Chapter 1:
There have been a variety of attempts to distinguish, if
only terminologically, the difference between the language of literature and
the language of not-literature. For example, Philip Wheelwright, in The Burning Fountain, opposes what he calls
"steno-language," or literal language, and "depth" or "expressive language," the
language of literature (3-4). While his terminology is far too limiting, if
not idiosyncratic, to be useful here, he does make two important points that
are worth appropriating. His first caveat concerns the distinctive traits
he ascribes to the two varieties of language; he asserts that "the differentiation
is by no means absolute but admits of the most varied and subtle degrees,
disguises, and overlappings" (73). He then adds that his concept of "steno-language"
is not to be confused with any particular body of discourse, notably that
designated "scientific." This latter concern is worth expanding;
neither literary nor instrumental language should be identified as belonging
to any particular genre(s) of texts. Both kinds of language appear in all
discourse; distinctions may only be made in terms of proportion.
As regards Wheelwright's caveat, while the proposed differentiation
between literary and instrumental language is perhaps less subtle, the vagaries
of language suggest that it would be foolish to believe that absolute distinctions
will always hold. Allowing for the difference between theory and reality,
Wheelwright defines steno-language as "the negative limit of expressive
language" (74). Literary language and instrumental language are similarly
dichotomaic; it is impossible for either to exist in pure form. Defining instrumental
as the negative limit of literary language nonetheless provides a useful approach
to understanding both; however, it first requires a definition of literary
language.
It is important to dispense immediately with a question
that may be present in the mind of the reader; the question, of course, is
triggered by the term "literary language" and would inquire after
a definition of literature. The answer is, for now, seemingly tautological:
a literary text is a text which demonstrates a propensity for the use of literary
language. The point of this is that the language of literature is not to be
defined by examining particular works, let alone the body of literature itself;
rather literature is to be defined as a body of texts that regularly employ,
indeed strongly favour, literary language. In this sense, the definition of
literary language is in some fashion prescriptive as well as descriptive.
The principle upon which the definition of literary language
will be based is the concept of multiple meaning, known variously as plurisignation,
polysemy, poly- or multi-valency; this text will use the term polysemy to
refer to these cases of multiple meaning, out of preference for the term's
brevity as well as its greater currency. This polysemy is not to be confused
with ambiguity, as Empson has been charged with doing in his 7 Types (Black, Perplexities 176). Rather, ambiguity is one
variety of polysemy, as is metaphor. The degree to which a text exhibits polysemy
is a measure of its complexity. Thus, a highly complex text (by definition,
then, a literary text) is characterized by the preponderance of metaphors
and ambiguities.[1] These two manifestations
and their larger conceptual framework are central to the definition of literary
language.
The connection between polysemy and literature has been
established, or at least asserted, by a variety of critics and theorists.
As Umberto Eco observes in The Role of
the Reader, the idea that a literary work has "an indefinite
reserve of meanings" is "the scope of the wave of American studies
on the structure of metaphor, or of modern work on 'types of ambiguity' offered
by poetic discourse" (54). Or, as N. Katherine Hayles plainly states,
"For someone steeped in literary analysis, it is a given that multiple
signification is a plus rather than a minus, or to use metaphors more appropriate
to literature, a story rather than a scandal" (CB 60).
Polysemy articulates a concept that is arguably, since the
New Critics at least, an essential part of a common approach to literary studies:
the idea that a work functions on multiple levels, or possesses depth (a popular
metaphor). Paul Ricoeur in The Rule of
Metaphor asserts that this kind of discourse operates on a primary
and a secondary level of meaning simultaneously; he claims that "a semantic
definition of literature - that is, a definition in terms of meaning - can
be deduced from the degree to which a discourse involves implicit or suggested
secondary meanings" (91). Eco, while admitting that "The notion
of textual level is a very embarassing one," and principally accepting
only the term "expression plane" to indicate the surface of the
text, does go along with "current theories" and claim as his ideal
text that which has the highest number of levels (Role 12-13). In both cases, polysemy appears
to be the gateway into these various levels, extending the text beyond its
apparent surface.[2]
These multiple levels contribute in part to establishing
two characteristics unique to highly polysemic texts: narrative discontinuity
and inconclusiveness. Both result from the same "problem," the proliferation
of meanings. As the number of meanings increase, the likelihood that they
all will contribute to a unity of narrative rather than breaking one down
is rather slim. As the narrative falls apart (or is never fully established)
the possibility of closure is eliminated, for closure demands an identifiable
progression of events. The inherent linearity of conventional narrative runs
counter to the nonlinearity of polysemy, which, whether one imagines it as
creating levels, depth, or systems of meaning, lends itself not to one prefabricated
structure but to many structures, contingent upon the individual reader.
