THESIS

mullIlllllllllllllllmlllglllwill:

This is to certify that the
thesis entitled
A TAGMEMIC ANALYSIS OF CONVERSATIONAL EXCHANGES
IN A SNAHILI FOLKTALE

presented by
Robert J. Dlouhy

has been accepted towards fulfillment
of the requirements for

a Master's degree in Linguistics

Major professor

 

Date %-4/ 4'73, @7/
/ / /

 

 

MSU

LIBRARIES
m.

 

 

 

300 A2 if"

 

RETURNING MATERIALS:
Place in booE drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

ll

 

A TAGMEMIC ANALYSIS OF CONVERSATIONAL EXCHANGES
IN A SWAHILI FOLKTALE

By
Robert J. Dlouhy

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Linguistics,
Oriental, and African Languages

1981

'l‘

ABSTRACT

A TAGMEMIC ANALYSIS OF CONVERSATIONAL EXCHANGES
IN A SWAHILI FOLKTALE

by
Robert J. Dlouhy

This study attempts to apply the revised tagmemic theory of Pike
and Pike (1977) to the analysis of narrated conversational exchanges in
a Swahili folktale. This revised theory does not allow the out-of-
phase relations between deep and surface structures that Longacre (1968)
demonstrates at the exchange level, because structure (surface relations)
and function (deep relations) are considered to be features of every
tagmeme. This study demonstrates that out—of-phaseness need not be
called upon as an explanatory device if the associations between
structures and functions are analyzed in sufficient detail. Such detail
is obtained by allowing any constituent of a turn to be labeled for the
exact function it carries in the exchange. The resulting detailed
analysis reveals that exchanges are linked to each other by shared
constituents, and that linkage can account for out-of-phaseness. An
emic analysis of the exchanges demonstrated that, for Swahili, function-
al relations within certain complex exchanges may be made explicit if
exchanges are viewed as capable of having exchanges embedded within
them. Embedding aids the explanation of the relationship among the
emic exchange classes, which, in Swahili, may be seen to form a hyper—

class.

ACKNOWLEDGMENTS

I would like to express my sincere gratitude to the members of my
committee, Dr. Carol M. Scotton, Dr. Barbara Abbott, and Dr. Ruth M.
Brend, for their helpful criticism and patience during the course of
this study. A special thanks is due to Dr. Brend for her openness to
my new and sometimes strange ideas about tagmemics. Special thanks
also go to my friend Mr. Darryl A. Salisbury, who spent many a long
night typing the diagrams and text of this thesis. My deepest thanks
are extended to my wife Carrie, without whose love and support the

project would have been impossible.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES .

1.

INTRODUCTION
1.1. Scope of this Study . .
1.2. Discourse Analysis as a Field of Study.

1.2.1. Discourse Analysis and Related Disciplines .
1. 2. 2. Approaches to Discourse Analysis .

1.3. A Swahili Folktale: Kisa gha_Punda wa_Dobi

1.3.1. Cultural Background

1. 3. 2. Swahili Literature .

1. 3. 3. Source of this Tale

1. 3. 4. Summary of the Story .

1.3.5. Further Comments on the Choice of this Story
TAGMEMICS
2.1 The Tagmemic Approach to Discourse Analysis
2.2 Discourse, Reference, and Function .
2.3 Further Comments on the Nature of the Discourse Level .

2.3.1. Hierarchical Organization at the Discourse Level .

2.3.2. Nuclearity and Marginality
2.3.3. Further Comments on Tagmemic Function

2.4. Methods of Analysis

2.4.1. Etic Description
2.4.2. Emic Analysis

2.5. An Example of an Etic Textual Analysis .
2.5.1. The Analytic Process
2.5.2. Discussion of the Analysis
ANALYTICAL PROBLEMS

Structural Labels .

Quoting Structures .
Problems with Function Labels .
Longacre' 5 Analysis of Exchanges .

wwww
#wNi—I

3.4.1. Longacre' s Deep and Surface Structures

3.4 2 Terminology for Exchange Structure and Function
3.4.3. Exchange Types . . . . . . . .
3 4 4 Out-of—Phaseness

‘ vi

I—|

«000000101 0'! (ON NI—I

3.5. Critique of Longacre's Treatment of Exchange Structure

3.5.1. Loss of Detail .
3.5.2. A Problem with Out- of— Phaseness
3.5.3. Modifications to Longacre' 5 Scheme

4. ETIC ANALYSIS OF EXCHANGES IN KISA CHA PUNDA WA D031

4 1 Etic Diagrams for Story and Exchanges
4.2. Exchange Linkage .
4 3 Structural Out- of- Phaseness

5. AN EMIC ANALYSIS OF EXCHANGES IN THE FOLKTALE

Contrasting Exchange Types . . .

Emic Exchange Classes as Forming a Hyperclass

Emic Exchanges in the Structure of the Folktale .
Exchange Structures and Linkage Type . . .
Normal vs. Non- Normal Class/Function Relationships .

0101010101
. . . . .
01$me

6. SUMMARY AND CONCLUSIONS .

Summary .

Findings Concerning Swahili
Conclusions .

Significance of the Study

01010101
thI—I

APPENDIX I: Text of the Folktale .
APPENDIX II: Abbreviations .
APPENDIX III: Etic Analysis
APPENDIX IV: Emic Analysis .

LIST OF REFERENCES .

TABLE 1:
TABLE 2:
TABLE 3:
TABLE 4:
TABLE 5:
TABLE 6:

LIST OF TABLES

Emic Exchange Classes

Classification of Emic Exchange Classes

Possible Expansion Sets of Exchange Hyperclasses
Exchange Structures vs. Linkage Type

Type of Linkage vs. Structure Linked

Normal vs. Non-Normal Class/Function Relations

. 63
. 63
. 66
. 69
. 69
. 71

FIGURE
FIGURE
FIGURE
FIGURE
FIGURE
FIGURE

FIGURE

FIGURE

FIGURE

FIGURE 10A:
FIGURE 108:
FIGURE 10C:
FIGURE 11:
FIGURE 12:

00

9:

LIST OF FIGURES

Paragraph 1 of PND Divided into Clauses and Translated
Discourse Structure for Paragraph 1 of the Folktale
Alternative Analysis of Sentence 1.3a. .

Alternative Analysis of Sentence Cluster 1.2.
Paragraph 2 of the Folktale, Original Analysis

Paragraph 2 of the Folktale, Revised to Include q. f. 's
in the Exchange Structure

Paragraph 2 Relabeled for Longacre' s Slots and
Functions . . .

Etic Analysis of Paragraph 2 Using Modifications to
Longacre' 5 Scheme .

Shared Linkage .

Partially Shared Linkage

Exchange 4.1.—4. Surface Structure (Longacre's Model)
Exchange 4.1.-4. Deep Structure (Longacre's Analysis)
Types of Linkage

Structural Out-of-Phaseness

vi

28

. 30
. 31
. 33
. 38

. 41

. 46

. 53
. 57
. 58
. 58
. 58
. 59
. 6O

1. INTRODUCTION

1.1. Scope 9f_this Study

This study offers a tagmemic analysis of conversational exchanges
found in the narrative text of a Swahili folktale, Kisa cha Pundb wa
Dobi (The Story of the Washerman's Donkey). Tagmemics, because of its
orientation to the hierarchical structure of language, is well suited
for the analysis of discourse and provides an analytical method which
allows consistent analysis of both higher— and lower-level grammatical
structures. Tagmemic research on conversational exchanges has already
been undertaken by Longacre (1968, 1976) and by Klammer (1971).
Longacre's work calls for the separate analysis of functional (deep)
and structural (surface) relations, and represents a departure from
Pike's tagmemic scheme. Klammer (1971) attempted to reconcile Longacre's
findings with Pike's work by viewing Longacre's deep structure relations
within the Pikean lexemic hierarchy. During the 1970's, however, Pike
merged part of his lexemic hierarchy into the grammatical hierarchy,
and established the new referential hierarchy into which are encoded
relations between cultural or real—world entities. Klammer's work,
although still relevant to exchange analysis, has become dated as a
consequence. This merger poses a further problem. Previous analyses
by Longacre (1968, 1976) have shown that the deep (lexemic) and surface
(grammatical) structures do not always coincide, and may be "out-of-
phase” with each other at the higher linguistic levels. Merging the
grammatical and lexemic hierarchies as proposed by Pike implies that
out-of—phaseness can no longer be tolerated. This study attempts to
solve this problem by allowing Longacre's deep structure labels for

function to name grammatical roles in Pikean tagmemes at the exchange

1

2

level. The resulting analysis shows the structure of exchanges in great—
er detail than those of Longacre, and allows a clear view of both link-
age between exchanges, and embedding of exchanges within other exchanges.
This improved analysis directly associates the structures of an utter-
ance with their functions, and thereby eliminates the need to explain
the relationships between structures and functions in terms of out-of-
phaseness.

The remainder of this chapter will outline some recent trends in
discourse analysis, and introduce the Swahili folktale. Chapter Two
will introduce tagmemic theory, and discuss methods of emic and etic
analysis. The third chapter will discuss some problems which arose in
the original etic analysis. Certain changes to the analytic approach
will be proposed here. Chapters Four and Five will discuss the findings
of the analyses (etic and emic) of the folktale. The findings of this

study will be summarized in Chapter Six.

1.2. Discourse Analysis as a Field of Study

1.2.1. Discourse Analysis and Related Disciplines

For purposes of this study, Pike and Pike's view of discourse
analysis as the study of linguistic organization at or above the sentence
level will be adopted (Pike and Pike, 1977). The range of such study is
obviously quite broad, and may include such topics a paragraph structure,
relations between paragraphs, dialog structure, plot marking, and extra-
sentential influences on sentence structure. This breadth contributes
to the usefulness of discourse analysis for application to syntactic,
textual, and semantic problems.

Before continuing to outline the range of these applications, it

3
would be useful to contrast discourse analysis with the traditional
disciplines of rhetoric, stylistics, and criticism. In many cases
discourse analysis may share the same object of study as these older
disciplines, namely, the structural analysis of texts, or the evaluation
of these structures for their effects. Characteristically, however,
modern discourse studies are closely bound to a linguistic theory of
one type or another, and serve the purpose of shedding light on lower-
level phenomena such as stylistic transformations, or describing text
in terms of a particular theory, as this study attempts to do. As a
result, both the means and ends of current discourse analysis are quite

different from work in the traditional areas.

