TIMES OF THE SIGN
Lectures on Discourse and Temporality in Recent Linguistics

By

Paul J. Hopper
Carnegie Mellon University

Humanities Institute, Santa Barbara
Seminar Series on Human Temporality, 1989-90
October 16-21, 1989



I. TEMPORALITY AND MODERN LINGUISTICS

In these lectures I will be using the terms "postructuralism" and
"postmodernism" in ways that will at times seem interchangeable. In
spite of some disagreement amoung users, it seems that generally the
term poststructuralism refers to an ill-defined set of approaches to
language that have emerged specifically in reaction to holistic
paradigms based on highly general, abstract, logical, a-priori
structure. The term "postmodernism," on the other hand, involves a
more general attitude of incredulity toward what Lyotard calls
'metanarratives' that can be appealed to in resolving the hopeless
complexity and moral ambiguity of communication in the modern world.
Lyotard speaks of a 'crisis of narratives' and a turn to fragmented,
'multiple language games,' and to 'institutions in patches.' Marcus
and Fischer summarize the postmodernist attitude by saying:

The key feature of this moment, then, is the loosening of the hold
over fragmented scholarly communities of either specific totalizing
visions or a general paradigmatic style of organizing research. The
authority of 'grand theory' styles seems suspended for the moment in
favor of a close consideration of such issues as contextuality, the
meaning of social life to those who enact it, and the explanation of
exceptions and indeterminants rather than regularities in phenomena
observed--all issues that make problematic what were taken for granted
as fact or certainties on which the validity of paradigma has rested.
(1986:8) 

Poststructuralism, then, can be seen as a single symptom of the more
general phenomenon of what Paul Rabinow has called the 'crisis of
representation' (Rabinow 1986) in postmodernism.

Linguistics in its current state nicely reflects this crisis. Should
language be 'represented', and if so, how? Since the generative
grammar paradigm is founded on the very concept of a mentally
represented system,it is vulnerable to postmodernist attacks on
representation such as that of Richard Rorty, who argues that
epistemology as the study of mental representations arose in a
particular context, that of 17th century Europe, and triumphed
eventually through its links to the professional claims of one group,
19th century German professors of philosophy. "The desire for a theory
of knowledge is a desire for constraint--a desire to find
'foundations' to which one might cling, frameworks beyond which one
must not stray, objects which impose themselves, representations which
cannot be gainsaid" (Rorty 1979:315).

The eventual concern in these lectures is with temporality. But to
start with at any rate I shall not be much concerned with the
linguistic representation of time as such, but with a more general
problem about the relationship between temporality and views of
linguistic structure. The question of temporality is not easily
distinguished from more obvious questions that concern linguists. We
need only think of the Saussurean dichotomy synchronic-diachronic to
recognize how profoundly the foundations of our science are linked to
time and the apparent opposition between system and temporality. And
so I shall also argue that certain recent developments in linguistics
which are drawing the field further away from the standard
representational paradigm can be viewed as simultaneously assimilating
the new perspectives to postmodernism in other fields, and that it is
perhaps time for linguists to recognize this and to be more receptive
to the fundamental conceptual changes and reassessments of goals
occurring in neighboring disciplines.Of all the reasons for suggesting
this, none is more cogent than the self-conscious political stance
recently adopted by generative linguistics in which the
representational paradigm has been elevated to the level of an
official dogma. The most glaring recent instance of this is the
editorial policy statement (Thomason 1989) by the editor of Language
(the field's flagship journal and the official organ of the Linguistic
Society of America) declaring that articles in certain subfields,
designated as 'central', which did not make reference to current work
in the generative paradigm were unlikely to be accepted for
publication. Typical rationalizations for this hegemonising claim do
not appeal to a need for conservatism or caution about too hastily
abandoning the traditional paradigm, as one might imagine, but instead
insist on viewing generative grammar as in the avant garde, and on
presenting competing approaches as pre-modern or 'unscientific'. There
appears, then, to be a need to force formal linguistics into a
defensive posture and to demonstrate its essential synchronicity with
structuralism, 'New Criticism', and other such movements in cognate
fields that are under assault.

James Edie (1972/77) has written an interpretation of Husserl in which
he links the Chomskyan notion of grammar to the phenomenological
project of explanation through a precise description of "structures of
consciousness" arrived at by introspection. Edie notes that Husserl's
quest for a pure grammar resumes a philosophical theme going back to
the Middle Ages and even before. He cites from the Logische
Untersuchungen:

  Language has not only physiological, psychological, and cultural-
  historical, but also a-priori foundations. These last concern
  the essential meaning of forms and the a-priori laws of their
  combinations and modifications, and no language is thinkable which
  would not be essentially determined by this a-priori. [cit. Edie
  1977:139-140]

According to Husserl, then, it is the grammatical which is the 'first
level of logical reflection'. This most abstract level of grammar,
logically preceding experience of the world and empirical grammar,
that is, the specific grammar of a particular language, is therefore
held to structure and encompass both individual grammar and perception
of the world. Edie defines the task of modern linguistics as that of
taking up Husserl's early project of an a-priori grammar. Thus Edie
observes that although the idea of a universal a-priori grammar has
its origins in medieval thought, Husserl's project of 'pure logical
grammar', which Edie praises as 'probably the most recent full-scale
proposal in this area from the side of philosophy', has been taken up
by Noam Chomsky in 'a program for the study of grammar which, if it
were to succeed, might seem to justify the earlier intuitions of
rationalist philosophers and to give a new grounding to this ancient
quest.' This 'pure logical grammar', Edie notes, is a formal grammar;
for the Husserlian term 'pure' is synonymous with 'formal'. 1137

If Edie is right, and Chomskyan formal linguistics has its
intellectual roots in Husserlian phenomenology, it would follow that
the premisses of formal grammar are only as secure as those of
phenomenology itself. If postmodernism can be said to have any
foundational texts, they surely include those by thinkers from
Heidegger to Derrida who have questioned and in their own ways
overturned or at least radically reinterpreted these premisses. Yet
formal linguistics has shown itself to be surprisingly robust, almost
one might say to be leading a charmed life, having passed virtually
unscathed through the postmodernist revolution and the
'deconstruction' of the texts of phenomenology. I shall not dwell on
the reasons for this invulnerability, but will note that it has much
to do with the extraordinary prestige of Noam Chomsky, whom even such
prominent thinkers as Pierre Bourdieu and Henry Staten seem reluctant
to attack. Bourdieu, for example, credits Chomsky with "affirming
their dependence of the structure of linguistic expressions relative
to their use and functions and the impossibility of making any
inference from analysis of their formal structure" (Bourdieu 1977:26).
(It should be added, though, that Bourdieu is elsewhere mildly
critical of Chomsky.) Yet Chomsky affirms this independence by the
very strategies of abstraction and reduction that Bourdieu's own work
was written to oppose! And Henry Staten, in his justly influential
comparative study of the thought of Wittgenstein and Derrida, declines
to engage Chomsky head-on at the very place where such an engagement
is almost obligatory for the development of his argument, on the
grounds that he believes Chomsky's discourse is scientific rather than
philosophic:

I believe Chomsky's version of implicit rules is plausibly tied to a
'scientific' research project, and that it is not vulnerable to
refutation by a purely philosophical critique... Chomsky's project is
not privileged merely through his claim to be scientific, but because
his text has been minutely and rigorously tied to the text of
scientific investigation... (Staten1984:178)

Probably almost any linguist could supply an ecdotal evidence of the
rudimentary level of understanding of the notion of grammar among
theorists in other fields. In my experience, many even quite
sophisticated social scientists and philosophers continue to regard
grammar as a set of established and well-known facts about a
language.One memorable example for me was when the subject of grammar
came up in a conversation I had with Hans-Georg Gadamer, during his
last visit to the USA, and he told me, if I remember correctly, that
he had never been good at grammar in school and had never taken any
interest in it philosophically! A rather common attitude is that
formal (i.e.,Chomskyan) grammar has presented facts about language
which can be interpreted in their scientific context. These facts may
be accepted or rejected in toto according to transcendental arguments,
but there is rarely any question about the empirical basis of the
"facts" themselves, how they were collected, what premisses under lay
their arrangement and analysis, and so on,