There are two contemporary critical models that expressly
address polysemy and closure. Eco's model, entitled the Open Work or Open
Text, is keyed to the inconclusiveness of a text, or as he describes it, "the
contradictory plurality of its conclusions, setting the readers free to choose"
("Reply" 140). He goes on to say that the role of language in such
texts is determined by this absence of closure, and provides for it "through
the ambiguity of language and impalpability of a final sense" (140-1).[3] Roland Barthes, in his "essay"
S/Z, identifies some texts as
writerly (scriptible) and others
as readerly (lisible); the characteristics
of the former type are extreme, making Eco's suspended conclusion pale in
comparison. A writerly text is maximally polysemic; as such, "it has
no beginning; it is reversible; we gain access to it by several entrances,
none of which can authoritatively declared to be the main one" (5). Barthes
further asserts that in such a text "there cannot be a narrative structure,
a grammar, or a logic" (6). The writerly text, he concedes, is rarely
encountered. Both Eco's and Barthes' models suggest that the presence of a
quantity of meanings, some of which are bound to be contradictory, undermine
traditional textual structures.
Of course, the polysemic text is not strictly a product
of contemporary writers; as Eco observes in Semiotics
and the Philosophy of Language, the Bible is exceedingly polysemic,
as are most sacred texts (149). Lance Olsen, in his study of postmodern literature,
groups Herman Melville with Gilbert Sorrentino and Thomas Pynchon as exemplars
of "the maximalism of plurisignification" (122). Who can forget
Moby Dick, the chapter on the
whiteness of the whale? After associating the color "white" with
all manner of things terrible and beautiful, Ishmael concludes:
But not yet have we solved the incantation of
this whiteness, and learned
why it appeals with such power to the soul;
and more strange and far more
portentious--why, as we have seen, it is at
once the most meaning symbol
of spiritual things . . . and yet should be
as it is, the intensifying agent in
things the most appalling to mankind. (Melville 195)
Literary language is, however, more characteristic of written
texts than verbal communications, or of texts derived from an originally oral
tradition. Barthes describes the connection between orality and classical
texts:
There is no genre, no written work of classicism
which does not suppose a
collective consumption, akin to speech; classical
literary art . . . is a product
conceived for oral transmission . . . it is
essentially a spoken language, in
spite of its strict codification. (WDZ
49)
The reason for this is rather
obvious; the complexity of polysemic language and the effort required to process
it is not compatible with the temporality of the spoken word. There is little
opportunity for pause to address an ambiguity, make metaphorical connections.
Eco, addressing the effort required to tease out key thematic meanings, suggests
another reason why literary language is not suited to orality: ". . .
in reading literary texts one is obliged to to look backward many times, and,
in general, the more complex the text, the more it has to be read twice, and
the second time from the end" (Sem 26). This is not, of course, to suggest
that orality precludes polysemy; and if the text is recited with any frequency
(a la Homer), it has a greater capacity to support polysemy, as the audience
will have opportunities to "look back" in the text. The written
word, however, remains the ideal home for maximally polysemic language.
Having characterized polysemic texts as inconcluisve, nonlinear
and narratively indeterminate, and simultaneously the gateway to a "depth"
of meaning, we must turn our attention now to two particular varieties of
polysemy, mentioned above as common subjects of inquiry: ambiguity and metaphor.