1.2.2. Approaches to Discourse Analysis

It seems that some treatment of discourse has been attempted by
proponents of almost every major linguistic theory in the United States
of America and Europe. One of the earliest treatments of discourse in
relation to syntax was by Mathesius (1939), and his work, along with
that of other members of the Prague School such as Danes and Firbas has
contributed to the development of the Functional Sentence Perspective
(FSP). With FSP, as described by Palkova and Palek (1978), is concerned
with the potential communicative function of a sentence in a particular
textual or situational environment. It has been brought to bear on
generative syntax in this country through Kuno's (1972) development of
functional syntax. Kuno (1978) uses discourse-based FSP principles to
act as constraints on Gapping operations.

Much of the current work relating discourse to generative grammar is

moving in the direction of showing how discourse factors influence

4
structures at the sentence level. There is much recent work exploring
the influence of discourse factors on pronominalization, of which a
study by Bolinger (1979) on the extra-sentential influences on choice
between full NP and pro-form is a representative example. Other syntac-
ticians are now looking to discourse features and constraints for moti-
vation of certain syntactic operations. The work by Erteshik-Shir
(1979) on dative-movement is representative of this approach. A number
of linguists, notably van Dijk (1972), have been much more ambitious,
and have proposed that a generative grammar could be designed to account
for textual as well as syntactic structure. Rieser (1978) explains that
under such an expanded model a text symbol could be rewritten into sym-
bols for sentence sequences, which could, in turn, be rewritten in
sentence symbols. He reports that generative text grammars have been
modeled in terms of interpretive semantic theory (Isenberg, 1971), and
generative semantic theory (Petofi, 1973).

The treatment of discourse plays an integral role in Systemic-
Functional (SF) linguistic theory. In a recent review of the SF
approach to text linguistics, Hasan (1978, p. 228) states that the
concepts of texture and structure are fundamental to the analysis of
texts. Texture involves cohesive devices such as conjunction, ellipsis,
and substitution which cause a series of lexical-grammatical structures
to be percieved as related to each other. Structure, on the other hand,
concerns those principles of textual organization which allow the hearer
to understand whether or not a text is complete. Hasan asserts that the
structure of discourse type (genre) is determined by its contextual con-
figuration, the features of the environmental situation affecting the

discourse. These features bear some resemblance to tagmemic role in the

5
higher levels of the tagmemic grammatical hierarchy, and in the referen—
tial hierarchy also.

Sociolinguists and psycholinguists have also attempted treatments
of discourse. Studies concerning the structure of conversation are
naturally in the domain of sociolinguists, but they represent a form of
discourse analysis as well. Schegloff's (1979) article about the in-
fluence of repair on the syntax used in a conversation is representative
of such studies. In this article, Schegloff suggests that repair, the
mid-sentence changes of syntactic course that result in pauses, "uh's,"
and restarts, affects the form of sentences and hence alters the cues
presented to listeners used for marking and turn-taking opportunities.
Psycholinguistic studies are often concerned with discourse as well.
Chafe (1979), for example, has put forth a "flow model“ to explain the
relation of thought to language. The model suggests that foci, or units
of stored information, are clustered to form thoughts. As a person
"thinks" he moves from focus to focus and may eventually move to a
different cluster of foci. Such a jump represents a sharp change in
coherence of the items being thought about, and, if such thoughts are
being expressed in speech, this jump may correspond to a boundary of a
discourse unit. In this manner, cognitive structure may contribute to

the structure of discourse.

1.3. A Swahili Folktale: Kisa aha Pundb wa Dobi

1.3.1. Cultural Background

Swahili is known today as a lingua franca throughout East Africa.
It origniated as the dialect of Zanzibar town on the island of Zanzibar

just off the coast of Tanzania (Whiteley, 1969). A number of languages

6
closely related to Swahili stretch along the Indian Ocean coastline
from southern Somalia to northern Mozambique. This dialect area has
long been distinct from cultural groups in the continental interior due
to trading contacts with Arabs, and the settlement and assimilation of
Arabs in coastal ports. The Afro-Arab culture based on trade contacts
with the outside world has become known as the Swahili culture. The
term ”Swahili culture“ is difficult to apply, however, since there is a
considerable amount of cultural diversity and social stratification
within the area (Ismail and Lienhardt, 1968). Swahili culture is quite
old, dating at least to the 11th or 12th century. The oldest mosque in
Zanzibar dates to 1107 A.D. (Knappert, 1971).

The East African coastal area was culturally distinct from the
interior in many ways. Coastal residents had greater material wealth
due to imported goods such as Chinese china. Economic life centered
around towns was also an innovation not found in the interior.
Linguistically, the Bantu dialects along the coast lost their tone and
acquired many loans from Arabic. The introduction and widespread adop-

tion of Islam also distinguished this area from the interior.
1.3.2. Swahili Literature

The Islamic religion had a great influence on the development of
Swahili literature. Historically, much Swahili poetry functioned to
express Islamic values, and was patterned after Arabic models (Knappert,
1968). There are many literary genres in Swahili, the most famous of
which is utenzi, a peotic form used for both religious expression and
epic narratives (Knappert, 1968). The earliest known Swahili verse

the Hamziya, is in this form, and dates from the 17th century. Other

7
genres include songs, riddles, proverbs, and animal fables (Knappert,
1971). The story subjected to analysis in this thesis, §Z§g_gflg_§ygég
gghgg§i_(Ih§_§tgry‘gfithg_Washerman's Donkey), belongs to the latter
class. Swahili literature continues to thrive in modern times, as
evidenced by the work of the late Shaaban Robert and others.

Knappert (1968, p. 7) argues strongly that the form and content of
most traditional Swahili poetry come from Arabic, Persian, or Indian
sources. Other scholars are able to clearly demonstrate African in-
fluences in more recent Swahili literature, as seen in Lienhardt's
study of Ismail's §yéfg_ag_NguvumaZi, a modern epic poem about medicine
men (Ismail and Lienhardt, 1968). In light of this debate, it is inter-
esting to note that Kisa aha Pundb wa Dobi bears a very close resemblance

to some tales found in the Sanskrit Panchatantra. The Sanskrit stories
without Heart 9:_Ears, found in Book IV of the Panchatantra (cf.
Edgerton, 1924). The Swahili tale studied here contains two stories,

one embedded in the other. The Swahili matrix story is nearly identical

in Sanskrit. Interestingly, the Panchatantra also has the story of the
ass embedded in the story about the ape and crocodile, so the overall
structure between the tales in the two languages is similar. The
Swahili tale substitutes a shark for the crocodile in the matrix story,
and a rabbit, the traditional African trickster, for the jackal in the

embedded story.

1.3.3. Source of this Tale

The tale §§§a_gflg_§g§g§_gg_§9§§_was collected by the Reverend
Edward Steere, Bishop of Central Africa, in Zanzibar town sometime
during the 1860's. Steere was among the first Europeans to study
Swahili, and his collection, Swahili Ial§§_(1870), was probably the
first Swahili literature translated into English. Steere produced
descriptions of a number of African languages, and did some translation

of the Gospels (Heanley, 1909).

1.3.4. Summary of the Story

Donkey) goes like this: A monkey, kima, and a shark, papa, become
friends because the monkey drops fruit to the shark from a fig tree
whose branches spread over a harbor. One day the shark offers to repay
the monkey's kindness by entertaining him. After promising not to get
the monkey wet, the shark takes him on his back and procedes homeward.
Halfway there, the shark tells the monkey the truth: his sultan is sick
and needs a monkey's heart to cure his illness. The monkey, realizing
that he is almost a goner, tries to deceive the shark; he says that
monkeys always leave their hearts in the trees when they come down to
the ground, and he didn't bring his heart with him. The shark believes
this and returns to the tree. The monkey escapes into the tree and will
not come down when the shark calls. When the shark asks why he won't
come down, the monkey asks, "Do you take me for a washerman's donkey,”
and the shark, of course, replies, "What's that?” The monkey then tells
the story of the washerman's donkey: A washerman, dbbi, had a donkey,

panda, which ran away and got fat. The hare, sungura, sees the donkey

9

and makes a plan for the lion to kill it so the two of them may have a
feast. The lion agrees to this plan, and the hare tells the donkey (a
female) that the lion wants to marry her. She agrees, and they go to
the lion's house. The lion tries to kill the donkey, but because he is
weak from a recent illness, she escapes. After some days, they try
again. The hare convinces the worried donkey that the lion was not
trying to kill her, it was only the way he converses. She agrees to
return, and is immediately killed. The lion tells the hare to roast
the meat, and save him the heart and ears. The hare roasts the meat
and feasts, and hides the leftovers. When the lion comes for the heart
and ears, the hare tells him that this was a washerman's donkey and
consequently had no heart and ears; after all, if it had a heart and
ears, would it have returned a second time? The monkey tells the shark

that he will not become a washerman's donkey; their friendship is ended.
1.3.5. Further Comments on the Choice of this Story

This story was chosen as the subject of analysis for a number of
reasons. The most important of these was that it exhibited a rather
complex structure that involved the embedding of one story within
another. Tagmemic theory makes claims about being able to express such
recursiveness, and it was felt that a story having this structure would
provide a good test of the theory. Another reason for choosing this
story was that it was provided with a good translation. Finally, it was
felt that the story would represent a form of Swahili that was free from
modern stylistic influences from the western world, since it was collect-

ed before the European cultural influence in the area was very strong.

2. TAGMEMICS

2.1. The Tagmemic Approach tp_Discourse Analysis

Because of their interest in translation, tagmemicists have long
paid attention to the structure of discourse. The result of this atten-
tion is a theory in which the treatment of discourse is well integrated.
To explain the nature of this treatment requires some background of the
tagmemic view of language; what follows, therefore, is a brief summary
of tagmemic concepts as they relate to the treatment of discourse.
Whenever possible, comments on the developmental history of these con-
cepts will be provided.

Kenneth L. Pike is known as the founder and prime mover of tagmemic
theory. Pike has stated that tagmemics originated in the period between
1947 and 1949 when he started a search for a unit of grammar that would
be analogous to the phoneme (Pike, 1976). The outcome of this search
was the creation of the tagmeme, as well as a new theory of language,
tagmemics.

The tagmeme was analogous to the phoneme in many respects. Like the
phoneme, it exhibited variability, being realizable in a number of ways.
The subject of a sentence, for example, might be realized by a pronoun,
noun, noun phrase, or clause. Also, like the phoneme, a tagmeme had
distributional constraints. The subject of a sentence can only occur in
certain positions in a sentential construction, for example. These two
characteristics allowed the tagmeme to be expressed as a slot-plus—filler
unit where the slot reflected the position of the unit in its including
structure, and the filler reflected the different realizations of the
unit. Finally, the tagmeme and phoneme were similar with respect to

their abstract nature. Just as a given phone, say [b], is not the same

10

11
as the phoneme it realizes, /b/, a phrase such as in Kalamazoo in the
sentence I have a girl in Kalamazoo is not the same thing as the ad-
verbial tagmeme which may be used as part of an abstract description of
that sentential construction. The phrase is a realization of that tag-
meme.