In other words, linguists who reject formal grammar and look for
support from post-modern theorists should not necessarily expect to
find either a familiarity with or a direct encounter with technical
linguistics. There are some exceptions, notably by Derrida; his book
Limited. Inc. could be described as a deconstructive analysis of
speech act theory, in particular of John Searle's version of the work
of Austin. The essay "The Supplement of Copula: Philosophy Before
Linguistics" (Margins of Philosophv 175-205), on Benveniste, should be
read by all proponents of 'cognitive grammar' and 'cognitive
semantics', particularly, one might say, by those who claim to have
been influenced by 'Continental' philosophy (Lakoff and Johnson 1 980:
xi). And so on. But on the whole, the quotations which I have read
from Bourdieu and Staten point to a rather superficial knowledge of
linguistics among post-modern theorists. Linguistics is in fact
generally equated either with mid-20th century Continental
structuralism or with the Chomskyan school, as if there were
uniformity of opinion and method in the field. Not uncommonly, either
of these schools of thought is liable to be referred to in
postmodernist writings as "linguistics" tout court. All too often the
laborious deconstruction of the foundational texts of linguistics,
important as it has been for the working out of method and the
exemplification of ideology, turns out to be irrelevant to linguistics
because these texts have already long been by-passed in linguistic
practice.

In recent decades, however, a non-canonical linguistics has been
emerging which in some important respects is bringing about an
assimilation of the ideology of linguistics to what might be called
the post-modern episteme. This assimilation is not to be seen as a
form of appropriation comparable to the earlier deliberate and
conscious appropriation of linguistic structuralism by anthropology
and literary theory. There explicitly formulated principles, the
everyday working methods of linguists, were coopted wholesale by
practitioners of other fields. Post-modernism is not a coherent set of
methodological postulates to be taken over and, so to speak, 'tried
out' on linguistics, but something more fragmented, more like an
attitude of mind. Stephen Tyler, discussing the practice
of-ethnography after the appearance of such basic postmodernist texts
as Derrida's Of Grammatology, says "Post-modern ethnography builds its
program not so much from their principles as from the rubble of their
deconstruction" (Tyler 1986:131). This fragmentariness is a natural
by-product of the critique of large, all-encompassing systems that
characterizes postmodernism.

The assimilation is therefore to be characterized rather as a
convergence of beliefs, postures, and practices which is occurring
gradually and without directly attributable influences. It is
occurring in spite of resistence from within and without linguistics.
(A recent attempt to thematize the assimilation [Hopper 1988] has
provoked an extremely sharp defensive response by a minor theorist of
postmodernism [Beck 1989, to appear, with a reply by Hopper].) This
means only, of course, linguistics is gaining the possibility of
linking with comparable post-modern and poststructuralist trends in
other fields. It does not mean that linguistics is or should 'go
post-modern', only (and I would emphasize this) that as it
increasingly moves beyond structuralism it has the potential for
letting itself be influenced, I believe in a very beneficial way, by
new habits of thought which are affecting fields traditionally
considered cognate, and thereby reciprocally opening up its own
thought to other fields. Linguistics deserves better than to be
presented to practitioners of other disciplines as Chomskyan syntax
and Lakoffian metaphoritis.

If I were to characterize the difference between structuralist and
poststructuralist approaches to language in the simplest possible
ways, I might say that structuralism totalizes and de-temporalizes
discourse by viewing texts retrospectively in terms of hierarchical
components. It is the same strategy as we see operating with the
sentence in the Chomskyan paradigm. The structure of the sentence is
assumed to pre-exist its utterance, and indeed the utterance is in a
bizarre sense quite redundant--all the structural work has been done
beforehand. But the important thing about structuralism, it seems to
me, is not so much its method, but the attitude towards texts that
that method presupposes.The structuralist has a bird's-eye view of the
entire discourse which is at issue, be it a sentence or a longer text,
and can move backwards and forwards in it at will, constructing
hierarchies and deep structures that supervise the entire text, and
moving to and fro between all texts. Poststructuralist thinking, on
the other hand, understands discourse in its temporal unfolding.
Consequently for the poststructuralists discourse is always seen in
medias res, in the midst of the social scene of its creation.

Structure, then, does not present itself to us as a fixed, uniformly
available code, but is ad hoc, disparate, and worked out 'on stage' in
an improvised fashion. Such an acting out occurs in time, not in
puppet-like response to some higher-level organizing principle.

A familiar theme begins to emerge from the distinction between
structuralism and poststructuralism. The kind of global access, the
totalizing perspective implicit in the structuralist's approach to a
text is indeed a possibility, but largely with written texts, which
permit to an unlimited degree spatial arrangement, diagrammatization,
what Jack Goody has called 'back-tracing'. For the generativist school,
in fact, and implicitly for many other schools of linguistics, the
spoken dimension of language is actually irrelevant. But it would be
wrong to suggest that for generative grammar writing and speaking are
equally irrelevant. In spite of its claims, generative linguistics is
not modality-neutral. Its whole analytic technique, and its
representational view of language, reek of the scriptorium. Most of
its sentences are simply impossible in spoken discourse, e.g.:The
excited astronomer looked at the new star with his telescope (found in
a recent textbook). (Indeed, as Wallace Chafe noted in a comment after
the lecture, they are impossible in written discourse also!)

In other words, the polarity structuralism/poststructuralism has every
appearance of being one more manifestation of the oral/written
polarity that has loomed so large in recent decades. The founders of
modern linguistics of course stressed their allegiance to spoken
language at every opportunity, with the self-righteous vehemence of a
presidential candidate proclaiming his devotion to family values. But
I think it is fair to say that 'spoken language' for them came down to
written language read aloud. The explicit interest in 'spoken
language' in the sense of 'orality' is owed to quite recent times, and
its gradual assimilation to mainstream linguistics is the work of a
rather small number of linguists--many of them in this room. The
dependence of modern linguistics on writing has been noted more than
once. Roy Harris, in The Language-Makers, goes so far as to say that
"the real 'discovery procedure' of modern linguistics is: Assume that
standard orthography identifies all the relevant distinctions until
you are forced to assume otherwise." He later continues:

  Suppose we strip away this superficial garb of the sentence, what
  lies beneath it? Something which must have all its words in place,
  their order determined, their grammatical relationships established,
  and their meanings assigned--but which simply lacks a phonetic
  embodiment: a string of words with the sound switched off. In short,
  a linguistic abstraction for which there is only one conceivable
  architype so far in human history: the sentence of writing. (p. 18)

The study of written texts as such is of course derivative of the
interest in orality. Once the distinction had been formulated, and it
is still far from precise, it became just as important to study the
written language as the spoken, though of course the context had
changed, since writing was no more to be seen as a degenerate form of
speaking than the reverse. Yet the study of written texts is not
unaffected by the changed attitude toward texts and language brought
about by the 'orality movement'--we cannot have structuralism for
writing and poststructuralism for speaking. We are now venturing far
into the territory of postmodernism, with its themes of writing,
intertextuality, and indeterminacy. It is here that linguistics can
benefit most from the insights of literary theory.

In a very early document of the Orality movement, Walter Ong noted
aphoristically: "Orality traffics in previous knowledge" (1965). He
was referring to the accumulation of lore that accompanies oral
knowledge, the quite literal accrual of bits and pieces of knowledge
and ways of doing things that are an inseparable part of orally led
existence. It is the conceit of literacy, perhaps, to see something
inferior in these prefabricated legacies of our own and others'
biographies, and to think of them as incidental to language rather
than its core. In the process of purifying language grammarians have
set aside the ready made and have preferred to act as if each
utterance were newly generated. This fiction of course is only
possible to the literate mind. As Ong and others have shown us, only
the literate mind can conceive of utterances out of their context, and
is equipped with that kind of ability to distance itself from language
that enables logical systems to be developed and applied to it. For
the oral scenarios in which speech occurs there is in fact little room
for a readymade, totalizing grammar. Speaking takes place 'on the
fly', as Chafe has noted. Formulaicity explains this improvised
quality better than does any notion of the mental representation of
grammatical structure. The short passage of spoken English that is
being circulated (see Appendix) consists almost entirely of formulas
and formulaic expressions. Even the glue that joins them, the function
words, could be said to be formulaic, in the sense that they provide
learned ways of providing for formulas to be used. And finally, let us
go the extra step and say that even individual words (like 'handicap',
'disturbed','relevant') are formulas, in that they occur in the
context of formulaic sequences. But formulas very often correspond to
what Chafe has called 'idea-units', short phrases centered around a
single idea, timed for a single breath pulse, and responsive in their
use to the contents of short-term memory. Formulas are in other words
grammaticalized idea-units.