William Empson's 7 Types
of Ambiguity (1930) was essential in establishing the preeminence
of ambiguity in literary studies; unfortunately, it is probably not as valuable
to the contemporary reader. Empson's intentionally vague definition of ambiguity
as "any verbal nuance, however slight, which gives room for alternative
reactions to the same piece of language" is too broad for most uses and
runs counter to contemporary critical application of the term (1). For example,
Paul Ricoeur argues that "ambiguity" characterizes a situation in
which a text provides for two possible meanings but can accept only one; the
two meanings, then, are in conflict, forcing the reader to decide while failing
to provide a context for such a decision (91). John Crowe Ransom, in an early
assessment of Empson's work, presses for this definition, emphasizing that
only two meanings should be
potentially acceptable (102).
This is an eminently useful distinction from polysemy, which
admits of multiple meanings without conflict, or at least without demanding
some committment from the reader before permitting further reading. However,
ambiguity is more commonly used as a synonym for polysemy, especially outside
of the literary-critical establishment. For this reason, it does not seem
sensible to try and reclaim Ricoeur and Ransom's definition; for the remainder
of this study, context will be used to identify the few situations in which
the more restrictive definition is intended or applicable. Otherwise, it is
probably more practical to read "ambiguity" as "polysemy,"
with an implication that the former generates an unease via uncertainty, while
the latter lacks any immediate logical/psychological conflict.
As suggested by the title of his work on the subject, Empson
establishes a typology of ambiguity; Donald N. Levine, with a sociological
purpose in mind, provides his own typology in The Flight from Ambiguity, distinguishing
experiential ambiguities from literary ambiguities, and those from others.
Such a structuralist approach, at least as concerns Empson, does not yield
any useful distinctions as concerns the definition of literary language. To
be sure, there are varieties
of ambiguity, ranging from the semantic to the syntactic, as any linguist
will affirm; if pressed, one might argue that the kind of ambiguity of greatest
concern to this study is semantic ambiguity. The only distinction, however,
that is absolutely necessary is one that differentiates between ambiguity
located in the text and ambiguity located, so to speak, in the reader.
The philosopher Max Black has coined the term "radical
ambiguity" to identify an ambiguity that is not resolvable by an "expert"
reader (Perplexities 183).[4]
As he observes, something may only appear ambiguous to an individual reader
because that reader lacks sufficient training or preparation; this kind of
ambiguity he describes as "subjective ambiguity" (181). This latter
type of ambiguity is not, as concerns the present approach, a genuine ambiguity;
the difficulty resides not in the text itself but in the mind of the reader.
Radical ambiguity, on the other hand, is based in the text; further, no amount
of skill or knowledge will enable a reader to disambiguate it. It is this
type that falls under the definition of literary language.
The above heading is taken from a passage in Eco's Semiotics; although in context it is used
to describe a particular difficulty involving formal semantics and metaphor,
the phrase aptly describes the state of metaphor studies at large. Metaphor
studies has become a fecund and multidisciplinary field, and within the literary-linguistic
arena, it has also become a hotly-debated subject. A casual survey of the
twenty-seven essays in the second edition of Metaphor and Thought (1993) makes clear that
the essential question, "What is metaphor?" remains unanswered.
Combining the vast amount of scholarship with the fact that few scholars seem
able to agree that they are all discussing the same subject, the highly-contested
nature of metaphor is obvious. With that in mind, it is beyond the scope of
this study to address the particularities of this debate. It is enough to
establish that metaphor is polysemic. To do this, however, some definition
of metaphor must be posited.
We shall construct a definition of a simple or "conventional"
metaphor, beginning with what Sam Glucksberg and Boaz Keysar call the "canonical"
metaphor (415). This metaphor is represented by the construction S
is P, wherein the "is"
may be explicit or implied; when the construction is read literally (instrumentally),
the statement it makes must not "make sense." The literal interpretation
of a statement may be impeded in several ways; most commonly, the statement
makes a patently false assertion. Thus, the statement
(1) Mary is a sheepherder
is not a metaphor by this
definition, but
(2) Tom is a thumb
is. This is not to say,
of course, that Mary is in reality a sheepherder; rather, the statement makes
a potentially true observation
about the world. In no reality of ours, however, might a man actually be a
thumb.[5] John R. Searle identifies three additional
interpretive impediments that distinguish this kind of metaphor: the statement
exhibits "semantic nonsense," violates "the rules of speech
acts," or violates "conversational principles of communication"
(103). In each case, the reader finds the literal reading inadequate or unacceptable.
