SOME THOUGHTS ON SCHEMATA 
Wallace L. Chafe 
University of California 
Berkeley CA 
I will try to summarize here a few 
ideas and questions that have arisen from 
efforts over the past few years to develop a 
model of what I call verbalization: the set 
of processes by which a person converts 
nonverbal knowledge into a verbal output. I 
have been thinking mainly in terms of a 
person who has experienced somethin~ which 
he later decides to tell about, although the 
examples given below will be of 
verbalizations of secondhand experience, 
acquired from hearing someone else tell 
about it. Even in the latter cases, 
however, I will assume that much of the 
knowledge being talked about has been stored 
in nonverbal form, so that the same kinds of 
verbalization processes must be applied. 
I find the most interesting processes 
in verbalization to be those in which the 
speaker has to make a choice of some kind. 
The need to make choices arises because 
experience is particular -- it involves 
tokens which are localized in time and 
space -- whereas language has to operate in 
terms of types. Elements of experience have 
to be interpreted as instances of types in 
order to be verbalized. And much of this 
kind of interpretation takes place while the 
speaker is speaking. He may have in mind, 
for example, some particular event or 
object, but he cannot express that idea in 
words until he has decided on an 
interpretation that will help to communicate 
more or less effectively what he is thinking 
of. Evidently some interpretation of this 
kind takes place during perception, and 
probably some of it is accomplished while 
the material is stored in memory. There 
remains much more, however, that has to be 
decided on while the speaker is talking. 
The most direct evidence that this is so 
comes from (I) the fact that the same 
speaker may be found to make different 
choices in verbalizing "the same thing" at 
different times and (2) the fact that 
speakers hesitate a great deal in speaking, 
these hesitations occurring at points where 
such decisions are being made. 
While all these choices are alike in 
that they involve the interpretation of a 
particular as an instance of a type, they 
seem to be classifiable into three distinct 
kinds, defined by the nature of the 
particulars as well as by the nature and 
function of the types to which the 
particulars are assigned. I like to think 
of verbalization as beginning with a speaker 
who has, at the outset, a holistic idea of 
some incident or story he wants to 
communicate through language. This initial 
chunk is too full of content to be 
communicated directly in anything more than 
a gross summary ("A funny thing happened to 
me this morning"), and must therefore be 
broken down into smaller chunks. This 
process of "subchunking" is evidently 
governed by the existence in the speaker's 
mind of patterns which I have been calling 
schemata. He might, for example, interpret 
89 
the initial chunk as an instance of schema 
X, which then dictates that certain smaller 
chunks may or must be included. Each of 
these smaller chunks may in turn be 
interpreted as an instance of some schema, 
and so on in a largely hierarchical fashion. 
My discussion here will mainly involve 
these schemata, but I need to say a little 
about the other two kinds of choices that 
are made during verbalization. What the 
speaker is aiming at in breaking down larger 
chunks into smaller ones is to arrive at 
chunks that are of optimum size to be 
converted into sentences. When he has 
reached that stage, the speaker is able to 
take a chunk and, instead of schematizing it 
further, to interpret it as an instance of 
what I have been calling a frame (in the 
sense of "case frame"). Typically the chunk 
at this stage will be an event. Instead of 
choosing a schema, then, by which this event 
will be broken down into subevents, the 
speaker chooses a frame by which the 
participants in the event are factored out 
and assigned roles within it -- roles such 
as agent, patient, or beneficiary. 
The third and last kind of choice that 
has to be made during verbalization is 
motivated by the need to find words or 
phrases that will communicate the ideas the 
speaker has of events and of the individuals 
involved in them. This is the kind of 
process that has been called categorization. 
The choice of a category, while it is still 
far from being fully understood, has been 
studied more than the other two kinds of 
choices and has at least the potential of 
shedding some light on the nature of all 
three. 
I will try at this point to illustrate 
something of the nature of schematization 
(and, in passing, of framing and 
categorization) with reference to the 
following American Indian folktale. It was 
recorded in the Caddo language (in several 
versions by the same narrator at different 
times), but the points to be made can be 
illustrated in English translation. 
The Wildcat and the Turkey 
It is said that once, long ago, Mr. 
Wildcat was digging roots in order to make a 
garden. Presently he heard someone talking. 