This thought then at once raises the question of where the
combinations that constitute formulas come from. The answer is: other
formulas. They are changed by a number of processes, for example
grafting and blending, brought about partly by ingenuity in response
to a new need, and partly by misunderstanding of an old formula whose
context (biography) has been lost. There are crowds of examples of
such malapropisms and folk-etymologies, and though they sound quaint
and anecdotal, they are more revealing about language than is
sometimes believed. An elderly colleague recently told a committee he
had gone through the document in front of them 'with a fine
tooth-brush.' Presumably 'fine-toothed' in the sense of 'having its
teeth close together' is not sufficiently current to resist the
overwhelming relationship comb-brush-toothbrush. Meaning has little to
do with this process--'I heard it through the grape-vine' was surely
once 'through the grap-line', a line passed between two sailing ships
to keep them in touch during rough weather.

Speech is produced through a stitching together--to use the Ancient
Greek metaphor of rhapsodein--of previously known material. Oral
speech has structure, but it is structure re-conceived. It is what
Derrida calls diffˇrance - structure spread out through
time,provisional, deferred, continually being remade in the social
scenes of its enactment. Its inhabitable dimension is not "mind" but
history and interactive discourse. The structure of oral speech is not
homogeneous,since the bits and pieces of previous knowledge out of
which speech is composed, the bricolage of discourse, is disparate,
heterogeneous, learned and reinforced and redistributed in as many
different contexts as there are verbal experiences. Among the members
of a speech community these verbal experiences will overlap in
important enough ways that expressions that have been worked out and
proven their value in one social context will be applied to another.
The expressions themselves then become part of the common pool of that
speech community's linguisti resources, and the use of them becomes a
significant criterion for defining the in-members of the speech
community vis-a-vis the Others.

The British novelist Jack Lodge, who in the 1970's spent a year in
Northern California, has described exactly this situation in his
article on Psychobabble, a form of slang which originated in Marin
County. Its users are wealthy, young, impressed by easy-talking
therapists, and of course faddish. Cyra McFadden's picture-novel The
Serial is essentially a lexicon which preserves the genre in much the
same way as the Iliad and the Odyssev preserve the formulas of Ancient
Greek oral literature. The core of Psychobabble consists predominantly
of a small number of verbs combined with adverbs and prepositions, and
occasional nouns. But nouns do not figure in the jargon that much. For
the speakers of Psychobabble are not much concerned with nouns in the
sense that outsiders would know them; when they interact with the
world of objects it is largely through brand-names, highly
particularized things that are as much symbolic of the in-group as the
language. Again, from The Serial:

  Harvey made a lot more money now...but they spent it on things they
  hadn't known existed ten years ago: Rossignol Startos and season lift
  tickets at Squaw; twin Motobecane ten-speeds; Kate's Cuisinart, which
  did everything but put the pate in the oven; Stine graphics;
  Gumpoldskirchner and St. Emilion (Harvey had 'put down' a case in the
  vacuum cleaner closet); Klip speakers and the top-of-the-line Pioneer
  receiver; Brown Jordan patio furniture; Dansk stainless and Rosenthal
  china; long-stemmed strawberries and walnut oil from the Mill Valley
  Market; Birkenstock sandals and Adidas (Kate didn't actually jog yet,
  but she was reading The Ultimate Athlete.)...

The striking thing about Psychobabble is that it consists almost
entirely of prefabricated phrases. Here are a few of them:

- he's into [bonsai trees, belt-buckle casting, meditation,
  and Zen jogging]
- dropped out
- where you're at
- where you're coming from
- get behind it
- get it together
- get it centered
- get it on with
- get off on
- lay a trip on
- hang loose
- hang tough
- spaced out
- grossed out
- weirded out
- blow away

That many of these formulas are now part of general spoken English
testifies of course to the widespread covert envy of the Marin County
lifestyle in the rest of California and elsewhere. But let us accept
that they were once characteristic of a particular social group living
in the Peninsular to the north of San Francisco. As Lodge notes,
Psycho-babble has provoked criticism as being just a hoard of clichˇs
representing prepackaged thoughts, ready-made, functioning to release
their users from the burden of creatively and authentically
extricating themselves from their constricted lives. He quotes from
several of the critics of the style, including R. D. Rosen, the
originator of the term itself. According to Rosen, Psychobabble is

  a set of repetitive verbal formalities that kills off the very
  spontaneity, candor, and understanding it pretends to promote. It's
  an idiom that reduces psychological insight to a collection of
  standardized observations, that provides a frozen lexicon to deal
  with an infinite variety of problems.

Note now that if we drop Rosen's prejudicial rhetoric, he is in effect
saying that Psychobabble is language. It is a 'set of repetitive
verbal formalities' (i.e., formulas), a 'collection of standardized
observations', and it 'provides a frozen lexicon to deal with an
infinite variety of problems.' Isn't this what language always does?
One wonders what on earth a language would look like if it instead had
a separate lexical item to deal with every separate problem.

Of particular interest here is what we should think of as the emergent
semantic structure of Psychobabble. Its vocabulary is strongly
spatial. It makes use of a small handful of verbs like 'get', 'hang',
'be', and a variety of prepositions ('on,' 'off,' 'into') and spatial
nouns of very general meaning that are tightly collated with
particular expressions('space,' 'pits,' 'wall', 'head'). This
relentless spatiality of the phrasal lexicon of Psychobabble in the
domain of personal relationships and feelings might to some raise
questions about cognition and the 'representation' of 'concepts' like
'mind' and 'other'. It appears as if social relationships and
psychological states are metaphorised as a static network in which the
ego stands in a spatial relationship to theirs and to various aspects
of his or her own psyche. These states and relationships could
therefore be said to be mentally represented, and to govern our
actions and perspectives from a fixed vantage-point.

I would suggest that there are alternative ways of seeing even the
most persistent metaphors. It is true that once a tendency towards
patialization has been begun it can become pervasive and influence
other directions of metaphorization; but this is no more than always
happens with verbal formulas, which always become contaminated and
blended with other formulas that are contextually adjacent. Once a
prime metaphor comes into use, others may cluster around it. If I find
that my ways of talking about 'life' consistently make reference to
words and expressions having to do with travelling, this doesn't mean
that my mind in some sense contains a representation of life as a
journey, but rather that a couple of very broad natural associations
like 'beginning' and 'end' attract a small network of more specific
formulas taken from other language games in which there is talk of
beginnings and ends. And these formulas will not always be appropriate
ways of talking about life, but will only be appropriate in certain
circumstances, circumstances that have been singled out by the
linguist precisely because these metaphors figure in them. And somehow
defenders of metaphor as a foundational theory look only at the end
product, what looks in retrospect like an equi-valent set of metaphors
magically clustering around a single story, and perhaps even repeating
themselves across different languages. Once again the reduction of
language and thought to metaphor involves hypostasization as a unitary
concept of material taken from functionally diverse contexts. It also
involves a view of language that segregates speech from its social
contexts and throws everything worthy of description and explanation
into the individual psyche. One suspects again that the kind of
representation of language implied in this attachment to concepts and
metaphors presupposes writing and literacy.

In this lecture I've suggested that there is an ineradicable
relationship between structuralist theories about language and the
deeply sedimented linearity and spatiality that literacy entails. I
can't help thinking there is more to this than simply getting the
right handle on language. Every time you try to think about written
language, questions of power and authority are raised. Written
languages that are fixed are socially privileged, and the
institutional settings served by them are hegemonizing. Consequently
even when they are spoken, perhaps especially when they are spoken,
written languages continue to mediate and symbolise power and control.
Frequently the very names by which they are known are suggestive of
power, privilege, and authority. The Greek standard is known as
kathaverousa, the pure language. Standard German is called
Hochdeutsch, or 'High German'. That this term ('High German')
originally referred to a geographical region is of course purely
coincidental; its reference is nowadays entirely social, the language
of the 'higher-ups'.So the supposed neutrality of the written standard
language is as phony as that of most institutionalized phenomena for
which political neutrality is claimed.