When the statement does not literally make sense, the reader
is forced to construct a relationship (or, as Bruce Fraser has it, an analogy)
between S and P that allows the statement to become figuratively
meaningful (332). A reader who is presented with (2) and immediately rejects
the literal notion that Tom is a thumb will likely attempt to understand how Tom and a
thumb may be related; the immediate question may be, "How is Tom like a thumb?" He may then conclude
that Tom is short, dextrous, at the center of things (from the idea of having
something "under one's thumb"), or a man that stands out in a crowd.
The establishment of this metaphorical relationship is the ultimate purpose
of the metaphor.[6]
Our everyday language is rife with this "canonical"
metaphor, although much of it has passed into the realm of idiom; for example,
few of us consider
(3) Paul is a jerk
or
(4) He is fuming
to be prime examples of
metaphoric usage. One does not ponder the potential relationship between "He"
and "fuming" in (4). Therefore, it is necessary to distinguish those
metaphors that we cognitively process as such from those we do not. This distinction
is commonly made according to the terms "alive" and "dead,"
respectively.
Black has proposed, however, the substitution of categories
entitled "extinct," "dormant," and "active"
for the mortality categories (MaT 25). In this schema, an "extinct" metaphor is
one whose metaphorical nature is "beyond resuscitation," as in example
(3); it would be a difficult task to revive that metaphor in any context because
it is thoroughly idiomatized. Black's "dormant" metaphor corresponds
to "dead," with the implication that given a suitable context the
metaphoric dimension may be reactivated; example (4) is arguably of this variety,
and the condition for its reactivation may be a context bearing references
to heat or gases. An "active" metaphor, corresponding to "live,"
is cognitively processed as a metaphor; example (2) represents this type.
This study is principally concerned with "active" metaphors.
Black further proposes differentiating active metaphors
according to two criteria: "emphasis" and "resonance"
(MaT 26).[7] Emphasis, for
Black, is a measure of how necessary
a particular metaphor is to a text, along with the relative lack of equivalent
substitutes; in his words, "Emphatic metaphors are intended to be dwelt
upon." Though he seems to be emphasizing the significance of context,
the term better describes how "important" the metaphor is; in other
words, how it impacts discourse. One can certainly be affected by a powerful
metaphor even outside its original text; to say "all the world's a stage"
is to invoke an extremely important metaphor regardless of the absence of
Hamlet. Black does a better
job of defining the antithesis of an emphatic metaphor, describing it as one
that is "'expendable,' 'optional,' 'decorative,'" or "ornamental"
(MaT 26).
The criterion of resonance describes the degree to which
a metaphor lends itself to "implicative elaboration." Thus a highly
resonant metaphor has the capacity for a variety of connections to be made
between the S and P. A minimally resonant metaphor, in contrast,
allows little but a superficial appreciation of the metaphorical relationship.
With these two criteria in mind, we may observe that
(5) All the world's a stage
is a highly emphatic metaphor
because it is significant in the context of Hamlet and his character; it is also a highly resonant metaphor
because we may make a wide variety of connections between "the world"
and "a stage," well beyond Hamlet's own identification of men as
"merely players." On the other hand,
(6) He's on fire
is not particularly emphatic
because one can easily imagine other metaphors that carry a similar meaning
("He's really cooking,") or other nonmetaphorical phrases that would
do nearly as well ("He's at his best"). Nor is (6) resonant because
it is difficult to imagine taking the metaphor beyond the simple idea that
both "He" and the "fire" are energetic, move rapidly,
consume themselves, and are potentially spectacular.
Black categorizes metaphors that are both highly emphatic
and highly resonant as "strong" metaphors; this would seem to be
the type of metaphor other critics variously identify as "poetic."