As a constituent of a larger construction, the prepositional phrase
in the example above illustrates another point which Pike had to treat,
namely that constructions could include other constructions in their
structure. This presented little problem, because any construction, no
matter what structure it is included within, can be described in terms
of constituent tagmemes. This solution, as a result, gave Pike a means
of describing hierarchical structure in the sentence.

However, some changes away from the prevailing view of hierarchical
grammatical structure were made as tagmemics developed. Pike (1976, p.
96) states that he made an important breakthrough when he rejected the
idea that a grammatical unit must have at least two immediate consti-
tuents, and concurrently adopted the view that a grammatical unit must
fill a “functional position" in a construction (Pike, 1974). This meant
that the tagmeme, along with having a place (slot) and shape (filler),
had a function in the construction which included it. This function was
constant for the slot and its fillers, and expressed in part the rela—
tion between the tagmeme in question and other tagmemes in the construc—
tion. Thus, to call a noun phrase a "subject” of some clause type is to
say that the noun phrase bears some special relation to the other consti—
tuents of that clause. It should be noted that the tagmeme carries the
function but the construction that fills it does not. Thus noun phrases

and prepositional phrases do not have functions in their own right, but

12
only as realizations of tagmemes. The function of the prepositional
phrase in I have a girl in Kalamazoo is different from that in The girl
in Kalamazoo is my wife. Pike relates that this change was significant
because it caused him to focus on the functional relation between a tag-
meme and its including construction rather than on its immediate consti-
tuents. This ability to look upward for a functional relation to an in-
cluding construction was later to have great importance for the treat-
ment of discourse.

Another useful consequence of allowing grammatical function to be a
feature of the tagmeme was that this feature could be very useful in dis-
tinguishing between constructions in cases where their structural differ—
ences were minimal or non-existent. For example, in the sentences John
hit the ball and John was hit by the ball, the subjects have different
functions. The Subject of the first performs the action, while that of
the second receives the action. These differences in function, together
with some significant structural differences, contribute to the contrast
between the two constructions.

It might appear that the analogy between the phoneme and the tagmeme
had broken down at this point, since no mention has been made of phono-
logical constructions, phonological function, or contrast between tag—
memes. Consistency between the phoneme and the tagmeme was maintained
as an outgrowth of Pike's early realization that a phonological hier-
archy based on the phoneme - syllable - stress group - etc., could be
formed (Pike, 1976). This concept was radically different from the then
current notion of linguistic hierarchy which had phonemes constituting
morphemes, morphemes constituting phrases, etc. A re-evaluation of

grammatical and lexical structures satisfied Pike that three

13
semi-autonomous, simultaneously realized hierarchies, the phonological,
grammatical, and lexical, could be posited. This view was set forth in
Hpmgp_Behavior (1954). Eventually, from the influence of works such as
Crawford's (1963) in phonology and Wise's (1968) in lexemics, it became
apparent that tagmeme-like units could be posited in the phonological
and lexemic hierarchies. The phoneme became the minimal unit in the
phonological hierarchy, but phonological tagmemes analogous to gramma—
tical tagmemes were allowed. In this way a closer correspondence be-
tween the two hierarchies was established.

The general scheme of language presented in Volume I of Language
tagmemicists. In this scheme, discourse structure is seen as structure
above the level of the sentence in the grammatical hierarchy. Discourse
structure is therefore a type of grammatical structure, as recently ar-
gued by Longacre (1979). Discourse constructions may be hierarchically
organized, having sentences within paragraphs, and paragraphs within
monologs, for example. These constructions may be described and con-
trasted in terms of their tagmemic constituents with their features of
slot, class (filler), and role (function). (Tagmemes may also have a
feature of cohesion which expresses co—occurance restrictions.) Dis-
course analysis parallels the grammatical analysis of constructions at
lower levels. This concept of discourse is allowed by the view of
function in hierarchy mentioned earlier. For every construction it is
possible to ask what larger structure includes it, and what function
the lower-level unit plays within the higher-level one. If this ques—

tion is asked repeatedly, sooner or later linguistic acts are seen as

14
parts of non-linguistic behaviors. Functional units of behavior were
seen by Pike as parallel to linguistic units, and called behavioremes in
Language... (1954).

Discourse has therefore been an integral part of tagmemics since
the theory began. In the first volume, first edition, of Language...
(1954), Pike discussed discourse genre as "utterance types" such as the
sermon, the after-dinner speech, a cantata, a joke, and so forth. One
of the earliest descriptive treatments of discourse was Loriot's study
of Shipibo paragraphs (1958) which treated the different ways sentences
could be linked together to form paragraphs of various functional types.
Longacre (1976, p. 8) states that this paper was very influential among
tagmemicists during the early sixties, particularly with respect to
Powlison's study of Yagua paragraphs and Gudshinsky's consultation work
for the SIL in Brazil. During the middle sixties, Waterhouse (1963) and
Pike (1966) contributed further to the study of paragraph structure by
producing reports on the dependencies between sentences in a text.
Discourse genre, as opposed to paragraphs, were first treated comprehen-
sively by Longacre (1968) and his colleagues at a workshop he conducted
in the Phillipines. This workshop produced many significant findings,
among the most important being a classification and analysis of features
for four different discourse genre: narrative, procedural, horatatory,
and dramatic. The first tagmemic treatment of conversational exchanges
is represented by the work on dialogue in narrative and dramatic dis-
course done at this workshop. Pike and Pike (1977) and Longacre (1976)
represent the most recent major theoretical treatments of discourse from
the tagmemic standpoint. Pike and Pike provide a useful etic (descrip-

tive) list of roles (functions) for tagmemes at different hierarchical

15
levels of discourse, and both they and Longacre pay increasing attention

to the relation between discourse and reference.

2.2. Discourse, Reference, and Function

Probably the least developed aspect of tagmemic theory, and the
aspect that has been subject to the greatest theoretical modification,
is the treatment of the referential hierarchy. This hierarchy was ori-
ginally called the lexical hierarchy by Pike (1954). It functioned in
part to express the taxomonic relations between real world object, Fido
is a dog; a dog is a canine; a canine is a mammal; etc., and the rela-
tions between paraphrased expressions, Fido; my dog; my four legged
frisbee catcher, all in reference to the dog that I own. Wise (1968)
expanded this concept beyond lexical items by showing that features such
as the chronological ordering of events are constant for a story despite
its grammatical form. She also proposed that participant roles similar
to Fillmore's (1968) cases could be encoded in this hierarchy. She re-
named the hierarchy the lexemic hierarchy, a revision which was accepted
by Pike during the late sixties and early seventies.

Klammer's work on conversational exchanges (1971) was done with
Wise's model of the lexemic hierarchy. The significance of Klammer's
work lies in his adaptation of Longacre's findings to the Pike-Wise
lexemic hierarchy. Longacre at this time (1968) had developed an analy—
sis of conversational exchanges in which “surface structures“ (grammati-
cal structures) were treated separately but parallel to ”deep structures"
(relations between constructs analogous to case roles such as actor, un-
dergoer, etc., at the lower levels). Longacre showed that deep and sur-

face structures were usually congruent, but sometimes diverged at the

16
higher levels of discourse, becoming out-of—phase, as he termed it.
Klammer was able to account for Longacre's case roles and deep structures
in the Pike-Wise lexemic hierarchy, and demonstrate out-of—phase lexemic-
grammatical structures as well.

Longacre maintains his deep-surface model to this day, but Pike and
Pike (1977) have extensively revised the lexemic hierarchy and renamed
it the referential hierarchy. This revision was apparently made so that
identities, events, purposes, and other forms of cultural knowledge
could be analyzed as they pertain to language. For example, what Pike
and Pike call the referential structure of a story is quite distinct
from its grammatical structure because if the story is told from a dif-
ferent point of view or with its elements in different order, the parti-
cipants, items, and events referred to in the story still maintain their
relationships. It is possible to tell a story backwards, but the chron—
ological relationships of its events will still remain the same. Thus,
the referential hierarchy may be viewed as expressing the relations be—
tween things (in the cultural world) referred to in a discourse.

Part of Pike's revision of the lexemic hierarchy involved moving
the account of deep case or participant role over to the grammatical
hierarchy. This transfer was due in part to Becker's (1967) development
of Pike's notion about the close relation between participant role and
grammatical structure, and it had the effect of establishing participant
role as a feature of the grammatical tagmeme. In the revised tagmemics,
grammatical tagmemes at all levels have a feature analogous to partici—
pant role. Pike calls this feature role, but I prefer to call it func-
tion because it is more descriptive and general, particularly for the

higher levels.

17

A number of points should be made about function. The first is
that it is in some sense a semantic feature of a grammatical unit (just
as case is), so the constituents of a grammatical construction are seen
as having semantic function within that construction. Second, the func-
tion of which Pike writes in his early works (cf. 1954) is not quite the
same as what I label function here. The original function represented
the relation of a constituent to its including construction. To use
Pike's example (1976), the relation between a and the and the noun
phrase in which they are included is constant in a boy and the boy. In
both cases a and the are determiners, and function to specify the noun.
In practice, slot names such as subject, object, etc., come to label this
type of relationship. In the past decade, the use of these labels has
lessened because it is difficult to find them for each level and they are
very easy to confuse with role labels. In this study they are eliminated
entirely. A third point is that function (role) is analyzable into com—
ponent features. Much recent tagmemic work has concentrated on finding
such features (see section 2.3.3.).

The shift of participant role or function from the lexemic to the
grammatical hierarchy raises a number of questions. First, since the
revision has dated Klammer's work, we may ask the general question of
whether or not the new scheme can account for the phenomena treated by
Longacre and Klammer. Secondly, we note that structure and function are
now united in the grammar and ask how we can specifically account for
the phenomenon of out—of—phaseness. This is the major problem treated in
the present study, and it will be shown that the revised scheme can ac-

count for the previously noted discourse phenomena.