Incidentally, it is not merely written languages that are privileged,
but alphabetic written languages. It has been a common conceit of
Western linguists that alphabetic writing systems are the logical end
of a long process of rational evolution whereby the linguistic units
represented by writing are progressively (!) reduced in size and hence
increased in flexibility and usefulness. Word signs gave way to
morpheme signs, which gave way to syllable signs, to consonant signs,
and finally to phoneme signs. So a few Western languages whose writing
systems were inherited more or less directly from that of Greek now
enjoy the benefit of having the symbols of their writing coincide
exactly with the minimal phonemic units of their languages. It seems
unkind now to suggest (Harris 1988;Linell 1979) that this story might
have to be told in reverse, that the linguist's 'phoneme' might turn
out to be nothing but an epiphenomenon of a grammatical tradition
which happens to have emerged in an alphabetic culture. It is surely
no coincidence that Saussure, searching for a way of explaining the
relational nature of the linguistic sign, could find no better example
than letters of the alphabet.

If language is to be appropriately recontextualized, I would suggest,
it must be dislodged from its status as a meta-narrative and seen in
the light of such Wittgensteinian concepts as 'language games,' and
the general resituating of speech into the contexts in which it has
its immediate origin. I shall return in a little more detail to this
point tomorrow. Speech thus conceived owes its being to a displacement
that is lateral (temporal, historical) rather than to one that is
vertical (metastructural, synchronic). The study of language is
primarily the study of the social and, inevitably, political scenes of
its creation and, secondarily, the ways in which these scenes become
interiorized in the individual's psychology. Speech is not to be seen
as what Kenneth Burke called the 'linearization of essence', but
instead owes the way it is to the way it was before, linearity without
essence and defined solely by the past. ('Only that which has no
history can be defined'-Nietzsche.)

I have tried to suggest in this lecture that the language "anschauung"
of formal linguistics has reversed this directionality, hypostasising
grammar and structure and treating them as a-prioris, as necessary
preconditions for discourse. It has claimed as its project not the
study of discourse, the ways human beings talk, but abstract
projections from discourse. Worse still, it has made these projections
not on the basis of real temporal discourse but on the basis of
exemplars of possible speech, cross-cutting heterogeneous settings and
speakers. These examples have been carefully preselected, not on the
basis of their distribution in discourse but on the basis of their
formal similarity as determined by a-prioristically arrived at
grammatical principles. Such decontextualized examples of language are
quite often unrepresentative of real speech. They moreover usually
conceal a prescriptivist bias. This is particularly true when it comes
to assigning grammaticality and ungrammaticality to various strings of
words; in Green 1974:63, to take one of literally thousands of
examples, the following pair of sentences is held to exemplify a
grammatical vs. an ungrammatical use of 'reason' (vs. 'purpose'):

(a) John is endeavoring to stop his hiccups, and Bill is acting like
    he is for the same reason.

(b) John is endeavoring to stop his hiccups, and Bill is acting like
    he is for the same purpose.

The linguist's argument here depends on a supposed grammatical
distinction between reason and purpose clauses. Since to stop his
hiccups is a purpose clause, it cannot be referred to by a word
appropriate for reason clauses, so (a) is wrong (in the linguist's
jargon, "ungrammatical," or "unacceptable.") But notice that it is
only by virtue of the explanation itself that we can assign
ungrammaticality to (a) and grammaticality to (b). Our linguistic
intuitions are useless in such cases--we have simply not encountered
enough contextualized examples to make the correct projection. Indeed,
from the linguist's point of view our intuitions are worse than
useless, since to most speakers of English the judgement is precisely
the reverse--only a total nerd would use 'purpose' in such sentences,
or indeed say anything like these sentences at all. But the
homogenizing strategy of the formal linguist methodologically excludes
'social information' of such a kind. In doing so of course its outward
motives are scientific, to purge from its data anything non-mental,
anything extraneous to the individual mind as an isolated
psychological unit, and by segregating language from its somatic and
social--in other words its temporal-- surroundings to study it in its
abstract purity. The effect of this move is a radically hypostasising
one in which quite disparate linguistic events are selected for their
formal--epiphenomenal--consistency. What is factored out in this
process is nothing less than the entire set of social and political
forces constitutive of speech as a mediator of human interaction.

Such connections between the post-structuralism of linguistics and
postmodernism have two important implications. One is that they open
up interdisciplinary discourse for linguistics. Linguistics has tended
to rest on its laurels after the triumph of structuralism, and to take
it for granted that other fields would continue to adopt its newer
methods and assumptions such as deep structure, competence, and so on
just as they so enthusiastically coopted structuralism in the 1950's.
When this did not happen, linguistics was in effect sealed off from
influences of the social sciences and literature. The time seems ripe
for a new interaction, and for linguistics to enter the debate about
'discourse' in its wider sense.

Secondly, representational linguistics, the view of language as
mentally represented and in principle insulated from social forces,
has meant that linguistics as such has had nothing to say about
society and political ideology. This is not to say that individual
linguists like Noam Chomsky have been silent on social and political
questions, but only that their activity as linguists has not been
engaged in this way. Rationalism is egalitarian in postulating
genetically prescribed abilities, but its corollary that language is
mentally represented and that linguistics is involved only with
abstract competence, not with existential performance has meant that
language not only can but in principle must be isolated from political
and social concerns. Consequently the argument from rationalism to
egalitarianism is too remote from language itself to count as a
politicization of linguistics. Poststructuralism will inevitably push
linguistics into the exploration of the social and political
implications of the poststructuralist view of language as an intrinsic
part of the discipline.

John Searle, in his lecture at the 1987 LI on speech acts, observed
that while you can make a promise simply by saying it, you cannot fry
an egg simply by saying it. Both of these can be disputed. Promises,
as is well known, can be made infelicitously. But it is not true that
I cannot fry an egg simply by saying it. An egg in a hot frying pan
may be at various stages of coagulation. Responsibility for a
declaration that one of these stages is that at which the egg is
'fried' could easily be vested in an individual, say the teacher of a
cooking class. It could also be agreed upon by two or more people in
collaboration. More importantly, without such an imposed or negotiated
definition there would be no point in using the word 'fry' about the
activity in question at all.

The point that I mean to make here by engaging Searle and his version
of 'speech act theory' that is one of the more recent incarnations of
structuralism is that if words and their meanings are social
constructs, the linguist's task crucially includes the study of the
power relations within which definitions of words are negotiated and
hegemonised. The current struggle over reproductive freedom has
brought such matters sharply into focus, revealing itself to be
primarily a struggle for control of the definitions of words like
'human life', 'child', 'fetus',and so on. The importance attached by
authority to the right to define was seen vividly in a recent lawsuit
in which the word malice was centrally at issue. The judge in that
case carefully explained to the jury the legal definition of malice,
but the foreman of the jury, a school teacher, later admitted to the
judge that he had read to the jury the dictionary definition of malice
found in Webster's dictionary! The judge sharply reprimanded the
foreman and the jury and declared a mistrial. This is the most
explicit case I know of in which authority went one step further than
claiming the right to define and, by enforcing the claims of the legal
professional establishment over those of the recognized civil
inventory of definitions, actually preempted the means of defining.

The study of language began in the ancient world as a very pragmatic
affair, the agonistic skills of persuasion, dialectic, and exegesis,
with all of their ethical potential. Its encounter with oral discourse
is destined to take linguistics beyond the concern with 'grammar' as
an a-priori mental construct and by re-contextualizing it in the
scenes of its formation restore to language its crucial temporality.
If these scenes are ineradicably social and can never be politically
neutral, this means that linguistics will in making this return to
orality at the same time reclaim something of its moral heritage.


