"Natural" and Constructed Dialogues
linguists have argued for decades over what kind of examples it is acceptable
to study. Back in the 1950s and 1960s Noam Chomsky was notorious for using made-up
examples like "Colorless green ideas sleep furiously," "Flying
planes can be dangerous," and "Mary is eager to please/Mary is easy
to please." In reaction against this practice, many linguists have insisted
that linguistic study be based solely on "real" language, language taken
from actual spontaneous conversation.
has typically meant studying snatches of conversation taken from some corpus or
other, such as the British National Corpus in the U.K. or the Linguistic Data
Consortium in the U.S. (For links to actual corpora, mailing lists for people
interested in corpora, and corpus analysis tools, go here.)
These corpora are painstakingly compiled by first taping actual conversations,
then transcribing and annotating them on paper (or in digital text form); increasingly,
also, the original audio- and videotapes are being made available, both on line
and on CD-ROM. When a linguist studies a conversation taken from a corpus, the
reasoning goes, s/he is studying actual languagenot made-up examples.
From a performative viewpoint, however, things look a little different. For example, look at this passage from the Michigan Corpus of Academic Spoken English (MICASE), part of a recorded and transcribed advising session in a professor's office that is interrupted by a knock on the professor's door:
Here S2 is the professoralmost certainly someone named Peter Nelson. S4
seems to be an advisee of his named Naomi Bernstein. S1, S3, and S5 (who doesn't
say anything in this extract) are other students; S3, probably named Karen, is
just leaving as this extract begins. In addition to the actual transcribed words,
we are given annotations signaling nonverbal behavior, enclosed in < >,
like entering and exiting, laughing, and pausing <P :07> (seven-second pause),
and overlaps (when two people are talking at the same time), enclosed in [ ],
as in "i mean you're famous too [S4: yeah ]." Because the transcript
gives us the exact words these people said, we get false starts like "there's
a there's a famous, i mean you're famous too [S4: yeah] there's another famous
there's an art critic, or someone who writes on art whose name is Naomi Bernstein."
Here Professor Nelson seems to be attempting to say "There's a famous art
critic named Naomi Bernstein" or "There's a famous Naomi Bernstein who's
an art critic" and stammering a bit, the way we all do in actual speech.
And of course
if you're interested in the syntax of ordinary speech, an exact structural record
of that stammering is a useful thing to have. How do people flounder around when
they can't quite articulate what they're trying to say?
For a performative study of speech acts, however, this transcript does not seem significantly more useful or even more "real" than a movie or play script. What is going on here, dramatically? What speech acts are Professor Nelson and his advisee Naomi Bernstein performing? In the exchange
are they doing? What kind of speech act is it for a professor to ask a student
"do you give money?" and the student to agree "yeah yeah"?
We don't know, of course, because we can't get inside their heads, but we can
guess, because we've been in enough similar situations that we can fill in the
blanks with what we take to be reasonable accuracy: they're almost certainly joking.
The student is asking the professor to perform an action that he is required to
do as part of his job, sign a drop slip; in this particular U.S. academic context
(probably: still guessing) bribery is neither condoned nor realistically contemplated;
it's a joke. When Naomi Bernstein says "yeah yeah" she is not promising
the professor a bribe; she is going along with his joke. Again, we don't know
this; it certainly isn't (and couldn't be) annotated in the transcript. We're
specifically, we're interpreting the words on the page along the lines of our
own familiarity with similar conversations. We're "adding" or "filling
in" the dramatic context and motivations "behind" the words on
the pagebecause we need to "know" those things in order to study
speech acts, and because, well, that's what we do. We always flesh written words
out with imagined dramatic contexts. (We also do it with spoken words. We "make
sense" of conversations we're in by "filling in" or "constructing"
a dramatic context for the words other people are speaking, guessing at what they're
intending and how they're performing their words.)
performatively speaking, the "reality" of this passage is not "in"
the fact that it was transcribed from an actual conversation. The reality, rather,
lies in, or emerges out of, this imaginative dramatic (re)construction of the
conversation that we do to it when reading it.
is an important and rather difficult point that bears careful unpacking.
For the constative
linguist, studying language as machine, the "reality" of language is
structure: not the actual words, but the structures "behind" or "beneath"
the words. This means that a transcript of a conversation that represents the
exact structure of the conversationall the words in their precise original
order and all the observable nonverbal behavior surrounding the wordsis
just as real as the original conversation, because the structures are the same.