18
2.3. Further Comments on the Nature of the Discourse Level

2.3.1. Hierarchical Organization at the Discourse Level

In the tagmemic framework, the upper grammatical levels are usually
considered to be those constructions above the level of the clause, so
what is studied here is the "supra—syntax“ of the story. Tagmemic theory
makes explicit claims concerning the existence of grammatical structure
above the sentence (cf. Longacre, 1979). Pike and Pike (1977) demon—
strate that this supra—sentential level is hierarchically organized in a
manner analogous to the lower grammatical levels. Although the exact
number of levels above the sentence is language specific, the Pikes
state that the basic hierarchical ordering of these levels is as follows:
sentence, paragraph, monolog, exchange, and conversation. A tagmemic
analysis of these levels would illustrate their constituent structure,
showing, for example, how sentences act as constituents of paragraphs,
how paragraphs act as constituents of monologs, and so forth. Both
Longacre and the Pikes realize that the hierarchical progression is not
as simple as stated above. Both have allowed level skipping and recur-
sion to occur within the hierarchy. Normally, we would expect a tagmeme
to be realized by a construction on the next lower hierarchical level.
When level-skipping occurs, the tagmeme is realized by a construction
which is on a still lower level. For example, a turn in conversation
does not necessarily require a monolog consisting of many paragraphs; a
single clause, or even a single word will do. Level-skipping such as
this is very common. 0n the other hand, a turn at conversation might
present the opportunity to present a narrative of some sort in which a
conversation is recounted. Telling a joke may represent such a situa-

tion. In this case, a conversation would be embedded in a monolog, and

19

recursion would occur. This phenomenon, like level-skipping, is very
common at the supra-sentential, or discourse level.

Both level-skipping and recursion are easily expressed tagmemically.
For example, each turn in a conversational construction is a constituent
of that construction and hence may be abstracted as a tagmeme. This
tagmeme may have fillers from monologs down to individual morphemes. If
a single morpheme is realized as the filler, level-skipping has occurred.
If a monolog is realized, and one of its constituent tagmemes is realiz—
ed as a conversation, then recursion has occurred. There is nothing in-
herent in the tagmeme which prevents it from being realized (filled) by
a construction that is of the same or higher level than the construction
of which the tagmeme is a constituent. The restrictions on the realiza-
tions of a tagmeme are due to the definition or specification of the
construction in which the tagmeme plays a role. A construction of a
given type K; which contrasts with some other constructions Y and Z, may
contain constituent tagmemes having different realizations (fillers) or

functions (roles) from those of Y or Z.

2.3.2. Nuclearity and Marginality

Pike and Pike (1977) apply two descriptive terms to the position
feature (slot) of the tagmeme: nuclear and marginal. The Pikes state
that a tagmeme is nuclear if it plays an essential semantic role within
the including construction, and if it tends to recur in all instances of
that construction. A construction in slightly varied forms might be used
to realize a number of different tagmemes; the unvarying constituents of
that construction would be nuclear to it. Thus, nuclear tagmemes tend

to be obligatory and functionally important to their including

20
construction. Marginal tagmemes tend to lack these qualities. A con-
stituent tagmeme of a construction must be either nuclear or marginal
with respect to that construction, and a construction may have more than
one nuclear and/or marginal tagmeme.

Two examples may serve to illustrate this point. First, at the
lower grammatical level, the constituents of the English noun phrase may
be considered. In a noun phrase, a noun is always found, determiners
are frequently found, and adjectives, prepositional phrases, and relative
clauses are not uncommon. No matter where the noun phrase is found, be
it a subject, object, prepositional object, etc., it always has a noun
and often has a determiner. These constituents are considered nuclear
to the noun phrase, while the others are marginal to it. From the func-
tional standpoint the noun plays a critical role since it is the "namer"
in the construction; it is saddled with the primary task of the construc-
tion. As a second example, the discourse level of a story may be consid-
ered. It will be noticed that certain constructions in the story present
information crucial to its development, while other constructions provide
background information, or make comments about the action or participants.
In most cases, the particular genre of the story and its essential seman-
tic content would remain unchanged if these comments were excluded. In
fact, different tellings of the story might include different comments,
or the deletion of some comment made in an earlier version, or entirely
new comments at points in the story where they were not included before.
Such comments are marginal to the structure of the story, while the less
changeable ones are nuclear to it.

Nuclearity and marginality are important characteristics of tag—

memes because they indicate the status of the tagmeme in relation to the

 

21

entire including construction. Determining nuclearity is part of the
process of determining emic (contrastive) construction types, since many
examples of the construction must be used, and contrast between construc-

tions may be based on the nuclear versus marginal status of tagmemes.

2.3.3. Further Comments on Tagmemic Function

Tagmemic function, or role, as it is called by Pike and Pike (1977),
is very important in tagmemic analysis. As mentioned in section 2.1.,
tagmemic function is important for determining contrastive construction
types. Section 2.4. will show that the identification of functional
units in a text is a very important part of determining that text's con—
stituent structure. Section 2.2. mentions that role expresses the gen—
eralized semantic function of the tagmeme in its including construction.
The identification of tagmemic function is therefore a crucial aspect of
both the theory and practice of tagmemics.

Although tagmemics provides a method for discovering the emic con-
struction classes of a language, it does not provide a unique method for
the determination of tagmemic function. Tagmemicists are very much aware
of this situation, and have been exploring the problem in recent years.
Most of this work has been done at the clause and sentence levels, how-
ever. Hale (1973) worked on role features for clauses of Indian and
Nepalese languages, while Longacre (1976) has developed an etic set of
case frames based primarily on English data. Sterner, Subharno, and
Pike (1976) worked on a feature system for role relations in complex
(multi-clausal) sentence constructions, and two-sentence clusters.

Their role feature system serves as the model for assigning roles to

sentence level constructions in the Swahili text under study here.

22

Pike and Pike suggest a list of possible tagmemic functions for
constructions at the higher levels of discourse. They do not maintain
that the list is comprehensive or exhaustive; tagmemic function might be
of any sort and take any label, and its description is left to the skill
and insight of the analyst. Among functions listed for sentence level
constructions they suggest question, command, wish, and statement. At
the paragraph level, they suggest topic, problem, illustration, hypo-
thesis, and many others. At the monolog level, they suggest instruction,
greeting, coordination, story telling, and many others. Many more func—
tions are suggested for the exchange and conversation levels.

The discussion above does not imply that each grammatical level has
its own exclusive set of functions. It is often the case that the same
function is found on more than one level of a hierarchy, especially if
the function is for a marginal unit. In the sentence 1.2. of the folk-
tale, for example, the function "item" occurs at all levels from sen—
tence cluster down to word (see Figure 2, page 30).

The tagmemic literature, with the exception of Fries (1970), has
not paid attention to the effects of level-skipping and recursion on the
roles assigned to constructions. Chapter Four will treat this problem
in detail, and demonstrate that a construction may have dual roles if it

functions as a constituent shared between two linked exchanges.
2.4. Methods pf Analysis

From the preceding discussions, it may be seen that a tagmemic dis—
course analysis of a text should accomplish two things. First, it
should shed some light on the constituent structure of the text itself

(etic analysis), and second, it should make a statement about the types

23

of constructions which are used in the text (emic analysis). The first
of these goals is readily achieved by examining the constructions which
make up the text, and determining their function. The second goal is
somewhat more difficult to achieve because it involves determining the
emic classes of constructions. A simple text would not be adequate to
achieve this goal for all structures in a discourse. However, if a type
of structure occurs frequently enough in a text, a limited but not
necessarily exhaustive set of contrastive classes may be discovered.
This study will present detailed emic and etic analyses of the narrated
exchanges in the folktale, but cannot make any broad claims about the
emicity of its exchange classes in relation to the rest of the language.

It should be noted here that the processes of etic and emic analysis
often appear to be arbitrary and ad hoc exercises in labeling units and
functions. The object of tagmemic analysis, however, is not to label
everything in sight, but to discover the least complex system of units
which account for the data. There is no mechanical process for doing
this, and the analyst must rely on what Longacre (1964) called the
"guess-and—check” method. In this process the analyst creates labels
freely, but constantly checks and refines them. The result of this re-
fining process is a system of units and labels (particularly for func-
tions of tagmemes) which account for the data in the most efficient way.
There is, of course, no guarantee that the analysis will be correct or
valid, but experience has shown that results for one language are often
relevant for other languages. A case in point is the set of functions
Longacre developed for turns in an exchange. This set was originally
developed on the basis of data from some Phillipine languages, but has

been applied with little modification to languages of Central America

24
and, in this study, to Swahili. This cross language relevance lends

some degree of validity to tagmemic theory and analysis.

2.4.1. Etic Description

Ideally, an etic analysis of a text's constituent structure would
be handled in much the same way as it is for clause-level constituents;
constituency would be judged by the substitutability of one unit for an-
other. While this is relatively easy to do at the clause level, it is
rather cumbersome for texts, since it is difficult to ask an informant
to substitute one paragraph for another. With an old written text such
as the one used in this study, this approach is almost impossible, since
the text is a static record of a discourse. A different approach must
be taken for the analysis of old written texts.

This alternative approach hinges on the basic assumption of tag—
memics that tagmemes have functions (deep or semantic roles). If the
function associated with a construction can be discerned, then the func-
tion of the tagmeme which the construction realizes is known, since con-
structions have functions only with respect to the tagmemes which they
realize. If the construction in its context can be identified as to its
type, then the three basic features of the tagmeme are known: its slot
(grammatical role), its function (semantic role), and at least one of
its fillers. For example, it might be possible to recognize a certain
section of a narrative text as acting to introduce or set the stage for
the story that follows. The function of the tagmeme which this section
realizes is now clear: introduction. The type of construction which
fills this tagmeme might be a paragraph, and it might be located at the

very beginning of the story. Thus, the tagmeme which represents this

 

25
constituent of the story has the features of introduction, filled by a
paragraph in story—initial position.

At the early stages of analysis very few functional units may be
recognizable. When one such unit is discovered, however, it is possible
to ask at least three questions: (1) In what construction is the identi-
fied construction included? (2) What other constructions are consti—
tuents of this larger construction? (3) What constructions act as
constituents of the identified construction? It may not be possible to
answer all these questions, but as often as not, they lead to the iden-
tification of a few more tagmemes. As more tagmemes are identified, the
questions repeat themselves, and gradually, by proceding upwards, down—
wards, and sideways in this manner, the constituent structure of the
text is fleshed out. The final result of the analysis is a description
in which the constituents are identified by their function, type, and
position. The analysis may be represented by a tree diagram in which
the nodes are labeled as construction types, and the branches are label—

ed for the function of the node to which they lead.
2.4.2. Emic Analysis

Once an etic description has been obtained, an emic analysis may be
undertaken to determine the constrastive construction types. An emic
analysis begins by sorting the constructions under study into groups on
the basis of their structural make-up and the functions of their con-
stituents. Constructions with the same order of constituents and having
the same function associated with each constituent are placed in the
same group. Any two groups contrast when their constituent structure

and internal functions vary from each other. Often, it is found that

26
one group appears to be some sort of expansion of another group; the two
groups may have a common core of constituents with the same order and
functions. If the functions of the core constituents are not altered
by the presence of the new constituents, the new constituents are con—
sidered optional, and the two groups are classed as one. When this sort-
ing is completed, tagmemic formulas for each construction class can be
made. Each tagmeme in a formula will be an abstract representation of
one of the constructions' constituents into which features of location,
function, possible fillers, and cohesion (co-occurrence restrictions)
are coded.