Appendix

Well no the problem is and this is what the psychologist has mentioned
to me. These kids wont wont show any hope like the see you take a
normal on the average retarded child I mean the one who does not have
any handicaps like blindness or deafness or something like that. he
will improve a little bit. maybe a lot. It depends on how badly
disturbed he is. but these people wont because they are still going to
no matter what happens they are going to be living in a fantasy world.
because they are blind. and they have to imagine and they keep asking
one question after the other and then nothing they say makes any sense
and nothing is relevant to the situation. and it never will be because
they well theres just such a sharp line of differentiation between the
normal blind and then the emotionally disturbed blind. (Carterette and
Jones 1974:422).
FORMULAS

the problem is 
has mentioned to me 
these kids 
you take a little bit 
may be a lot 
it depends on 
no matter what happens 
they are still going to 
living in a fantasy world 
one question after another 
nothing they say makes any sense 
relevant to the situation 
sharp line of differentiation 
emotionally disturbed...
(etc.)





II. TEMPORALITY IN DISCOURSE

Yesterday I suggested that linguistics was facing a crisis that could
be said to reflect a more general situation, what Paul Rabinow has
called a 'crisis of representation' afflicting the human and social
sciences. I suggested that linguistics had until recently remained
insulated from these crises because of its decision to exclude orality
from its purview, and that the new concern with orality, as opposed to
the mere equation of spoken language with written sentences read out
loud, was tantamount to opening linguistics to postmodernism and
thereby linking the projects of linguistics with those of current
thought in cognate fields.

I don't wish to set an agenda for linguistics as it moves into its
poststructuralist phase. For one thing this would be a somewhat
arrogant thing for me to do, but more importantly because agendas and
paradigms in general don't seem to be the appropriate way to think
about linguistics any more. As the idea of 'a language' becomes less
coherent, less a total concept, the idea of a scientific
conceptualization of the procedures of linguistics becomes less
cogent. Fifty years ago Rupert Firth aptly compared linguistics as an
academic field to geography, with its lack of clear disciplinary
boundaries and highly variegated subject matter, each aspect demanding
slightly different methods and making slightly different assumptions.
A foregrounding of 'language' might then be the only thing that holds
linguistics together. (If this seems an obvious statement to make,
consider that formal linguistics has attempted to preempt the term
linguistics for its own very narrow project of sentence calculus.) But
linguistics is not the only field that is concerned with language, and
it is

not at all clear that language is a sufficiently autonomous notion to
be a cohesive force. Consequently it seems to me that a
poststructuralist linguistics will not enter into a paradigmatic
struggle with formal grammar and then prevail by displacing it.
Rather, we can expect that fragmentation will be the only order of the
day, and that the critics of post-structuralism will continue to taunt
its practitioners with the charge of incoherence and lack of a
paradigm; and that these same critics will continue to hypostasize
their own paradigm as exclusively constitutive of the field of
linguistics.

Yesterday I talked about the role of time and temporality in the
differing views of language given by post-structuralism and formal
linguistics. There what was at issue was the limitations of an
atemporal view of language, and how these limitations became
insuperable once language was considered from the perspective of
orality. In today's talk I will return more explicitly to the notion
of time in linguistics and language by considering two views of tense
and aspect. To some extent this discussion will bear on the whole
question of categories in linguistics, and it is perhaps with
categories that we should start.

One of the standard assumptions of all grammatical studies,
including,that is, of linguistics, is that the forms of a language can
be classified according to partly formal and partly functional
criteria in to nouns, verbs, and a few other categories, the number
and nature of which are language-dependent. That these grammatical
categories overlap semantically with real-world percepts like actions
(for verbs) and things (for nouns) has long been recognized, and the
well-known work of Roger Brown with children has shown a broad
ontogenetic basis for the semantic overlap. In a joint article
published a few years ago (Hopper & Thompson l984) Sandra Thompson and
I suggested that the nature of linguistic categories could not be
determined in the abstract, since different discourse contexts called
for different degrees of conformity to a categorical prototype, and
these prototypes themselves derived from negotiated, interacting
discourse rather than from a-priori abstractions. I won't here
recapitulate that entire paper, but will mention only one part of it
that will serve as a basis for what I have to say. Consider the
so-called noun 'fox' in a suitably bucolic utterance like This morning
down by Beecher's Brook we started up a pair of foxes.'.'Fox' here
conforms neatly to the semantic prototype of a noun as Roger Brown and
Talmy Givon have independently described it. lt is thing-like in the
sense of having perceptible contours and persisting over time. It
appears indisputable that we can speak of 'fox' as instantiating the
category Noun, and thereby qualifying for such noun-like perks as
articles, quantifiers, adjectives, possessors, plurality, and so on.
The problem is that neither the formal nor the semantic features of
'fox' appear in every context in which the form 'fox' appears. For
example, in constructions in which the form 'fox' is incorporated into
'hunt', like "My Dad took part in a fox-hunt last year", "They're out
fox-hunting","Fox hunting is cruel", and so on, 'fox' can't have any
of its noun-like perks; it can't have articles, quantifiers,
adjectives, and so on. Worse still, incorporated forms are drained of
the very semantic features held to account for their noun-ness, since
they no longer refer to things but for general classes.

An exactly similar argument applies to the form 'hunt', which appears
now as a noun, now as a verb, and now as something between the two in
'fox-hunting'. Here, in a parallel way, we cannot define 'hunt'
semantically in terms of action independently of its context. In both
cases the point is the same: there is no discourse-independent way of
characterizing nouns and verbs; there are only a categorical forms
which accrue morphological and syntactic peculiarities according to
their discourse behavior.

This behavior, we have suggested, has to do with the reporting of new
events and the introduction of new participants. In view of what I
have been suggesting about the temporality of discourse it would have
to be added to this that the notions 'event' and 'participant' are
themselves categories. Consequently they too are not guaranteed a
priori, but are established interactively. Categories are therefore
provisional and emergent, and the means of instantiating them are
intrinsically unstable. It is only the linguistic strategy of
detemporalizing the data of language that has permitted stable,
autonomous, a priori categories to be identified and hypostasized from
discourse.

Yet the words 'fox' and 'hunt' are recognizably words of English. They
may not be assignable to categories in advance of discourse, but it
surely cannot be denied that they are words, composed of segmental
phonemes (such as [faks] and [h‘nt]). They qualify as members of the
class of 'the smallest units that can be uttered separately'
(Palmer).They are 'minimal free forms' (Bloomfield). They are 'an
autonomous syntagm formed from inseparable monemes' (Martinet). Words
and these mental phonemes that comprise them form a common basis for
formal grammar and poststructuralist approaches to language. We can
therefore in a sense touch rock bottom with categories by grounding a
view of language and hence of linguistics on a common foundation of
words and phonemes. The next stage of my argument will try to show
that even this apparently obvious step might be treacherous. I will
start by asking the question 'Where do words come from?'

Questions like "Where do words come from?" seem baffling from two
opposites points of view. On the one hand their answers are obvious.
In this case, words come from other words. When ever we are in a
position to trace the genealogy of a word, we find that it goes back
to some other words. Modern English 'fee' goes back to Old English
'feoh', meaning money, to the reconstructed Proto-Germanic word *fehu,
and eventually to an even more remote reconstruction
Proto-Indo-European *peku 'wealth, cattle', at which point the
historical trail peters out. 'Nocturnal' was adopted wholesale from a
Latin word with a root noct- meaning 'night' and a compound adjectival
suffix all of whose elements, including its case ending, are probably
of Indo-European date. Even acronyms like have an in direct source in
other words.

On the other hand, the question of where words come from seems
unanswerable. The 'other words' themselves stretch back into an
infinity ago, and their ultimate source is linked to that most
mysterious genesis, the origins of language itself. Words are always
already there. While we can say sensible things about the histories of
individual words, about the origins of words as a class there doesn't
seem much point in asking questions at all.

Yet from the 'bird's-eye view' I talked about last time the study of
language change also tells us that the normal course of events in the
evolution of words is governed by attrition. While this course can be
allowed or even arrested internally by analogical extension and
externally by borrowing and adaptation, left to its own devices the
controlling drift of change is loss. Such loss can be quite trivial or
it can be very drastic. Modern English 'stone' still recognizably
reflects Old English,, Proto Germanic *staynaz. On the other hand, as
Joseph Greenberg has pointed out, the segment [m] in the phrase
"yes'm", which in practice is little more than a nasal offset to the
final [s] of [yes], was once the Latin 'mea domina' (madame >
ma'am...). It seems clear that one possible outcome of attrition is
zero, the actual disappearance of the word, as a result of
cliticization, affixation, and eventual absorption into a stem.In
asking, then, where words come from, we might re-phrase the questions:
"Why are there still words?" That is, instead of accepting the
existence of words, we might look at the strategies found in languages
for the renovation of vocabulary, for the replacement of worn-down
words.