Mr. Turkey was standing there. Mr. Turkey 
said, "Well, well. You are busy. What are 
you doing?" '!I'm digging roots to make a 
garden. What are you doing?" "Nothing. I'm 
just looking around." "You'll be in my way," 
said Mr. Wildcat, and he caught him, 
plucked him, and said "Go over to my house 
where my wife is. Tell her to cook you so I 
can eat you for lunch." 
Mr. Turkey went off and came to where 
Mrs. Wildcat was. She was pushing a cradle 
and singing. Mr. Turkey said, "Your 
husband over there sent me to tell you that 
you should make some parched corn for me. 
After you've made it I'll go along." She got 
the corn, made the parched corn for him 
quickly, and he left. 
He went some distance, until he came to 
a place way over there where a little deer 
was lyinK by a big log. Mr. Turkey climbed 
up on top. The little deer yawned and said, 
"kakkudikE;nu?,nu?." \[This means in Caddo 
"I'm sleepy." The last syllable, nu?, which 
the deer repeated, means "turkey".\] "Oh," 
said the turkey, "he knows my name!" The 
deer kept yawning and repeating, 
"kakkudik~:nu?,nu?." 
Just at that time Mr. Wildcat arrived 
home. "Did you cook him?" he asked. "What 
do you mean?" asked Mrs. Wildcat. "The 
turkey I sent." "No. He came and said I 
should fix him some parched corn. After I 
made it for him he put it on his back and 
left." 
Mr. Wildcat followed him and caught up 
with him some distance away. 
That is why he eats him raw. 
The initial chunk from the speaker's 
memory, the chunk which constitutes the 
entire story, seems first to have been 
broken down into the plot-plus-moral schema. 
That is, the speaker knew -- or 
decided -- that this was a story to be 
organized into a plot followed by a brief 
sentence in which the plot was said to be an 
explanation of some well known general 
phenomenon; in this case the fact that 
wildcats eat turkeys raw. There is much to 
be said regarding this initial schema, 
including the possibility that it is not a 
native American schema at all, but one that 
was introduced from Africa by way of Uncle 
Remus type stories. But the point here is 
that this story is one instance of a schema 
that came to be very frequently used among 
these and other Indians. 
The first breakdown of the plot itself 
is into five episodes, each clearly 
definable in spatio-temporal terms. That 
is, each has to do with something that 
happened at a coherent location during a 
coherent segment of time. It may be that 
the schema to which the organization of 
these episodes conforms is unique to this 
particular story. On the other hand, 
further study of Caddo stories might show 
that the set-up in Mr. Wildcat's garden, 
followed by the trick played on Mrs. 
Wildcat, followed by the diversio~ with the 
deer, followed by the discovery of the 
trick, and ending with the disposition of 
the turkey conforms to a more generally used 
schema. It would not, at least, be 
surprising to find other stories in which 
these elements are put together in the same 
way. 
In any case the schema by which each of 
these episodes is itself broken down is one 
that is used very frequently. This is what 
may be called the visit schema. In it, 
protagonist A is engaged in some background 
~ctivitv as the episode opens. Then comes 
protagonist B's arrival on the scene. There 
follows a conversation between the two. 
Then some action is taken, usually by 
protagonist A but affecting protagonist B, 
and finally there is the departure of the 
latter. This visit schema appears over and 
over, not only in this story but in numerous 
other stories that these people tell. 
Furthermore, it appears to be a favorite 
pattern for the organization of much of 
Caddo real experience, in which visiting is 
an extremely common activity. Certainly it 
is of considerable interest that schemata 
are not just ways of' organizing what one 
talks about, and not even just ways of 
interpreting .reality, but also ways of 
organizing one s everyday behavior. 
Given this schematic organization, we 
might also use this example to glance at 
framing and categorization, the other two 
verbalization processes involving choice. 
As an illustration of framing we can take 
the clause in the last sentence of the 
second paragraph which reads, "\[she\] made 
the parched corn for him." In this version 
the speaker chose to frame the event in 
terms of an agent (Mrs. Wildcat), a patient 
(the parched corn), and a beneficiary (Mr. 
Turkey). At another time when this speaker 
told the same story she expressed the same 
event with the words, "She pounded it." Here 
the frame was different, being no longer 
benefactive but expressing rather an action 
performed on an object. And in this version 
a different categorization was chosen also. 
The event was interpreted as an instance of 
pounding, not of making. 
But to return to schemata, there are a 
number of unanswered or only partially 
answered questions regarding schematization, 
of which I will mention here only three. 