Occasionally, systematic relationships of structure or function may
be noted between contrasting classes of constructions. For example,
within Swahili clause roots we see a regular pattern of verbal deriva-
tion: kusoma (to read), kusomea (to read to), kusomewa (to be read to).
When fully inflected, each of the verb stems will form the nucleus of a
clause root that contrasts with others, even though the contrasting forms
are obviously related in a regular fashion. These related constructions
can be seen to form a paradigm or expansion set of clause roots. Such a
paradigmatic set of related yet contrasting constructions is called a
hyperclass in tagmemics. It will be shown in Chapter Five that the con—

trasting exchange classes in the folktale form a hyperclass.

2.5. An Example of an Etic Textual Analysis

An example of an etic tagmemic discourse analysis by means of func-
tion and construction identification will now be presented. This example
should amplify the points already made, provide a clearer idea of the

type of information a tagmemic analysis presents, and illustrate the

27
various options for labeling structures and functions that arise during

an analysis.
2.5.1. The Analytic Process

The first paragraph of the story will be the subject of this simple
analysis. This paragraph and its translation are presented in Figure 1
(page 28). We cannot tell if it is actually a paragraph; that decision
involves the contrastive analysis of construction types. It is seen
that this is a paragraph-like structural unit of some kind because it is
composed of sentences, and is distinct from other structures containing
dialogues that follow it. Also, the sentences that comprise it all have
a common purpose, to provide background information for the story. The
unit appears to be an introductory paragraph for the folktale. We could
say that its function is simply to provide an introduction to the folk-
tale, but a closer look at its constituent sentence—groups allows a re—
finement of that functional label. These constituents express the rela—
tions between the main participants, the location of the story, and the
relevant habitual actions of the participants. A more apt label for the
function of this paragraph might therefore be setting rather than intro-
duction. With these considerations in mind, we tentatively posit a para-
graph having the function setting for this position in the folktale.

The story has many constituents; one of them has now been isolated. It
is in story—initial position, has the function (seemingly) of stating
the setting, and takes the form (seemingly) of a paragraph.

We may now ask what constituents comprise this paragraph. The
orthographic form of the text provides four “sentences,” but detailed

examination reveals that each orthographic sentence contains more than

1.1a.

1.2a.

1.3a.

1.4a.

28

Aliondokea kima
he-went monkey
’% monkey went off"

akafanya urafiki na papa.
he—did friendship with shark
Hand made friends with a shark."

Pana mti mkubwa,
there was tree big
’flhere was a big tree,"

jina lake mkuyu,
name its fig-tree
’halled a fig-tree,”

umeota katika kilindi;
it-grew at deep water
”it had grown near deep water;"

matawi yake nusu yako mjini,
branches its half over town
’its branches were half over the town,"

na nusu yako baharini.
and half over sea
’hnd half over the sea."

Yule kima kila siku kwenda akila kuyu,
that monkey each day went eating figs
’Every day that monkey went eating figs,"

na yule rafiki yake papa huwapo chini ya mti.
and that friend his shark be-there beneath of tree
'hnd his friend the shark would be beneath the tree.”

Humwambia,
he-say-to-him
"He would say to him,”

Utupie nami rafiki yangu vyakula;
throw—to me friend my food
"Throw me some fbooL my friendk”

humtupia siku nyingi na miezi mingi.
throw—to-him days many and months many
”he was throwing to him for many days and many months."

FIGURE 1: Paragraph 1 of PWD Divided into Clauses and Translated

29
one sentence-like or clause-like unit. Because of this, each orthogra—
phic sentence here can be called a sentence cluster. The characteristic
that sets each of these clusters apart is their function in the para-
graph. The first sentence cluster describes participant relations; the
second, the initial location of the story; the third, habitual actions
of the participants; and the fourth, habitual subsequent action of the
participants. Each sentence cluster in this paragraph has the role men-
tioned above: participant relation, location, action, subsequent action.
The setting paragraph isolated earlier is now seen to have four major
constituents in the form of sentence clusters, each of which has its own
function.

The sentence clusters may now be examined. As seen in Figure 1
(page 28), the first sentence cluster, which functions to express par—
ticipant relations, is composed of two clauses. The first clause (1.1a.)
is a formulaic story opener (cf. Johnson, 1939), so it may be said to
have the function aperture (Longacre, 1976, p. 213). The second
structure (1.1b.) is an expanded clause which will be termed a sentence
stem. This sentence stem functions to state the relationships between
the primary participants of the story, so the function label given to
this structure will be situation (see Figure 2, page 30).

Skipping ahead to the third sentence cluster, we find that a more
complex analysis is required here. Structure 1.3a. will be called a
sentence stem because it has both a main and subordinate clause. Nor-
mally each clause is considered to have its own discourse function, so
the participial adverb akila (eating), could be taken as the purpose of
kwenda (go). The structure of this sentence is diagrammed in Figure 3

(page 31). It is not clear, however, that akila functions to express

30

wetness .eea

 

mrmpxeoa egg to H samcmmama Lot

 

 

 

 

 

 

.p_ms .pm.m.u
inllalnillu .mepo.m
pom mam .pc.m.n
.m.c.:
. m L
p .pc._o.n
.eom.ecu .p;.m.e
.mepu.m
.pom .nao.h:o
.pom .pm.m.u
.Lo .pc.m.e
.meFo.m
pave vac nco .ao .p;.m.u
.ooF .meFo.m .oo_ .p;.m.e
Ewp_ .me_o.m mam: .pc.m.e
sop, .pa.m.u
. .pc.m.u
.Fmg.pga .me_c.m ecoo
.pL.Fo.v

.L@ a

mngoszm OMLJOUmwQ "N mmDGHm

.mspxos mics wed ammo
miss 99% E®& on Smssn mm
.TmcwE vawe
as emcvxc :xwm wwasz:I.o
«weoexk me “cook ms sexes
McFaxmx>
:mcmz wxwwmc ram: mwazps.n
«mum wmxos mm
.wwnEm3::.m¢.H
.omsp exp
epeosma semen we @Nxos
cases was weaves was was
.wpe ma rewgu omega:
mama mxmz wxwmmg w_3> mc.p
«mmwk mxwpem
9:83 moxtos pomp mow madam
.zzsx vaxm
mvcwzx :xwm mFTx mex wF:>.mm.H
.oom esp ammo mama %Ns£ wee
.wcwcmcmn oxwx 3mg: m:.m
seen as“
some mama mo£p§8§® new %Nd&
wcwws oxmz :mzc mxmx asepwe.u
«genes goat and: mewaosm
.wu:w_wx mx_pwx muowE:.o
«mmspimw% 8 common
.zxzxs oxm_ newn.n
«moan mes e was msox&
.mznsxs wee m:ma.mN.H
.«&d£m e xpwe afieowsk owes wee
.mamg m: wxwwmcz mX:mmmxm.n
kKo 9:83 moxios w
mex mxouc0wF<.mH.H

31

 

 

1.3a.Yule kima n,p, _____._..____j!fl§fl;_
That monkey .
kila siku adv. t‘me d.s st. act.
every day
kwenda v. act. v.p. __E££;_

go
akila kuyu cl.
eating figs

FIGURE 3: Alternative Analysis of Sentence 1.3a.

the purpose of going. Kwenda akila could be an idiom or a compound
verb, and in that case the structure would be much simpler, the sentence
stem having the function action. For the sake of simplicity in presen-
tation, compound verb will be assumed here. At this point, some in-
determinancy in the analysis results in an arbitrary choice being made.
However, this is not the fault of the theory or method; simply more in-
formation about compound verbs in Swahili is needed to make the choice.
Sentence stem 1.3b. also expresses an action, but it is coordinated
with the action of 1.3a. The function of 1.3b. will therefore be called
coordinate action, and, since 1.3a. and 1.3b. are conjoined, the sen-
tence cluster they comprise will be called a conjoined coordinate
sentence cluster. The analysis is presented in Figure 2 (page 30).
Sentence 1.4. includes a quotation and a statement of the speech
act's result. The quotation has two constituents, the first a clause
root humwambia (he usually tells him), which functions to introduce the
quote. This is given the functional label quote frame. The quote it-
self proposes action by the monkey, and is expressed by a sentence
having a verb in the subjunctive form. This structure will be called a
subjunctive sentence stem. The quote frame and quoted sentence form a
larger sentential form which will be called a narrative quote sentence.

In sentence cluster 1.4., the narrative quote sentence functions to

32
convey a request. 1.4b., a declarative sentence stem, functions as the
proposition which realizes that request. See Figure 2 (page 30) for a
diagram of this analysis.

Sentence cluster 1.2. presents more complexities, because it is
tempting to interpret clauses 1.2b. and 1.2c. as non—restrictive and
restrictive relative clauses, respectively. Steere's translation reads,
"There was a great tree, of the sort called mkuyu, which grew near the
deep water..." Swahili marks relative clauses in either of two ways,
marking the verb with an affix, or using the relative pronoun amba-.

In clauses 1.2b.-c., verbs carry no relative affixes, and no amba- forms
are present. The only possibility is that the amba- forms have been de-
leted, but it is unknown here if that can be done. Given this uncertain—
ty, a simpler course will be followed. The clauses 1.2a.-c. will be
considered of equal status, the first functioning to state the existence
of some item, the second naming that item, and the third specifying its
location. Clauses 1.2d.-e. together describe the spatial distribution
of the tree's branches, and each of these clauses states a portion of
that distribution. Sentences 1.2d.-e. therefore form another conjoined
coordinate sentence cluster. A final problem occurs when we try to de-
cide the structural status of these four constructions within the in-
cluding cluster. It is possible to consider them equals, resulting in
the structure diagrammed in Figure 4 (page 33). However, the semi-colon
following 1.2c. may suggest an intonational clue that 1.2a.-c. is a
cluster in its own right. Functionally, this cluster would specify the
item as opposed to the coordinate cluster which specifies the distri—

bution of the item. This analysis is diagrammed in Figure 2 (page 30).