As concerns this study, however, the most important aspect of the strong metaphor
remains its quality of resonance. It is the metaphor's resonance that is the
principal source of its polysemy.[8] Although all
active metaphors are polysemic, the more resonant the metaphor the more polysemic
it is.
Defining instrumental language as the negative limit of
literary language results in several antithetical assumptions, the first of
these being the idea that instrumental language is monosemic. Ideally, then,
only one meaning is present in or offered by an instrumental text. As this
ideal cannot exist, however, an instrumental text is better defined as one
that avoids or limits the presence of ambiguity and strong/poetic metaphor.
Just as polysemy has in recent history been associated with literary texts,
a genre of texts that traditionally exemplifies instrumental language is the
scientific, as Andrew Ortony observes in the opening essay of Metaphor
and Thought:
Science is supposed to be characterized by precision
and the absence of
ambiguity, and the language of science is assumed
to be correspondingly
precise and unambiguous - in short, literal.
(1)
Ortony ties the prevalence
of this notion to the rise of logical positivism in the earlier decades of
this century; it is, in fact, a semantic positivism. Donald Levine further
associates instrumental discourse with man's increasing desire to cognitively
master his world; via the "determinateness of representation" offered
by monosemic language, he is able to discipline "internal sentiment"
in favor of rationalized science (39).
In the above quotation, Ortony uses "literal"
to describe scientific language. The term "instrumental" is proposed
here as an alternative, as it better articulates the use and function of monosemic
language: "instrumental" identifies language as a tool, like a hammer,
used after a specific fashion for a specific purpose. More significantly,
it suggests that language used instrumentally is used unconsciously with respect to its nature.
One does not imagine a carpenter pondering the form and manufacture of the
hammer each time he uses it; he does not consciously study or evaluate his
swing. Literary language, as we shall see, creates such a linguistic consciousness.
Instrumental texts are by no means limited to the disciplines
of science; the attempt to use language monosemically may be observed in a
variety of other texts, including many of those classified as fiction or literature.
Eco, for example, uses the James Bond novels of Ian Fleming and Superman comic
strips as examples of what he classifies as Closed Texts (Role
8). These texts "aim at pulling the reader along a predetermined path,"
and in the examples cited here, attempt to arouse specific responses at various
points in the narrative. Whereas the Open Text is characterized by a suspended
conclusion, a Closed Text contains a particular, predetermined resolution.
The Closed Text is arrived at one of two ways, depending on the author. If
an author uses language unconsciously, assuming that his words are inherently
monosemic, he will in all probability create a text that is linearly narrative
and closed. The other possibility is an author who is well aware of the polysemic
potential of language and who actively works to limit it; this type of authorship
is more common to the sciences and other monosemically-oriented discourses,
whereas the former is typical of the genre fiction Fleming writes. Charles
Olson extends this criticism to all realist fiction, suggesting that it "neutralizes
words" (99).
As Eco opposes the instrumental Closed Text to the polysemic
Open Text, so does Barthes oppose the "readerly" text to the "writerly"
text (S/Z 4). In Barthes' case,
however, the writerly text as an ideal does not generally exist; rather, all
texts are readerly to varying degrees, which is to say that all texts bear
some authorial intention, contain some imposed limitations. As noted in the
discussion of literary langauge, the ideal writerly text is maximally polysemous
and does not have "a narrative structure, a grammar, or a logic"
(6). Inverting Barthes' definition of the writerly text, we find that his
ideal readerly text (monosemic, highly structured) is an instrumental text,
and comparable to Eco's Closed Text. Also consonant with Eco's model is Barthes'
assertion that readerly texts are committed to "the closure system of
the West" (S/Z 7).
Barthes adds a useful nuance to the definition of instrumental
language when he observes that the ostensible ideal of much of Western discourse
is to "arrange all the meanings of a text in a circle around the hearth
of denotation" (7). Denotation, carrying the authority of truth and objectivity,
is privileged over connotation, which is by definition multiple in meaning.
Instrumental language is then the language of power within Western discourse;
literary language, in contrast, abandons authority and embraces the subjective.