Since constative linguists are interested only in the structures underlying actual
language use, they have to limit their study to examples taken from actual language
usewritten "representations" of actual conversations. A movie
script would be an imitation of actual conversation, not an exact structural representationhence
useless for analytical purposes.
the performative linguist, studying language as drama, the "reality"
of language is people doing things with words: not the actual words, certainly
not the structures behind the words, but the actions performed with those words
by real people in real situations. A transcript of a conversation is not "real"
just because it is structurally identical to the original conversation; it "becomes"
("feels") real to a reader who is able to "act it out" in
his or her imagination, to flesh it out with a dramatic reconstruction. (Remember,
the performative reality of language is always that feeling.) A reader reading
a corpus transcript imaginatively, dramatically, performatively is himself or
herself doing things with words: the act of interpreting the passage is itself
a dramatic action, a speech act. The interpretive or analytical actdeciding
that the professor's apparent request for a bribe is just a joke, sayis
an action performed with those words in a real situation (the situation of reading
the passage) by a real person (you or me). While we are reading the passage and
"dramatizing" it in our heads, it is real, because we are doing that.
And this means that
it makes no difference to the performative pragmatician whether the the language
s/he studies is "real" or "made-up"whether the written
representation of conversation being "imaginatively dramatized" is a
transcript of an actual conversation or something written by a screenwriter or
a playwright. The imaginative "performance" or dramatization that makes
the passage real, gives it what reality it ever has, is the same for both. Given
any piece of dialogue of any kind, in fact, not just writtenincluding taped
conversations and even real people sitting in a room and talkingwe are going
to have to "add" our own imaginative dramatization to the words on the
page or the words being spoken for the conversation to be real to us.
that sense the only two significant differences between any two dialogues for
the performative pragmatician are that (a) some dialogues feel more "plausible"
than others and (b) some dialogues are richer, dramatically more complex than
This is a subjective judgment, of course. Each interpreter will apply his or her
own experiences with language to a conversation and decide whether it "sounds"
or "feels" like real conversation. When I read movie and TV scripts,
some feel very plausible and thus real; others feel implausible and unreal. Other
people reading those same scripts may feel differently about them. The same goes
for corpus transcripts, and even for actual conversations. I sit in an airport
gate area eavesdropping on a couple conversing behind my back, and think: this
is not a plausible conversation. They're doing something other than "naturally"
or "spontaneously" conversing. They're playing some sort of game. They're
playing roles. Or else they've just said these same old tired things to each other
so many times for so many years that they no longer believe them, no longer feel
them; this is just a stale representation of a spontaneous conversation.
course, for the performative pragmatician all those things are "doing things
with words" too, and thus interesting objects of study. The point is that
real people sitting in a room talking make a conversation "real" for
the person with the tape recorder, the corpus compiler, and the constative linguist
using the corpus. For the performative linguist all conversation is "real,"
but not all "real" conversation is plausible as spontaneous conversation;
many other things may be going on that are worth studying. And in fact see Derrida
(1989) for his famous notion that all supposedly "spontaneous" conversation
is in fact a reiteration of past conversations, steeped in the iterability of
all languageand that without that iterability it would be impossible for
us to understand each other. Bakhtin (1934) too insists on the saturation of all
words in previous conversations, and the orientation of all utterances toward
an answering word.)
b. Dramatic complexity. For the performative pragmatician, all language use is dramatic, because all language use is people interacting with words. But some dramas are more complex than others. In the corpus transcript above, for example, the passage I discussed
is dramatically more complex than, say, this:
the first passage the two speakers seem to be operating on more levels at once:
they are (we think) pretending to negotiate a bribe but actually joking. At the
end of that passage Peter also seems to be pretending to be a foreigner ("you
drop class?")perhaps (interpreting some more) because when Naomi repeats
Peter's line "you give money" back to him it sounds vaguely unidiomatic
to him, and he exaggerates the unidiomatic sound of his own phrase for humorous
purposes. In the second (we think) they are saying exactly what they mean: Naomi
Bernstein says she has to drop the course because she can't handle the workload
of five classes and needs Peter Nelson to sign her drop slip; Peter Nelson echoes
her"okay. so workload"and (presumably) points his pen at
the line where he's supposed to sign and says "here?" This difference
makes the earlier passage much more interesting to the performative pragmatician
than the latter. There's more going on in it, dramatically, than in the latter.
There's more to study in it.
as this example shows, plausibility and dramatic complexity are found in "real"
conversationsconversations occurring spontaneously outside our own dramatic
imaginations. But the "realness" of the conversation, whatever that
means, is less significant for the performative pragmatician than the experiential
(subjective) sense of plausibility and dramatic complexity. This means that, while
performative pragmaticians have nothing against transcriptions of "actual"
conversations from corpora or other sources, they also have no reason to favor
such transcriptions. Any use of language that seems plausible and dramatically
complex will dosince the true "reality" of the conversation will
be handled by you, the reader.
in fact, as I say, the bulk of the examples used in this book are taken from movie
and TV scripts. This is largely because a successful movie or TV show will almost
always be more consistently complex dramatically than a spontaneous conversation.