When a morpheme loses its grammatical-semantic contribution to a word
but retains some of its original form, and thus becomes an
indistinguishable part of a word's phonological construction, I shall
speak of 'morphological residue'. Examples of morphological residue
are very plentiful wherever documentary evidence or reliable
comparative reconstruction gives us insights into the earlier
histories of words. A good example from English is 'seldom', from an
Old English adjective seld'rare, strange'. The second syllable of this
word contains the Old English dative plural suffix [-um]. While the
continuation of the earlier suffix is historically compatible with its
adverbial function, there is no Modern English suffix -om in
adverbials or anywhere else. Syn-chronically the suffix is now simply
part of the segmental constitution of the word 'seldom'. More common
than frozen case affixes are instances of words which have become
fused together so that one or more of the original components survives
only as part of the phonology of the new word. Modern English "alone"
is based on Middle English al one,(al='all'). As a result of the
phonetic split between the [w"n] of 'one 'and the [own] of 'alone',
there is now no synchronic relationship between the two words, so the
initial 'al' is purely phonological. In its aphetic(attributive) form
'lone' and the derived forms 'lonely', 'lonesome', aninitial phoneme
[I] remains as the sole relic of the earlier morpheme 'al'. Complexes
of prepositions are especially prone to this kind of reanalysis. In
languages with long written histories it may be possible to identify
several layers, e.g. English 'about' representing 'on'('at'?)+ 'by' +
'out', 'besides' = 'by' + 'side' + 's', where the 's' is a genitive
singular with adverbial sense. In the English word 'about 'there may
be some morphemic status still adhering to the initial 'a-'which
figures prominently in other locative adverbials like 'ahead','away',
'around', and many others, but the 'b' is unlikely to be understood as
anything except an unanalyzable phonological segment. The same is true
of the -i- in handiwork (OE handgeweorc), where only a specialist
would recognize the prefix 'ge', and of German bleiben'remain' (Middle
High German bel ben), with a completely absorbed prefixbe-, as well as
Glaube 'belief', gleich 'like', Glied 'limb', Gl ck'luck, happiness',
Gnade 'grace', and others in which an initial ge-prefix has likewise
been absorbed.

It is significant that demorphologication sometimes results not in the
addition of new segments. Modern English 'alive' owes its voiced
fricatve[v], by contrast with the [f] of 'life', to an earlier dative
suffix which caused a fricative to be intervocalic and thence to be
voiced. As is well known, such new alternations were an important
contributing factor in English to the emergence of a full contrast in
voicing in the fricatives, so that while no new segments accrued to
individual words,the paradigmatic inventory of the phonological system
was indirectly increased.

It seems, then that the segmental make-up of words is historically
quite fragmentary. It is not only that stems and affixes may merge,
but that the stem/affix distinction itself tends to become blurred. It
happens quite often that what was once an affix comes to be the most
prominent part of the word, and conversely that the earlier stem
becomes phonologically subordinate. The word for 'pig' in Germanic
(Gothic swein) consists of an initial consonant cluster sw- and a
nucleus. Yet the cluster sw- reflects the Proto-Indo-European word for
'pig' (*s-) and the nucleus an old adjective suffix (cf. Latin su-nus
"'porcine'). In sets of words like 'glow,' 'glimmer,' 'glint'; or
'snor,' 'snooze,' 'snort,' we are hard put to say how the analysis is
to go. Are gl- and sn- prefixes or stems? And are swine, swig, swill,
which may or may not be historically diverse (cf. Latin suillus
'pig-like,' presumably , *suin-l-os), to be analyzed as
morphologically complex?

Questions of this kind, of course, are not novel, but in recent
linguistics they have not been accorded the same degree of attention
as those concerning more 'obvious' paradigmatic morphology. The study
of morphology in fact tends to privilege precisely the more recent
accretions to stems, in which the relationship of two or more terms
(such as stem and affix, parts of a compound) is still transparent and
even paradigmastic, By factoring out the older layers from its
purview, morphology buys for itself considerable simplification and
generality, and at the same time licenses the study of 'phonolgy,' a
field which concerns itself wit hthe rules governing just these newer
accretions. But structural morphology also incurs the risk of setting
up artificial boundaries between 'new' morphology and 'old'
morphology, and between structure as 'fixed' and structure viewed
either as 'fossilized' (old) or as 'variable' (new).

This stage of morphological residue is obviously only the last in a
process of form and meaning change that need more study. The
phonological segments that we identify as 'stems' appear to be nothing
other than the debris of former affixes. If so, then all our 'words'
consist ultimately of morphological residues which sometimes together
simply carry the meaning of an earlier stem and sometimes modify it
quite drastically; these earlier 'stems,' of course, themselves
consist of phonological segments which are the morohological residue
from yet earlier stems and/or affixes.

If this process has no anterior limit, it is also not synchronically
restricted either. The accretion of phonological segments through more
or less redundant affixation is a constant process; we often do not
recognize it as such because the accruals are usually quite
respectable morphemes, such 'up' in 'upon,' 'with' in 'within' (used
as a preposition), -ate in 'orientate,' 'ir' in 'irregardless,' and so
on. But the effect of these accretions is quite often not so much to
add new meaning where none was before, but to increase what might be
called the 'bulk', the sheer physical length of the word. These
accretions counter balance the attrition brought about by normal wear
and tear on words. The important thing is that they are ongoing and
can potentially affect any word, given the right discourse
circumstances. Accretions can moreover be added spontaneously, made
and dropped in an ad hoc manner --the ones that gain circulation in
the community are exceptions.

We must, then, recognize that behind the seemingly fixed,
self-contained,robust structure which we are inclined to attribute to
the parts of a language there lies a crumbling, unstable framework
that is forever being restored by the collective action of speakers.
There is to morphology and lexicon an emergent aspect which is
entirely similar to that of grammar(Hopper 1988). The apparent
peculiarities of word families like that of gl- and sn- families of
Modern English, which have usually been discussed only anecdotally,
are part of a significant larger picture of a process of segmental
reformation that cannot be marginalized without losing sight of an
essential property of language: what Anttila has called its 'constant
resystematization' through and during actual usage. Joseph Greenberg
has noted:

  "It is not so much that the noun designates persisting entities
  as against actions or temporary states of persisting entities. It
  is that nouns are continuing discourse subjects and are therefore
  in constant need of referential devices of identification."
  (1978:78).

It is an observation that is significant not only for its content but
for the insight into the relationship between discourse and structure
that it reveals. The structuralist, cognitive notion of linguistic
forms which correspond to (or "mirror") concepts or ideas (such as
'persistingentities') is replaced by one of forms caught up in
continual change in response to the on-going referential needs of
discourse. Greenberg's work on degrammaticization points the way to a
perception of words in precisely the same terms, not as a fixed,
structured listing of available formes, as our 'dictionary' metaphor
might compel us to think of it, but as more like a patchwork of
segmental fragments reflecting layers of earlier morphemes, morphemes
that are constantly being accumulated in response to the collective
needs of discourse, and in turn eroded. 

At the beginning this digression I suggested that two linguistic units
might qualify as generally accepted across different linguistic
paradigms: the word and the phonological segment. I then gave some
reasons for thinking of words as being continually in statu nasciendi,
as emergent forms. Might not the same be true of segments? The
Saussurean 'image acoustique', the idea of a segmented mental reality
underlying the linguistic 'phoneme' might then be seen as one more
example of the spatialization of the units and structure of language
which the new attention to orality is destined to deconstruct. I will
not explore this possibility in depth, but some relevant
considerations familiar to all linguists are: 

- Segments are capable of considerable 'stylistic' variation. These
  variants may or may not become phonologized; 
- Sequences of segments may merge into a single segment (e.g. Latin
  [qu]> [kw)
- how do the in-between stages of such a process look? 
- Prosodies may override intersegmental rules. In such cases the
  segment seems to be dispensable.