The implications of this distinction are significant; for now, however, we
must simply let stand the conclusion that the ideal instrumental text, with
Barthes' contribution, is both monosemic and denotative.
With the definition of instrumental language established,
we have come full circle. The continuum marked by the extremes of polysemy
and monosemy is now available as a useful scale with which to evaluate individual
texts according to their "openness" or "reserve of meanings."
Yet the operational differences between the two varieties of language have
not been fully determined. To do so, another theoretical approach is required.
Contemporaneous with the rise of the New Criticism in literary
circles was the appearance of a new science dedicated to the mathematical
study of communication; originally called communication theory, the field
is now better known as information theory (Pierce 1). Just as the New Critics
called attention to the structure and function of language in literary contexts,
information theorists began to examine closely the core processes of communication.
Given that art and science have often demonstrated parallel concerns and roughly
simultaneous developments, it is not surprising to see a common interest arise
in a closer examination of communicative acts.
The lens of information theory permits a new approach to
the differentiation of literary and instrumental language; more significantly,
it reveals their core communicative functions. Information theory also provides
us some new tools with which to judge the complexity of texts as well as individual
polysemies. In general, this theoretical approach will serve a supplementary
role, offering a variant understanding of the creation of textual meaning.[9]
It is necessary to begin with a definition of a key concept,
"entropy." As Hayles observes, the word has variant meanings according
to context as well as discipline (CB
38). Its origin is in physics, where it is associated with the second law
of thermodynamics; the second law observes that in closed systems heat energy
will tend to dissipate. Entropy, in this sense, is a measure of this dissipation.
For example, a cup of hot coffee left upon a table will gradually lose heat
until it reaches room temperature. The coffee is a "closed system"
because it cannot replenish its own energy; outside intervention (in this
case, human agency) is required to restore lost heat. The coffee's entropy
increases as its temperature decreases.
A more general way of approaching entropy is to conceive
of it as disorganization. In order for a system to function, it must be organized
and ordered; over time, however, the tendency is for disorganization to increase,
ultimately leading to the disappearance of the system altogether. A maximally
entropic state is one of undifferentiation; all things are constant and identical.
Bridging the entropy of thermodynamics and the study of
communications involves equating communicative acts with organization; information,
in a sense, also becomes equated with energy. Norbert Wiener, one of the pioneers
of information theory, offers several lucid observations that provide a radically
different way of looking at messages:
Messages are themselves a form of pattern and
organization. . . . Just as
entropy is a measure of disorganization, the
information carried by a set of
messages is a measure of organization. In fact,
it is possible to interpret the
information carried by a message as essentially
the negative of its entropy,
and the negative logarithm of its probability.
That is, the more probable the
message, the less information it gives. Clichés,
for example, are less
illuminating than great poems. (21)
An important issue must be addressed here. In the context
of information theory, "information" does not equal "meaning."
It is better understood as a measure of potential meaning; thus a highly informative
message contains a larger array of possible meanings than a minimally informative
one. Returning to the definition of literary language, this array of meaning
is equivalent to multiple meaning. We may then define polysemy as a highly
informative message, as well as the reverse, so that a highly informative
message is the definition of polysemy.
This definition gives rise to a seemingly counterintuitive
observation: scientific texts, principally instrumental in nature, are less
"informative" (in this sense) than literary texts. If we consider
it fully, however, the reason becomes clear. A monosemic text, a minimally
informative text, carries (perhaps) only one meaning; the purpose of a scientific
treatise may be to identify and explain just one idea. Purely instrumental
language nearly guarantees that the specific meaning will be fully communicated.
Literary language, on the other hand, with its polysemy, will communicate
a number of meanings in addition to the intended meaning. Potentially, the
specifically intended meaning may disappear into the text, lost among the
various competing meanings available to the reader. Instrumental language
may not communicate very much information, but it is clear as to what it communicates;
literary language communicates infinitely more but the meanings are not all
of equal weight nor are they all intended. Recalling an example from the section
on metaphor,
(2) Tom is a thumb
and adding a new one,
(7) Tom is short
it is obvious that while
(2) is uncommon and less probable as compared to the relatively familiar (7),
the latter is much clearer as to its meaning. The metaphor in (2) allows for
a variety of meanings, as demonstrated earlier; it potentially communicates
much more than (7), but it may also fail to communicate (7) as one of its
meanings. Therefore, if one absolutely must communicate the notion that Tom
is of diminutive height, it is more appropriate to use the instrumental statement
(7) rather than the metaphorical (2).