Spontaneous conversations have their moments of dramatic complexity, but they're
rarer than in a well-written screenplay, where the screenwriter has "condensed"
dramatic complexity into every scene, every interchange, every line. A good screenplay
is (or can be, depending on who's watching) just as plausible as a taped or transcribed
spontaneous conversation, but it has fewer "flat" stretches, fewer places
where the drama is fairly straightforward and thus (to use a technical term) boring.
Another perspective on this issue is brought to the table by conversation and discourse analysts who are committed to studying "natural" discourse but also see some value in studying "constructed dialogue" such as novels and plays. Lakoff and Tannen (1984: 323) write, for example:
approach to "constructed dialogue" is fundamentally constative: the
authors speculate that there is a single "internalized model or schema for
the production of conversation," and that therefore by studying constructed
dialogue we may discover "the ideal model of conversation strategy."
Lakoff and Tannen seem here to be pushing conversation analysis in the direction
of Chomskyan competence: the ideal model of conversation strategy would be the
deep pragmatic structure of the ideal speaker's conversational competence, which
actual speakers as well as novelists and playwrights and screenwriters then "exemplify"
in "literal use" (i.e., Chomskyan "performance").
is also, however, a fleeting performative moment in this passage, in the authors'
suggestion that a tape or transcript of natural conversation strikes us as less
natural than a scene from a novel or movie or play. The performative interpretation
of this impression would not be that there is a deep-structural model at work
here, but that we are ourselves constantly generating "naturalness"
by performing language in our heads. "We" meaning language users, of
course: speakers and listeners in a conversation; scholars studying language;
anybody, at any time we try to interpret language.
this point of view, what makes taped or transcribed "natural" conversation
seem "unnatural" is that we have to work hard to generate "naturalness"
for a conversation that we are not currently participating inthe added distance
that comes from studying a conversation as an outsider adds interpretive or performative
difficulty as well. When we interpret, we want to zero in on the most important
stuff and ignore everything we take to be peripheral, the "noise." (This
is an important part of all brain function: imposing an interpretive order on
sense-data, distinguishing "message" from "noise" and suppressing
awareness of "noise" so as to simplify the process of constructing a
coherent interpretation of the "message.")
we engage in conversation, we have access to all the complex contextualization
cues that could help us impose a coherent interpretation: not just body language
and other physical cues but relational cues as well, various emotional and somatic
"sympathies" or "vibes" or "connections" between
or among the conversational participants. This helps us sort through the cues
and discard or suppress the ones we take to be irrelevant, focus in on the ones
we take to be most important.
conversation is a form of data-compression: we compress all the contextualization
cues we detect in a naturalistic setting down to visual and auditory data on a
two-dimensional screen. Audiotaping compresses this data even further, eliminating
the visual; and transcribing it compresses it further still, eliminating the auditory
as well, and representing all naturalistic cues with artificial symbols (letters,
punctuation marks, conventional transcription notation, etc.) When we analyze
videotaped conversation, therefore, we have to work harder to generate naturalness;
and with audiotaped or transcribed conversation, the imaginative effort required
to generate naturalness ("perform the conversation as natural'")
becomes progressively harder still.
that a similar increased imaginative or performative difficulty comes when you
try to engage in conversation in a language you have only studied in the classroom:
there the contextualization cues are potentially present, but since you don't
know how to read them yet, can't "activate" them as present, they might
as well not exist.)
the novelist or screenwriter is doing, then, is not "imitating" an ideal
model of conversational strategy but giving the reader or viewer a little extra
interpretive assistance to compensate for the inevitable data-compression that
results from not being there. In a play, actors (with the director's help) add
their interpretive assistance too; in a movie, so do editors, sound technicians,
all the hundreds of other crew members who work together to create the imaginatively
"powerful" and "lifelike" images on the screen.
in fact is what some cultural prophets of doom are talking about when they say
we are "amusing ourselves to death" (Postman 1986) by watching so much
television and so many movies and reading so few books: reading a book requires
a greater imaginative effort to generate naturalness, to perform the dialogue
as natural, to make it "come alive." And it's even harder to make transcribed
"real" or "natural" conversation come alive, because there
nobody is trying to make your performative projection any easier. You get no help
at all. You have to do all the work yourself.
that Lakoff and Tannen seem, like Chomsky, to be moving in a performative direction
by taking into consideration conversational strategieswhat people do with
wordsbut back off from the most radical implications of that direction by
idealizing the strategies, imagining an ideal model that guides us, perhaps in
some sense "wields" or "operates" us, like the operating system
of a machine.
In a truly performative methodology, the "realness" or the "naturalness" of any conversation is an interpretive construct, something made up by any real interpreter. This means that in terms of "realness" or "naturalness" it doesn't really matter what kind of dialogue you study: interpreters will always provide that (or at least will try very hard to). All that matters is how much assistance you give interpreters in constructing naturalness: just the script? The script plus a video clip? The script plus a video clip plus context?
More: Teacher's Guide index