Dennis Tedlock, in The Spoken Word and the Work of Interpretation, has
some interesting things to say about phonology that I will return to
later in this talk. But one remark of his is full of relevance to what
I am saying here about the construction of words:

  "There are even cases where a phonemic distinction is important
  with respect to content in some words and melts away into a
  continuum in others; that is to say, one might construct one set
  of evidence to support a phonemic discontinuity and another set of
  evidence to support treating two sounds as an allophonic variant
  of purely phonetic (as opposed to phonological) interest." (214)

Cases of this kind point to a construction of phonemes comparable in
microcosm to the construction of words I have been arguing for here. A
recent paper by Pagliuca and Mowry points the way to a radical
rethinking of the segment which amounts to an incorporation of
temporality into phonetics and phonology. And if this is a valid
perspective, it is a further indication that the forms of our written
language, in this case the alphabetic letter, have presented powerful
prototypes for our view of linguistic structure. I don't know how far
it's possible to pursue this line of thought without deconstructing
not only linguistics but all academic disciplines. I do not intend to
be nihilistic about linguistics or anything else, but only to insist
that in 'writing language' we should begin from the basis of its
lability, that is, its non-rigidity, and the social, interactive
construction of its concepts and categories. This means adopting what
Marcus and Fischer (1986) have called the 'jeweler's-eye view' of
language, seeing it up close, observing it in medias res, in the
scenes of its social creation. This need not preclude the opposite
perspective, the 'bird's eye view' and the study of stable
after-the-fact panoramas, but the two perspectives need to be
considered together. In particular, the panoramic perspective must be
placed 'sous rature', always subject to revision in its minutiae. It
is this point that will form the nucleus of the rest of this lecture.

The panoramic perspective on language suggests un syst¸me o¯ tout se
tient, a system of paradigmatic forms related oppositionally. These
related paradigmatic forms constitute what is commonly called the
'morphology' of the language, the inflections, compounds, and
derivations formed from identical stems. But from the point of view of
interactive discourse such systems are limited and transitory. Seen in
close-up, they do not have the kind of totalising, global validity
attributed to them by structuralist theories. The sort of long-term
word-building processes that I have just described make of morphology
a question of the gradual accrual and decay of molecular elements, an
on-going assembling and dismantling of form that resists reduction to
a stable system. Yesterday I quoted a saying from Nietzsche's Birth of
Tragedy: Only that which has no history, Nietzsche wrote, can be
defined. Just so: No discrete level of morphology, or syntax, lexicon,
etc., can be defined because language is temporal and historically
situated in the scenes of its use, governed by previous scenes of
usage. The illusion of stability and fixedness that exists is to be
seen as an epiphenomenon of the internalization of institutional and
political forces that resist the intrinsic tendency toward lawlessness
of the market-place.

There is no better example of what I mean here than the linguistic
category of verbal aspect. We find in many languages a distinction
that may be stated in its roughest approximation as that of point and
line, of action seen from the point of view of its termination and
action seen from the point of view of its internal constituency. These
two points of view are called 'perfective' and 'imperfective'. In his
recent book Aspect and Meaning in Slavic and Indic, Ranjit Chatterjee
notes the difficulties posed by ascribing to 'aspect' a categorical
status and deriving its uses from consistent general meanings of the
forms. He documents the weaknesses of approaches to aspect that assume
any kind of semantic constancy across different realms of usage, such
as for example the 'gestalt' or 'prototype' theories that measure
actual instances as approximations to ideal types, types that may
never occur at all in actual contexts. Chatterjee gives the example of
a sports commentary; the language is Czech:

  S'v dsky 08' hr c' rozj z'd, klic'kuje, padne a op t vst v

  Swedish player moves dodges falls & again gets up
                    away

  'The Swedish player moves away [ipf], dodges [ipf], falls [pf],
  and gets up [ipf] again' [C 59]

Utterances like these violate practically every generalization that
has ever been made about aspect. A sequence of foregrounded events,
punctual, dynamic, completed, and so on constitute a canonical context
that should evoke the perfective aspect. The single perfective, padne
'falls', has the same present tense reference as the other verbs, yet
as a perfective it should not be able to refer to the present.

In such utterances, Chatterjee shows, the distribution of aspects is
more a question of lexical choice and context than of morphological
meaning ('morphosemantics'). For example, klic'kuje 'dodges' lacks a
perfective counterpart; rozj-z'd- 'moves away' is imperfective because
it is ingressive, and this fact creates a context for the following
imperfections. Aspect as a morphosemantic category is neutralized.
There are only lexically diverse sets of verbs that may or may not be
morphologically related, and there is no more point in explaining
these than in explaining the choice of any other lexical item in a
specific context. Chatterjee points out that when decontextualized as
a sentence, the utterance could be forced to conform to the
descriptive canons of Czech aspect by the addition of an adverb
suggesting slowness, e.g. se pomalu. Now, however, the perfective form
padne is incongruous and must be replaced by the imperfective pad .
But now "the original 'ice hockey' flavor would also be absent" (p.
62). So normalization has taken place at the expense of the link with
the specific language game within which the aspectual forms interact.

Two comments are in order. One is that as this single suggestive
example is multiplied, we will find that the choice of aspect is in
the end determined not by previously established form-meaning
correlations but by the kind of language game in which they figure.
Here is what Wittgenstein has to say about language games:

  "I shall in the future again and again draw your attention to
  what I shall call language games. These are ways of using signs
  simpler than those in which we use the signs of our highly
  complicated everyday language. Language games are the forms of
  words. The study of language games is the study of primitive
  forms of language or primitive languages... When we look at such
  simple forms of language the mental mist which seems to enshroud
  our ordinary use of language disappears...We see that we can
  build up the complicated forms from the primitive ones by
  gradually adding new forms... Now what makes it difficult for us
  to take this line of investigation is our craving for generality."
  (PI p. 17)

The apparent intrinsic meanings of the aspectual forms are then
hypostasisations of their functions across different language games.
The applicability of this line of thought to linguistic categories is
clear,but especially so with aspect, about which one investigator has
written (echoing a widespread sentiment among aspectologists):

  Numerous attempts have been made to devise a system with
  definite rules for the use of the aspects. But the final result
  has invariably been that this was found to be impossible, since
  there would have to be a special rule for practically each kind
  of action. (Pontoppidan-Sjvall, cit. Chatterjee 44)

Within language games the rules for aspect may be relatively simple.
But rules for the use of aspects in one game may not carry over to
another game, or may carry over only partially.

A second comment concerns the interactive nature of the categories
that are being proposed. For agreement on aspectual meanings implies
agreement between interlocutors on the kind of language game that is
being played. For example, whether a sequence of events is reported
with 'imperfectives' and occasional present-perfectives or with
perfectives will determine whether it is understood as a commentary on
something occurring or a historical report. Agreement of this kind is
typically contractual, predicated on the jointly arrived at perception
of the participants. Somewhat closer to home, the well-known use of
the narrative present in English has little to do with 'time' in
either the objective or the linguistic sense, but with shifts in how
the participants in the discourse perceive the events, whether the
speaker is aiming to create involvement with the interlocutor, and so
on. Again the 'use' of the so called 'present tense' in English is not
necessarily constant across language games. In some contexts it will
suggest a temporal deixis relative to the speech act. In others it is
generic, in still others narrative, and so on. Explanations along the
lines of the speaker bringing a past event into the present in order
to lend vividness to the report confuse different language games,
borrow a generalization from one language game in order to account for
usage in another quite different one. Again Wittgenstein makes the
point:

  "The idea of a general concept being a common property of its
  particular instances connects up with other primitive, too simple,
  ideas of the structure of language" (Blue Book p. 17).