Information theory offers an alternative way of looking
at polysemy by using the term "noise" to describe the presence of
multiple meanings. Noise suggests something not immediately implied by polysemy:
the unintentionality and unpredictability of multiple meanings. Though the
term is generally considered pejorative, it must be remembered that in an
engineering context (home to information theory) predictability and regularity
are ideals; the celebrated richness of a polysemic literary text is antithetical
to these goals. Returning to the above example, (2) is too noisy a message
to effectively communicate (7). Information theorists speak of "channels"
through which messages are communicated; knowing this enables us to understand
William Paulson when he writes, "Insofar as literary texts are both communicative
and ambiguous, they are noisy channels" (CaO 42-3). Instrumental texts, conversely,
are ideally noiseless communication channels.
Another important aspect of information theory addressed
by Wiener relates to probability. Entropy (disorganization) corresponds to
an absence of information in a message; the more probable the message is,
the less information it carries. The most probable message, then, is the most
entropic. From this we may conclude that Joyce's Finnegans
Wake is arguably the least probable and most informative
"message" ever communicated; in contrast, the polite "hello"
of the hallway is both a highly probable message and an ultimately meaningless
one (excepting, of course, for the purposes of social cohesion). In this light,
complex texts are less probable (common) than simple ones, and literary language
is less probable than instrumental language.
If we pursue this to its logical end, however, we would
have to accept that the most informative message is one that is completely
random. We feel intuitively that this cannot be the case. Take, for example,
the following statement, which was created by pulling the first word appearing
on pages 17, 27, 37, 47, and 57 of Pierce's introductory text:
(8) Square even wish events the.
This semi-random combination
of words should yield more information than any conventionally created sentence.
In one sense, it does: as there is no obvious meaning immediately communicated,
we are forced to explore a wider range of possible relationships between the
words. Square may be a proper name; on the other hand, (8) may be projecting
a world in which squares have sentience and the capacity for wishing. "Events,"
apparently, are happening, and Square/square arguably has some feeling about
them. It would not be difficult to go on at some length.
This minor example, however, makes clear that a maximally
informative message suffers from an overabundance of meaning; it is so improbable and lacking in conventionality
that we cannot grasp some core intention, a locus of ideas. Given this situation,
we can better understand why polysemy is labelled "noise" by information
theorists.
Pierce plays around with another approach to generating
a text sequence, though in this particular experiment he limits the randomness;
without going into any detail as to his method, one result it generated is
offered here:
(9) It happened one frosty look of trees waving
gracefully against the wall
This less-random statement
strikes one as much more comprehensible than (8), perhaps even somewhat "poetic."
As Pierce observes, "Poor poets endlessly rhyme love with dove, and they
are constrained by their highly trained mediocrity never
to produce a good line" (263-4). In other words, their messages are highly
probable, to the point of being cliché. In some cases, it seems, a randomized
approach generates superior messages to a conscious human one. Comparing (9)
against (8), we increase the probability of the message and decrease its information
content in exchange for greater readability and understanding.
Returning to the issue that necessitated these examples,
we find that although maximum information is communicated by a maximally improbable
message, this condition is not necessarily desirable. Taking this into account,
information theory mathematically describes the ideal message as one that
is fifty-percent probable. Hayles translates this into layman's terms: "Maximum
information is conveyed when there is a mixture of order and surprise, when
the message is partly anticipated and partly surprising" (CB 53). Put another way, for a message to
communicate effectively it must not be so improbable and unexpected that its
recipient does not know how to process it, nor should it be so common and
expected that it is not worth processing.