 

33

item

 

1.2a.Pana mti mkubwa, d.s.rt.
There was a big tree,

b.jina lake mkuyu, d.s.rt.
called a fig—tree,
.umeota katika kilindi; d.s.rt.
it had grown near
dee water;
d.mat§wi yake nusu yako d.s.rt. —EQ£L—
mjini cnj.crd. dist.
its branches were s.clstr.
half over the town
.na nusu yako baharini. d.s.rt. —9934—
and half over the sea.

name

 

place

 

('1

. loc.
clstr.

 

 

I'D

FIGURE 4: Alternative Analysis of Sentence Cluster 1.2.

2.5.2. Discussion of the Analysis

The analysis presented above is partial and incomplete. A compre-
hensive analysis would have determined emic (contrastive) construction
types and accounted for how each construction found in the text was re-
presentative of some emic construction type. The analysis presented
here is a necessary first step in that direction, and it does provide us
with a fairly reliable skeleton of the grammatical structures found in
that particular paragraph.

A major implication of this type of analysis is that each identi—
fied construction is the realization of some tagmeme. Thus the identi-
fication of a sentence cluster functioning to express location (1.2. on
Figure 2, page 30) implies an emic paragraph type, one of whose tagmemic
constituents allows location as a role, and is fillable with a sentence
cluster. Each branch of the diagram implies a tagmeme which is capable
of being realized by the construction type and role for which the branch
is labeled.

A number of indeterminancies were noticed with respect to function

labels, structural configurations, and construction labels. In the

 

34

cases of structural indeterminancy more data was needed. For example,
the problem of possible deletion of relative pronouns in clauses 1.2b-c.
simply needs more data for resolution. In cases where the construction
type is unknown or the name of a function is not at hand, it should be
remembered that the labels themselves are not important; what is impor-
tant is that a structure or a function has been isolated, and can be
judged the same as or different from other constructions or functions.
The final assignment of labels is left to the final stage of the con—

trastive analysis.

3.1.

3. ANALYTICAL PROBLEMS

Structural Labels

One of the problems encountered in the analysis of hierarchically

structured data is that the constructions found at the different levels

all require names. In this study the names assigned to sentential con-

structions between the clause and paragraph levels are not critical, yet

a consistent and descriptive classification of these structures is need-

ed. The following etic (descriptive) scheme will be applied to the

Swahili sentences of this folktale:

1.

The Swahili sentences will be classified as declarative,
interrogative, or subjunctive. Interrogative sentences are
marked by an interrogative suffix on the verb or a question
word within the sentence. The subjunctive is marked by the
suffix -e- on the verb. The subjunctive in Swahili is often
used as a polite imperative, and as an expression of intention.
Four levels of sentential complexity will be posited between
the clause and paragraph levels (cf. Pike and Pike, 1977).

The simplest structure will be the clause root (cl.rt.) which
is the verb plus various tense, agreement, and derivational

affixes. For example:

Example 3.1-1.

a. akamwambia and he told him (declarative clause root)

b. twende let's go (subjunctive clause root)

The next level of complexity is the sentence root (s.rt.) which is

a clause root plus subject, object, or modifying phrase. This is the

basic “unexpanded” sentence which shows no embedding. In Swahili the

35

36
sentence root and clause are at the same level; because this analysis
looks downward, structures at this level will be called sentence roots,
although they could just as well be called clauses. Examples of sen-
tence roots:

Example 3.1—2.

a. Ntakuleta nyama kesho. I will bring you meat tomorrow.
(declarative sentence root)

b. Sasa twende zetu. Let's get on our way now.
(subjunctive sentence root)

The next level above the sentence root is the sentence stem, in
which two or more sentential or clausal roots are linked with some sort
of grammatical device. This device may be conjunction or some depen-
dency relation.

Example 3 1-3.

a. Ntakuleta nyama kesho tuje kule. I will bring you meat
tomorrow so we may eat. (declarative sentence stem)

When the constituents of a sentence stem are linked by a conjunc—
tion the structure may be called a conjoined sentence stem. There is a
special linked sentential construction using the —ka— tense which
usually expresses sequential action. Constructions of this sort will be
called linked sentence stems. For example:
Example 3.1—4.
a. Sungura akaondoka, akaenda mwituni, akamwona punda. The
hare left, went into the fbrest, and saw the donkey.
(linked sentence stem)
Finally, a sentence cluster is any group of sentential constructions
which are not grammatically linked, but which have coordinated functions.

For example:

 

37

Example 3.1-5.

a. Fadhili zako nyingi, nataka kwenda kwetu nikakulipe
fadhili. Your kindness is great, I want you to come to
our place so that I may repay it. (sentence cluster)

Note that orthographic punctuation cues are unreliable for determining
sentence clusters. Often sentence clusters are included within ortho-
graphic sentences.

The terms dialog, conversation, exchange, and turn require brief
discussion. Dialog is used here as a general term for conversation
between two people. Conversation, however, is a technical term for the
structure formed by one or more linked exchanges. Exchanges are the
basic structural units of conversations, and consist of at least an
initiating turn and a resolving turn. Turns (occasionally referred to

here as utterances) are simply the speeches made by the participants.
3.2. Quoting Structures

In the original analysis of the folktale it became quite evident
that the use of narrative quote sentence (n.q.s.) as quoting structures
created a highly regular and predictable layer immediately below the
exchange level. Paragraph two of the folktale, diagrammed in Figure 5
(page 38), is a typical illustration of this situation. This diagram
shows that each turn of the exchange is filled by an n.q.s. consisting
of a quote frame (q.f.) which identifies a speaker and signals the
onset of a quote, and the quote (qt.) itself. 0f fifty—nine narrated
conversational turns in the folktale, only two did not exhibit the
q.f -qt. pattern. One turn, 9.4., reversed this sequence, while another
turn, 4.4., eliminated the quote frame (see Appendix 3, page 85).

Despite the regularity of this pattern, it poses two problems. The

Jam)»:

.mwma_m:m chwmwso .wrmpxrom use to N cgwsmmcma ”m mazmHa

 

.pa._o.nm
.m.e.=

 

.pL._u.U

 

. .uc.m.nm
.LHWFU Eoca

 

. . . . .pa._o.e
IuHMflmil m a : can

 

.BL._U.U

 

.um.m.u

.pL.Fu.w

Idmmmmll .m.c.:

 

.u;.m.v

. .pc.m.nm
.pm.m.u cm;
.pc.m.e

.oca

 

.sa.m.e
.p>=e

 

.pc.m.n
1v.a.>vm

 

 

 

.Li.xm

.om mmeu
.mu:w3p.n
revs wNou me new
.mvnEmzewx<.m¢.N
.xom
space wa3 genes k0 mean 8 no:
.mueazxwm__ whee m_ m:oe.o
.sos sees Naps N
.m:x:;uzxmpz.n
«Sex on ween me new
.mwnEmzemx<.mm.m
.mNGEwEB fixcm m3
«songs ox» ones om s.cow o3
.wccwn m3
use»: .wcwmme wwmcvzpmc wmm:.o
«om N NNws 38m
.ahaeeazxepz.g
anaemewre mmxtoz
.znwhmem mE_¥.mm.N
.mmospsex axom memos mos H on
.VFTmew marpzxmewc.w
ooeNm ass on 8508 on 30m «:83 H
:umZx mucmzx mxmpm:.u
apnoea we momefirwx $30»
.emcesc oxeN e_e£eea.c
ammxecz worms «somm
.mevx MN_—:3Em¥m cama.n
new o:o rm£B
oat; =¥_m epeI.eH.N

39
first is that not only sentences are quoted in this manner, but greater
and lesser structures as well. Note in Figure 5 (page 38) that the
quote in turn 2.3. is a sentence cluster, not a dependent or conjoined
sentence stem. Note also that the same quote frame, akamwambia (and he
said to him), is used to frame an adverb in sentence 8.5. and an entire
monolog in sentence 7.1. This evidence would require either the estab-
lishment of classes such as narrative quote sentence clusters and nar-
rative quote monologs, or the creation of a structure within narrated
exchanges which marks the quote and labels the speaker.

Consideration of the second problem will shed further light on this
choice. The second problem is that the n.q.s. analysis of Figure 5
(page 38) is totally predictable and hence inelegant. All turns in a
narrated exchange are filled by an n.q.s. This regularity suggests the
same alternatives as before; either the quote structures should be dif-
ferentiated so a variety of structures fill the turns, or a special
quoting structure used within the exchange should be created.

Because the creation of a variety of quoting structures would add
considerably to the number of contrastive classes, it is desirable to
opt for a quoting structure which is part of the exchange structure.
This alternative gains additional support from the existence of other
structures which must be included within the exchange, such as the
adverbial phrase 2.1a. in Figure 5 (page 38). This phrase acts as a
time frame for the entire exchange and is a 'sister' of the exchange's
constituent turns. It is proposed that quote frames be considered
immediate constituents of narrative exchanges, and therefore sister con-
stituents of the quoted structures directly under the exchange node.

This solution simplifies the hierarchical structure of exchanges,

 

4O
eliminates the need for many additional construction classes, and helps
to directly associate the function of a conversational turn with the
structure that fills it. A revised analysis of paragraph two is pre-

sented in Figure 6 (page 41).
3.3. Problems with Function Labels

Two important but related problems surface during the analysis of
function in discourse. One has to do with the choice of terms used for
labels (what should function X be called?), while the other concerns the
appropriateness of the label for the particular level of structure being
considered. Both problems are evident in Figure 5 (page 38). Notice
that the description of sentence cluster 2.1c.-e. requires labels of
function at three levels. During analysis this creates a problem of
choice because it is initially assumed that the function of each con-
struction at each of the levels is different. The analyst normally asks
”What does this construction do within the including construction?" but
the answer is sometimes misleading. In Figure 6 (page 41), for example,
it was decided that the function of sentence cluster 2.1c.-e. was
'invitation' (invt.), but careful examination will reveal that this
could serve to describe the compound sentence 2.1d.-e., or even the
sentence root 2.1d.