The particular case of aspect has been introduced in this lecture
because of course it is a category (well, if it is a category) that is
temporal par excellence and has come close to engaging the attention
of philosophers such as Vendler, Kenny, and Mourelatos, in the context
of the theory of actional types, as well as of formal linguists like
Dowty, who has devised a formal calculus of time. Such logical
analysts over look the essentially social nature of ideas like event,
process, and state, and attribute to them an a-priori mental
permanence and transcontextual constancy that they do not in fact
possess. The foundational nature of events that follow in temporal
sequence can easily be subverted by asking whether 'temporal sequence'
itself is not in fact determined by our perception of 'events'. Our
perception of 'events' in turn is crucially dependent on socially
arrived at agreements over questions like what, if anything,
'happened'. (Darren Newtson's work suggests that such perceptions are
subject to all kinds of contracts.) Time, then, does not present a
preexistent framework, a flux against which things happen, but rather
time is secondary to sequencing. Time is defined by the sequentiality
of events rather than the other way around (Dry). It follows that
sequential time is primary only to the extent that eventive discourse
is privileged, for example in scientific events (Wood 1989). This
privileging of scientific sequentiality has had important consequences
for the linguist's approach to tense, aspect, and modality. It has
meant that 'tense' has been singled out as a primary category, and
separated from aspect and modality.

Chatterjee in his Wittgensteinian treatment of aspect considers the
charge that by making aspect a lexical property of verbs he has
effectively eliminated aspect as a morphological or grammatical
category. Unwilling to go this far, he suggests instead that "there is
an implicit, tentative or 'fuzzy' quasi-system against which aspect
games can be and are played. If one wanted to innovate conceptually,
it could be suggested that Grammatical meaning, as much as lexical
meaning, is capable of metaphorical extension, if not metaphorical in
essence" (61).

There comes about, then, a tendency that pulls against the tendency to
lexicalization of members of a paradigm. This is the speaker's
creative tendency to mobilize forms that were useful in one language
game when involved in another. There is a leap-frogging effect whereby
perceived success in the use of a form in one context, i.e. in one
language game, encourages one of the players to repeat it, either in
the same game or in another, different game. Yesterday I mentioned
David Lodge and Marin County Psychobabble. In a footnote to that
article, Lodge describes an oral exam back in England that he
attended. In discussing the candidate's performance, Lodge used the
psychobabble expression 'couldn't quite get it together', at that time
a completely new phrase. Within half an hour two of the other
examiners had also used the expression 'get it together'.

Furthermore there is layering, the creation of new forms when the old
ones have crossed some kind of threshold of semantic recoverability.
The past tense 'sped' in English no longer seems to work when the new
language game about driving a car over the legal speed limit is being
played. Consider an exchange I heard some while ago:

A: How did you get here this fast? It's 20 miles and you called me
   only half an hour ago.
B: Well, I speeded most of the way.

It is in this way that the old and the new are allowed to coexist:
dreamed and dreamt, brothers and brethren, and so on.

This 'jeweler's-eye view' of structure is not intended to abolish the
linguist's legitimate concern with regularity, but only to
contextualize it by restoring to it its temporal dimension, which I
think is the crucial issue here. The perspective I am emphasizing is
that of particularity in social interaction spreading in concentric
circles through repetition in other contexts. Regularity is then to be
seen as transtextuality, perhaps on a small scale, perhaps quite
global. If it is global we might interpret it as 'grammatical' and as
being a fixed, stable part of language that would hold still while we
examine it. But the danger here is that our method will encourage us
to isolate just those parts of the language that present themselves as
maximally fixed and stable in this way and afford them the central
place in our analysis, in other words, to hypostasize them as grammar
and morphology, and ascribe to them a unified location in the
individual's mind.

I've tried in these two lectures to present from a linguist's
perspective the dialectic between two views of language. One of these
is mentalistic and representational, the other social and interactive.
I've tried to show that the different attitudes toward temporality
constitute a crucial division between these two views. I've also
suggested that whatever movement can be observed in the direction of
the social-interactive view of language, of grammar and mind as social
constructs, accompanies a return to spoken language, that is, orality,
as the linguist's primary concern. This is something like saying that
the concern with orality constitutes a genuine new paradigm for
linguistics. For the reasons that I have discussed I do not like to
think of linguistics in this way. Behind any talk of paradigms lurks a
specter of power, authority, and total mobilization of effort that,
whatever its ideology, needs to be resisted rather than welcomed.
Perhaps the overthrow of the old paradigm does not necessarily have to
be accompanied by the instatement of a new one in the Kuhnian sense.
Perhaps the very rejection of totality in linguistics means that the
only new paradigm will be the absence of paradigms, the sort of
fragmentation that I talked about earlier, a kind of disciplinary
anarcho-syndicalism. There would then only be a collection of
practices, a set of linguistic games linked family-resemblance style
to some very general assumptions. An appropriate place to conclude
would be with some of these assumptions, which might be broadly
presented as follows. (I owe much to conversations with Rom Harrˇ):

(l) The Assumption of Temporality. This is rather like Winograd and
Flores's 'historicity'. It is the 'lateral' as opposed to 'vertical'
displacement of language that I talked about yesterday, the idea that
language owes the way it is to the way it was rather than to some
abstract, supervisory schema. This assumption is at the heart of the
next two.

(2) The Assumption of Contextuality. Meaning is tied to contexts, and
cannot be segregated from these contexts. That is, there are no
decontextualized meanings. Forms take their meanings not from a
discrete mental lexicon but from the contexts in which they occur and
pass them on to other, new contexts.

(3) The Assumption of Indexicality. The individual is situated
temporally, spatially, and socially. What is on hand confronts a
specific consciousness, a real person in a geographically locatable
place and a historically fixed time. Speech reflects what Bakhtin
called 'the unitary and unique event of existence' (cit. p 181 of
Clark & Holquist).

Of these three assumptions, the assumption of temporality is central
to the other two. Viewing language from the perspective of time
constitutes a radical change in the nature of linguistics. The
detemporalization that comes with a linguistics of writing has brought
about a focus on one aspect of language, that represented by the
historically situated conventions of writing. It would be naive of
course to think that by substituting a new set of conventions for
writing language we shall be able to achieve some kind of ideal form
of representation. We must recognize that eventually such a new form
of writing will inevitably recapitulate the errors of the old. It
would constitute a new paradigm for linguistics, and we have seen that
a new, holistic paradigm is scarcely the solution. But there is a need
for experimentation in new kinds of writing. One is the continual
refinement of alphabetic writing as a tool for representing spoken
discourse, in other words a refinement of transcription. The unique
importance attached to transcription by the Santa Barbara linguists
and a few others is symptomatic of this need. Dennis Tedlock, in his
book The Spoken Word and the Work of Interpretation, notes that such
phenomena characteristic of real speech as contouring, timing, and
amplitude have been downgraded by linguists because they do not lend
themselves easily to reduction to discrete units. The meanings
imparted by these factors are continuous, they are gradated as a
continuum of things like in completeness and finality that the simple
period (full stop) too easily dichotomizes into either finished or not
finished. The linguist's strategy of treating such phenomena as
'suprasegmental', in other words as 'supplementary' (to use a word
full of post-modernist resonances), presupposes that the meanings they
convey are somehow detachable from word meanings and sentence, and are
secondary. In this regard it is interesting to note that the Serbian
gusler whose oral epics have become well known through the work of
Parry and Lord were unable to recreate their stories except in the
rhythmic style of their chanted pentameters. Deprived of their gusla,
the monotonous fiddle that accompanied the song, they lost their
memories and were unable to piece together their formulas.
Extrapolating from this observation of Lord's, it seems quite possible
that the kinds of meanings that are factored out in our written
representation of speech are inseparable from meaning as a whole. Yet
these meanings that have been factored out are precisely the temporal
ones. Evidently in order to be considered phonologically relevant a
feature must be atemporal; in Tedlock's words, "the one problem that
all these features (namely pauses, intonation, stress, vowel quantity)
pose in common is that of temporality, and a given feature will be
accepted phonologically to precisely the degree that a way may be
found to reduce its temporality to instantaneity" (204). Tedlock
suggests that might be thought of as a 'temporalized' form of writing.
A storyteller's narrative, for example, might be recorded on a
continuous moving scroll that reproduces not the 'words' arrived at by
linguistic analysis but the variations in pitch, amplitude, voice
quality, and duration, including the pauses and their precise length
that in here in the entire performance.

This seems like a good place to finish up. It is on a note of
optimism, in that it suggests that post-structuralism and
postmodernism are not going to destroy linguistics any more than they
will destroy any other discipline, at least not just yet. It also
suggests that ingenuity and commitment will find new ways,
theoretically and practically, to investigate language, but that
language so investigated will be situated and temporal.