This fifty-percent point is also useful when applied to
specific polysemies. For example, in the discussion of metaphor it was resolved
that a strong metaphor is one that is highly resonant and therefore highly
polysemic. Yet we can conceive of a metaphor such as
(10) Tom is a widget
that is improbable and polysemic
as well as extremely difficult to process. One would probably not consider
(10) a useful or successful metaphor. In the case of metaphors, then, a metaphor
that is fifty-percent probable is one that will communicate ideally. Without
belaboring the point, we may observe that this holds true for ambiguity as
well; too many conflicting potential meanings make it impossible to usefully
entertain any.
Of course, probability cannot be measured for specfic messages,
be they texts or metaphors, for a wide range of reasons unsuitable for discussion
here. We cannot know whether or not something measures up to the fifty-percent
criterion. This does not, however, diminish the inestimable usefulness of
this scale or information theory as a whole. Regardless of its direct applicability,
it provides us with a new way of thinking about communication on all levels.
When we begin to understand messages in terms of being common or uncommon,
and directly relate that to information content, we supplement the traditional
linguistic or literary-critical approach. We are also better able to estimate
the relative value of various texts across genre and discipline.
Information theory and the fifty-percent criterion suggest
an interesting parallel in the literary language-instrumental language continuum.
If the ideally communicative message is one that is both moderately anticipated
and moderately surprising, then arguably the ideally communicative text is
one that is half-literary and half-instrumental in nature. If we consider
the instrumental language used in a text as the structure, the foundation
upon which literary language is arrayed, we have a useful model for the arrangement
and ultimate processing of the ideal text. The instrumental language functions
as a baseline against which literary language plays; it also provides a monosemic
thread for the reader to hold to in the midst of the polysemic chaos. In light
of the fact that neither literary language nor instrumental language exist
as absolutes, certainly not as ideal texts, this even proportion gains credence
as a realizable formula.
However, it is too soon for this conclusion; we have yet
to incorporate the reader. In preparation for that, it is probably useful
at this point to chart the various attributes specific to literary and instrumental
language:
Literary Language
polysemic
radical ambiguity
strong/poetic
metaphor
connotation
open/indeterminate
less probable
less entropic
more information
more potential
messages
noisy
Instrumental Language
monosemic
unambiguous
conventional/extinct
metaphor
denotation
closed/closure
more probable
more entropic
less information
fewer potential
messages
clear
[1] Another term which carries a useful metaphoric dimension is "density," whch suggests scientific parallels in terms of weight and material composition.
[2]I am hesitant to endorse terms such as "depth" and "levels," implying as they do a hierarchical structure which presumably exists across the work. I would prefer instead to propose the conceptual framework embodied in the term "systems of meaning," which suggests a multiplicity of meanings dependent not upon a contiguous level of some kind but on the presence of other meanings which have one or more shared semiotic characteristic(s).
[3]This provides for an interesting question; one might wonder to what degree a text may be polysemic and yet maintain the appearance, if not the actuality of a determined conclusion.
[4]This term seems to share company with other contemporary "radicalities" posited by Ihab Hassan, who identifies "radical innocence" as characteristic of the postwar novel and "radical irony" as characteristic of the postmodern novel.
[5]This excepts, of course, the possibility that someone has named his or her thumb, in which case contextually we would comprehend that (2) is potentially true.
[6]Identifying the nature of this relationship and its generation in the mind of the reader is the site of much of the current disagreement over metaphor.
[7]Don R. Swanson also differentiates more powerful metaphors from lesser ones by the number of connections the metaphorical relationship allows for, but where Black uses a second criterion of emphasis, Swanson focuses on "the speed or the suddenness" of the discovery of these connections as a distinguishing characteristic (163).
[8]A metaphorical statement is polysemic in another way, as well: it has both a literal and metaphorical meaning.
[9]As is the case with metaphor, the version of information theory presented and used here is extremely reduced in scope; only the most essential and relevant elements have been retained. Unlike metaphor, however, this reduction is not the result of competing theories or inadequacy of definition; rather, it is due to the simple fact that information theory, based in mathematics, rapidly turns unwieldy for the layman.