This vagueness in application of function labels relates to a pro-
blem of appropriateness of these terms to the level at which they apply.
Are 'invitation' (2.1c.—e.) and 'clarification' (2.1b.—c.) appropriate
functional labels for the exchange level structures? Also, can these
terms be seen as members of an integrated set of functional labels for

the exchange level? If such an integration is attempted these functions

41

.mgspozcum omcmsoxw cw m_.c.c mu=_ocw op ummw>mc .mFMHxFQm one to N zamcmmgma no mmeHa

 

 

 

 

 

 

 

 

 

 

 

 

.om macaw
14me .pL._o.am .mvcmzp.n
«Sec wmop me new
. .c .pg._o.u .mwnEmzsmx<.mq.m
i .xom
£osop NNvB emcee %o momfi e no:
.50; .uc.m.am .mpmazxmeF whee m— w:oe.o
lanai 529m— £3. mass 3.3 L.
.oga .pL.—u.u . .m:x:;u=xmpz.n
See on Beam me new
. .c .pL.Fo.u .mwnEmzecx<.wm.m
i .mmuswse weoN o3
«emcee men 092% om bison ma
.mgcmp m3
.xm.gc .cmg .pm.m.u mam»: .wcwhma wwm:w:pm: wmmc.u
.pm.m .om H Nmes 30m
.mwzm .unco .mmzc .ps.~u.w .mwmucmzxmgz.n
domsmSmio moxioE
..wm .uL.m.u .znwnsmxm me¥.mw.N
i .mmmrw:w« 830m memos mes H on
.zmg .wc.m.nm .TFT;UM$ maw_:xmuw:.w
. mco .pm.m.u moemm $30 on osoo on son “:83 H
.me—u.m m .0; .wc.m.v :pm3x mucmZx mxwpmc.u
.u>:w «nemgm we mmmefiewx Leo»
.Eoco .pe.m.e .emee»: oxeN __e;eea.u
«merges on ween «seem
.$.c .pc.m.u .mewx me_:zamxm mamq.a
.m.c .pm.m.u mow use sees
as: vase 9:; it. 332w

 

.Lt.xm

42
would seem rather inconsistent with the quote frame function discussed
earlier. This problem of consistency and appropriateness of function
labels at a given level can only be solved by detailed analysis of
function at specific levels of the grammatical hierarchy. Such work has
been done by Hale (1974) at the clause level; Sterner, Suharno, and Pike
(1976) for sentence clusters; and by Longacre (1976) for conversational
exchanges. Longacre's treatment of exchange level structure and function
is based on a large amount of work done on Philippine and Mesoamerican
languages, and has general applicability to other languages. A modified
set of his exchange level function labels will be used in this analysis.
The following sections will describe and critique Longacre's treatment of

exchanges and explain why certain modifications were made to the scheme.

3.4. Longacre's Analysis 9: Exchanges

 

3.4.1. Longacre's Deep and Surface Structures

Longacre (1968, 1976) made a significant departure from Pikean
tagmemics when he adapted the terms deep and surface structure to de-
scribe the relations of slots to each other and functions to each other
within a hierarchy. This proposal has caused considerable discussion
among tagmemicists, but it is fair to state that Longacre's surface struc-
tures correspond to Pike's slots, while his deep structures correspond
at least in part to Pike's roles (functions)(Brend, personal communica-
tion, 1979). The significant difference between these approaches is that
Longacre allows the separate analysis of functional relations and struc-
tural relations, whereas the latest version of Pikean tagmemics has
structure and function inextricably bound together. For Longacre,

structural and functional relationships are generally congruent, but may

43
be 'out-of'phase' at certain points.

This paper attempts to justify the Pikean model, but Longacre's
well developed set of function labels for exchanges will be borrowed
with certain modifications. Justification for these changes will be
made in the critique of Longacre's analysis of exchanges given in

section 3.5.
3.4.2. Terminology for Exchange Structure and Function

Longacre (1976) provides a set of labels for both structure and
function in conversational exchanges. His structural labels include
initiating utterance (IU), resolving utterance (RU), continuing utter-
ance (CU), and terminating utterance (TU). An initiating utterance
opens an exchange, while a resolving utterance closes or at least re-
solves the issue brought up within the exchange. A continuing utter-
ance is a structure which extends an exchange by failing to resolve the
previous utterance and forcing the other participant to make a resolving
utterance. A terminating utterance closes an exchange after it has been
resolved. TU's are often optional. The exchange analyzed in Figure 6
(page 41) illustrates these four slots. The first turn (2.1c.—e.) is an
initiating utterance in which an invitation is presented. The second
turn is a continuing utterance because it fails to resolve the IU and
causes the other participant, in this case the initiator of the exchange,
to resolve a new issue. The third turn in this exchange is the resolv-
ing utterance since the issue raised in the previous turn is settled
here. The final turn is the terminating utterance since it closes the
exchange by signaling agreement on the resolved issue.

Longacre's terms for function (deep structure) form a small but

44
very useful set which is organized according to which turn (IU, CU, or RU)
the function applies. Initiating utterances may take three basic func—
tions: proposal, question, and remark. Proposals (pro.) are utterances
which ask or suggest some action on the part of one of the exchange's
participants. Under this term are included utterances which act to ad-
vise, suggest, invite, plan, threaten, request, or command. Questions
(ques.) ask for information, but are not considered to be proposals.
Utterances which have an interrogative form, but which function to re-
quest some action function as proposals, not questions. Remarks (rem.)
function as statements or declarations which the hearer may comment upon.
Remarks do not request action or information.

Resolving utterances have their own set of functions each of which
corresponds to a function listed for initiating utterances. Resolving
functions include answer, response, and evaluation. Answers resolve
questions, responses resolve proposals, and evaluations resolve remarks.
These functions may be used in terminating utterances as well, although
TU's may also express acquiescence (acq.) or rejection (rej.). In
Figure 6 (page 41) the IU (2.1.) functions as a proposal, the RU (2.3.)
functions as an answer to a question put forth in the previous turn, and
the TU functions as acquiescence.

Continuing responses act to let a participant in a position to re-
solve an exchange prolong it by countering the preceding utterance in
some manner. The three functional labels for utterances which do this
are the counter-question (c.ques.), counter-proposal (c.pro.), and
counter—remark (c.rem.). The definitions of these functions are the
same as they were for the IU's except that their use is directed towards

the purpose of 'countering.‘ The function of a CU need not counter the

45
same type of function in the preceding utterance. In Figure 6 (page 41),
for example, the second turn (2.2.) is a CU functioning as a counter-
question, although the preceding turn is an IU functioning as a proposal.
Figure 7 (page 46) shows the structure of paragraph two relabeled with
Longacre's terms for exchange slot and function. Slot names, where used,

are indicated in parentheses below the branch line.
3.4.3. Exchange Types

Longacre's labels for the different turns in an exchange and their
possible functions enable him to present an insightful discussion of
exchange types. What he calls a simple exchange is the most basic type
of dialog, and IU—RU turn sequence. Sentences 8.4. and 8.5. of the folk-
tale provide an example of a simple exchange in which the rabbit schemes
with the lion to get some meat:

Example 3.4.3—1.

8.4. rabbit: I'll bring you an animal tomorrow so that we
may both eat. (IU, pro.)

8.5. lion: Good. (RU, resp.)

In some cases a simple exchange may allow an optional terminating
utterance. Consider this translation of an exchange between the rabbit
and the lion:

Example 3.4.3—2.

12.1. rabbit: Well lion, did you get her? (IU, ques.)

12.2. lion: No I didn't. She hit me with her hoofs and ran
away, and now I have many bruises because of my
lack of strength. (RU, ans.)

12.3. rabbit: Sorry to hear that. (TU, eval.)

The TU is optional because the exchange seems resolved with or with—

out it. The general pattern of simple exchanges may be formulated as:

46

 

.mcoweoczw use wee—m m_mcommc04 com vamaeFmL N :amcmmcma um mmsza

.om m.pmq
mwww .pc.~o.nm .wncmzh.n
.Ee: :Nop o: w:w
.pL.Fu.e .mwnEmBme<.m¢.N
.som
:oxop NNQB seeds k0 Qo:fi 8 no:
. .pc._o.nm .wumasxwmwF whee m_ wcop.u
mmmw .me_u.m Eosa qsom oxen NNea H
.pc._u.u .m:x:;ozxmuz.a
.51: B 38 a: as.
.pg._u.u .mwnEmZme<.wm.N
.mNBEw:d w:dN m3
«amass m:p op:e om p.:ofi m3
.eccmn m3
.pm.m.u mme: .wahme wwmcwsee: wmmc.o
.8 A SS 8m
.pL._o.w .mhmecmzxmpz.n
JemmmSm:e mox:g§
.pc.m.u .znemmem mew¥.mm.m
.mmm:fi:ex 830m :Rxpm: has H om

. .us.m.nm .Wchvmw waw_:xmpwc.m
. u .pm.m.v_ cm; oodmm :30 on uses on 30m «:83 N

ADHV .L . m: . a .pg.m.e :prx mvcmzx exmpm:.u
.oga pm_o m o; seesaw we mmo:w:e« :30»

.pc.m.u .wmch: oxmw Wszuea.o
«mmx:oé on meow «Le:m

. . .pL.m.v .mex mwwpzzemxm mama.n
. .G .pm.m.u m new m:o :m:&
w we?» .a.>um oxen :xwm mme.wH.N

.Lm.xm

 

 

.L..U

 

 

 

.ocq

 

.xo.Lc

 

_ .cmg

 

m20~ .pm.m
.mo: .0 .umco

.ch

.mmzc

 

 

 

.EULU

 

 

47
simple exchange = +IU +RU iTU
An exchange that contains one or more continuing utterances is a
complex exchange. The rabbit's proposal to the donkey can serve as an
example of this type of exchange:

Example 3.4.3—3

9.2. rabbit: I have been asked to come here to propose to
you. (IU, pro.

9.3. donkey: By whom? (CU, c.ques.)
9.4. rabbit: By the lion. (RU, ans.)
9.5. donkey: Well, let's go. (TU, acq.)
As in the case of simple exchanges, the terminal utterance may or
may not be used. The general formula for a complex exchange will be:
complex exchange = +IU +cun +RU iTU
Longacre also notes that exchanges may be linked together to form
compound dialogs. In this paper, such compound dialogs are considered
conversations. In compound dialog exchanges are linked together when
one exchange is resolved and a new one about the same or related topic
is initiated. In its simplest form a compound dialog is a series of
exchanges concerning the same topic. Here is a hypothetical example
from English:

Example 3.4.3-4.

EX1 A: Where's my book? (IU, ques.)
B: You left it on the table. (RU, ans.)

EX2 A: If I had cleaned up last night I would have seen it.
(IU, rem.)

B: Well, it was Jane's turn to clean. (RU, eval.)
There are more complex ways in which exchanges may be linked, but

Longacre does not mention them. These complex linkages involve the

48
sharing of constituents between exchanges. Linkage accomplished by
sharing constituents is very common in the folktale, and will be dis-

cussed at length in sections 4.3. and 5.4.