First, what are the essential ingredients of 
a schema? Studies of actual verbalizations 
suggest that a schema must define a set of 
event (and situation) types (such as 
background activity, arrival etc). A chunk 
of (direct or recalled) experience is 
evidently interpreted as an instance of the 
schema when some of the events included in 
the chunk conform to these types. How a 
person establishes such conformity is not at 
all clear, but presumably the process is 
similar to that involved in categorization. 
A schema must also involve the relations 
between these included events: temporal and 
causal relations, for example. In the visit 
schema the arrival must occur while the 
background activity is in progress, and so 
on. One would like to know more about the 
kinds of relations that are found here. And 
the chunks within a schema quite clearly 
have different degrees of salience. In the 
visit schema the arrival and conversation 
events seem to have the highest salience, as 
judged by the fact that they are usually 
mentioned while other elements may not be. 
It would be nice to know more about the 
determinants and effects of salience. 
The second question is best understood 
against the background of another example. 
(One problem with discussing discourse is 
that each example occupies an entire page.) 
90 
I 
I 
I 
I 
I 
I 
i 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
London Bridge 
Once there was a man in Winchester who 
dreamed that if he went to London Bridge he 
would hear something that would be very 
useful to him. A week later he had the same 
dream again, and a week after that he had 
the dream a third time. So he decided he 
would see whether there was anything to it. 
He traveled to London and stood on London 
Bridge for many hours, until his patience 
was nearly exhausted. Finally he was 
accosted by another man who asked him why he 
had been waiting there so long. After some 
hesitation, he told about his dreams. The 
other man laughed, and said that he too had 
had a curious dream the night before, to the 
effect that if he went to a town named 
Winchester and dug under a certain tree he 
would find a pot of gold. He ridiculed the 
whole idea, however, and said that he did 
not even know where Winchester was. The 
first man decided there might be some 
advantage in secrecy, and said that he did 
not know either. But he immediately 
returned home and dug under the tree which 
the man in London had described. He found a 
pot full of gold, and on the lid of the pot 
was an inscription in a language which he 
did not understand. The pot together with 
its lid were put on display at the village 
inn, and one day a scholar stopped at the 
inn and was able to translate the 
inscripion. It said, "Look lower; where 
this stood, is another twice as good." The 
man who had found the pot heard of this and 
returned to the tree, where further digging 
uncovered another pot which contained twice 
as much gold as the first. 
At the highest level this story 
consists of two major parts, the second of 
which begins with the words, "and on the lid 
of the pot was an inscription..." If the 
story ended before these words (as in fact 
in some versions it does), it would still be 
a perfectly good and complete story. But of 
special interest is the fact that there is a 
schema in the first part which is used again 
in the second. This schema consists of a 
puzzle being presented (through the dreams 
or through the inscription), of a solution 
to the puzzle being provided by some 
stranger, and of a payoff resulting from the 
solution. But this schema seems less 
obvious than the visit schema in the Caddo 
story. It is not so directly defined by the 
overt events, and is something that people 
who hear the story are not likely to be 
immediately aware of. It is also less 
directly reflected in the way the story is 
told. (One may note that the two instances 
of this schema, the two major parts of the 
story, are not even separated by a sentence 
boundary, let alone a paragraph boundary.) 
The question, then, is whether some schemata 
are more abstract than others. Again, the 
determinants and effects of the degree of 
abstractness are not wholly clear. 
The third question is a methodological 
one. I don't believe it is realistic to 
search for a discovery procedure for 
schemata. Given a particular verbalization 
91 
there is unlikely to be a recipe anyone can 
apply to it to come up with a schematic 
analysis. Like other theoretical entities, 
a schema takes imagination and intuition to 
ferret out. But we might nevertheless hope 
that there are objective ways of validating 
and demonstrating the existence of a schema, 
once hypothesized, so that different 
investigators will agree. I gave the London 
Bridge story to a class of fifty people who 
had been told a little bit about what 
schematc analysis involves, to see how many 
of them would come up with the 
puzzle-solution-payoff analysis to which my 
own intuition had led me. Just over one 
quarter of the class produced analyses 
essentially identical to mine. Probably the 
percentage could be raised with further 
training of the analysts. But one is left 
with the question of how objective any such 
analysis is, and how the methodology of 
schema research can be made to conform to 
acceptable scientific standards. 
