists.
11. 11. Chomsky reads the structure of (10) as,
(10')It was (an ex-convict (with (a red (shirt)))) that he
was warned to look out for.
12. 12. The examples are from _Aspects of the Theory of Syntax_.
13. Explicit performatives are speech-acts in which the act being
performed is named by the utterance by which the act is performed.
These are to be opposed to _implicit performatives_, in which,
although the same act may be performed, the act is not named.
Examples of these might be the explicit performative, 'I make X a gift
to you'; and the implicit performative, 'Here you go.' These might
both be used for the purpose of making a gift, but only the first
names the act performed.
14. When we use 'force' in this discussion we are really intending to
refer to the illocutionary act being performed by a given utterance.
For simplicity we have left Austin's description of locutionary,
illocutionary, and perlocutionary acts out of our main text. In
brief, the distinction is as follows. Locutionary acts are the
issuing of certain sounds with the intent and possibility that they be
taken as sounds in language. Illocutionary acts are acts which are
accomplished by or in making a locutionary act. The prime examples of
these are explicit performatives, in which "to say something is to do
something." Perlocutionary acts are acts which are achieved as
consequences of performing a locutionary or illocutionary act. We may
often think of these as achieved _by_ performing the locutionary or
illocutionary act, but not normally _in_ performing the latter. In
general, perlocutions are non-conventional, though, of course, no
sharp line exists between conventional and non-conventional actions.
Hence, our examples of forces may ease into perlocutions. In our
discussion of force, we give examples of cases in which the force is
other than the "normal" force of an utterance. By this it should be
understood that a different illocution is being performed than that
which is normally associated with the given utterance.
15. See the above footnote for a brief discussion of locutionary and
illocutionary acts.
16. The conventions which are not considered by semantic theories are
not, of course, those conventional semantic links which, on some
accounts, define language itself. Rather, these neglected conventions
are those which concern the performance of illocutions.
17. This example is from Davidson's essay and, I guess, before that
from a Sheridan character. The context should make it clear enough
what is going on.
18. There is no point in demonstrating that Saussure speaks of dialect
continuums, in the main text; Saussure says almost exactly the same
sort of thing as the other authors we quoted. However, to put the
readers mind at ease we can give this example.
"It is impossible, even in our hypothetical examples, to set up
boundaries between dialects. The same applies to related
languages. The size of the territory makes no difference. We
would be unable to say where High German begins and Low German
ends, and would find it just as impossible to draw the dividing
line between German and Dutch, or between French and Italian.
There are extreme points where we may assert, 'Here French
predominates, here Italian,' but in the intermediate regions the
distinction would disappear. (p.204)"
19. As an example of the importance of communication about
communication we can point to Austin's estimate that there are names
of speech-acts of a number in the order of 10^3^. If the vocabulary
of English is about 200,000 words, then this means that anywhere up to
5% of our lexicon may be names of communicative acts. This is
certainly an important portion. It is, of course, unclear with this
information alone how much of our actual speech is about communicative
acts. But this is also probably of at least the same order.