3.4.4. Out-of—Phaseness

Longacre coined the term "out-of—phase" to describe situations in
which the relations between functions in a dialog do not coicide with
the relations between grammatical structures. This situation arises
when a grammatical structure has more than one function relative to its
including constructions. The following English example, patterned after
one presented by Longacre, demonstrates this:

Example 3.4.4-1.

1. A: Would you please move that box over there? (IU, pro.)
2. B: Sure boss, could Fred help me? (CU, res./pro.)

3. A: Don't strain yourselves; use the forklift. (CU, pro./
pro.)

4. B: Thanks. (RU, resp.)

The second and third turns in this exchange have dual functions. In
the second turn, the speaker both responds to the original proposal "sure
boss...“ and makes a counter—proposal ”...could Fred help me?“ In the
next turn Speaker A counters this by saying, "Don't strain yourselves...”
and then makes a new proposal ”...use the foklift."

The following chart adapted from Longacre (1976) illustrates why

he describes such structures as being out of phase:

49

Example 3.4.4-2.
Surface Deep

1. A: IU - pro. ]
, _ resp.
2. B. cu term]
, _ c.pro.
3 A. cu {pm
4. B: RU - resp.]

These three naturally formed pairs of deep structure relations are indi-
cated by the square brackets, and they are incongruent with the four
surface structures. This is one of the reasons why Longacre separates
deep structure (relations between functions) and surface structure (re-
lations between grammatical units).

Example 3.4.3-2. may be reanalyzed to show an out-of—phaseness
similar to that of example 3.4.4-1. In this case, the RU (12.2) is
analyzed as carrying two functions, answer and remark:

Example 3.4.4-3.

12.1. rabbit: Well lion, did you get her? (IU, ques.)

12.2. lion: No I didn't; she hit me with her hoofs and ran
away, and now I have many bruises because of
my lack of strength. (RU, ans./rem.)

12.3. rabbit: Sorry to hear that. (TU, eval.)

Multiple function such as this is common in the folktale. Out-of-
phaseness, together with the linkage shown in the previous section con—
tributes to the complexity of exchange structure in the folktale.
Longacre points out that out-of—phaseness makes dialog seem ”effective"
and suggests that discourse of literary quality may demand such struc-
ture rather than tolerate it. Perhaps this greater effectiveness is due
to a greater load of functional information carried on each structural

unit.

 

50
3.5. Critigue ef_Longacre's Treatment pf_Exchange Structure

 

Longacre's treatment of dialog structures and functions is both
simple and elegant. His function labels are few in number, yet appro-
priate and comprehensive for the exchange level. His analysis of dialog
is straightforward and gives us insight into dialog structure and func-
tion. Despite the importance of his contribution, Longacre's treatment
of dialog is subject to at least two criticisms from a Pikean viewpoint.
First is that his analysis fails to accurately associate the function of
a particular utterance in a dialog with the structure within that utter-
ance which serves that function. Second, the purported existence of
out-of—phase (incongruent) relations between deep and surface structures
seems to represent an anomaly, since such structures appear to be "in-

phase" at all the lower levels.

3.5.1. Loss of Detail

The first problem of Longacre's analysis of exchanges is that func-
tions are not clearly associated with particular structures in cases
where a complex filler of a turn is analyzed as having more than one
function. This situation can be called loss of detail, and it is clear-
ly illustrated by example 3.4.4-1. Turns two and three of this example
each have two functions. This suggests that the fillers of these turns
each have two functions. Taken as whole entities, these two structures
do have two functions apiece. However, we notice that each filler is a
complex unit. Turn two is a sentence consisting of a phrase of_agree-
ment followed by an interrogative clause, while turn three is a sen-
tence cluster consisting of two imperative clauses. It is clear that

each of these constituents is associated with one of the exchange

 

51
functions. "Sure boss...” functions as a response to the proposal in
turn one, while "...could Fred help me?" is a counter-proposal. “Don't
strain yourselves..." is the counter-proposal in turn three, while
”...use the forklift," is the new proposal.

When complex fillers of conversational turns are examined, we find
that many of the sub-constituents can be labeled for a function differ-
ent from that of the entire utterance. Often, an answering utterance may
be prefaced by a rhetorical question, or a responding utterance may in-
clude a remark. Noting the functions of the constituents of a complex
filler of an exchange may yield important information about the construc-
tion's internal structure and linkage to other exchanges. Longacre's
analysis, however, does not permit a sufficient amount of detail to make
this evident. Given the potential usefulness of this functional detail
for the study of such things as linkage between exchanges, it is surpris-
ing that Longacre allows the loss of so much of it. This loss of detail
may account for why he mentions only the simplest type of linkage be-

tween exchanges.

3.5.2. A Problem with Out-of—Phaseness

A second criticism of Longacre's treatment of exchanges concerns
his demonstration that deep and surface structures may be "out-of—phase“
or incongruent with each other. Out-of—phaseness is interesting in its
own right, but it is also important for Longacre's theory insofar as it
helps demonstrate the necessity for the separate treatments of deep and
surface structures. It was an apparent "out-of—phaseness” between
phonological and grammatical structures that stimulated Pike to explore

the notion of having separate hierarchies in his theoretical treatment

52
of language.

Out-of—phaseness is therefore known and accepted within tagmemics.
However, Longacre's out-of—phaseness appears only at the higher levels
of discourse, the exchange level and above. The question is why doesn't
it appear at lower levels also? It seems inconsistent to find this
phenomena occurring only above a certain level. The view taken in this
study is that a more detailed analysis of the associations between
structures and functions will eliminate the need for out—of—phaseness.
The following section proposes a minor revision of Longacre's scheme

which allows for a more detailed analysis.
3.5.3. Modifications to Longacre's Scheme

Two problems with Longacre's analysis of exchanges have been raised.
The first is that details of function within the exchange may be lost,
while the second is that out-of—phaseness occurs only at the highest
levels. A minor modification of Longacre's scheme will now be proposed
to help alleviate these problems for the initial etic description of
exchanges in this folktale. This proposal is to apply Longacre's func-
tional labels to any relevant structure or sub—structure within an ex-
change. This will allow a complex filler and its constituents to be
labeled for exchange function. Figure 8 (page 53) presents paragraph
two of the folktale reanalyzed in this manner. It should be stressed
that this modification is made to improve the etic description of the
exchanges; the emic analysis will reduce the detail to those elements
responsible for contrast. The enhanced etic analysis will, however, per-
mit an emic analysis which preserves any necessary detail and captures

functional relationships without divorcing them from structure.

 

53

.wsmgom m_weccm:04 op mcowpmuwewuos mcwm: N camemmgma to mwmapece crew ”w mmawHa

.om m.pmm
.pL._u.nm .mucwzh.n
“Se: ENQp o: w::
.pc._u.u .ewnsmzsmx<.e¢.N
.30m
:oxop wa3 mouse %o mo:fl e no:
.u:.m.nm .mpwazxemWF whee «F wcop.o
«30m oxep mmee H
.u;.—o.u .m:x:;u:xepz.n
«Se: on ween o: w::
.pL._o.v .mvnsczewx<.mm.N
.mmose:8 @:8N as
«seeds m:p oe:e om p_:ow o3
.occma m3
fidawe .pm.m.u mam»: .wahmE wwmceapm; wmmc.o

 

.amme

 

.nvoU

 

.mcm

1ll|I||||l .mero.m

.mcm

 

.oem

 

.Wud

.Xw.L:

 

«om H NNeB 30m
.pL._u.m .mhmvcwzxmpz.a
«wommsm:o mmx:o§
.pL.m.u .znwnwam me¥.wN.N
.mmm:c:w« exom :prm: nee H on
.Ew; .pc.m.nm .wpwguem waszxmpwc.w
.pm.m.u moeNm :xo on 8506 0% 30m p:83 H
.pL.m.u :pwzx mestx mxmpmc.v
«modem we mmo:fl:w« exom
.pe.w.e ._mee»e oxaN e_eeeea.a
«mox:o§ on meow «:8:m
. . .pe.m.v .mewx er_:zsexe mamg.n
.pm.w.u w New m:o :m:&
wEFp .a.>um oar; :xwm mpm:.mfi.N

lulllgwnlllln .pm.m
mm: o .ewco

 

.mmnc.o

 

.n—..U

.OL LHme m 0L

 

.EwL

 

.L..U

 

.Le.xm

 

54

A second minor modification will be to eliminate the use of slot
names (IU, RU, CU, and TU) in this analysis. The reason for this is
that identification of slot names has no effect on the emic or etic
analysis. Slots are easily identified by their position and nuclearity
feature. On a chart they can be easily numbered. Slot names often have
functional overtones, and confusion over slot and functional labels may
result. In Figure 8 (page 53) slots are identified only by their spa-

tial distribution in the graphic display; slot labels are not used.

 

4. ETIC ANALYSIS OF EXCHANGES IN KISA CHA PUNDA WA DOBI

4.1. Etic Diagrams fpp_$t0ry epg_Exchanges

An etic description of the folktale's upper—level discourse struc—
ture is presented in Chart 1 of Appendix III. This chart diagrams the
folktale down to the level of conversation and paragraph, and clearly
shows the embedding of one story within the other. Names for the con-
structions and many of the functions have been adapted from Pike and
Pike (1977), while the remainder of the functions have been borrowed
from Longacre's discussion of plot structure (1976). The numbers at the
terminal nodes correspond to numbered paragraphs of the story's text.

Examination of the etic charts of the folktale's exchanges prompts
several general remarks. First, two-thirds of the exchanges in this
story are linked to other exchanges. Two exchanges are said to be link-
ed when they relate to similar topics and are juxtaposed. Linked ex-
changes from what Longacre calls a compound dialog, and a compound dia-
log forms at least a part of a conversation. Unlinked exchanges are,
for the most part, embedded within paragraphs. Node 3 of the story
(Appendix III) illustrates a typical linked exchange cluster, while
Node 2 typifies the unlinked exchange.

Another observation is that fifteen of the twenty-two exchanges in
the story make use of continuing utterances, and may therefore be con—
sidered complex under Longacre's scheme. Node 2 is a typical complex
exchange. Complex exchanges are often linked together to form compound
dialogs. Node 6 provides an example of this.

Within exchanges are structures similar to quote frames which de—
scribe time, place, or action at some point in the exchange, usually the
onset. When such structures do occur at the beginning of the exchange

55

 

56
the action expressed functions to set up or frame the exchange; hence
such structures are called exchange frames. Although these structures
function directly in the narrated exchange, they are often grammatically
linked or subordinated to other structures in the exchange. In this
case the filler of the exchange frame will have dual functions, one
function as the exchange frame, and the other specific to the structure
in which the frame is included. Node 2 illustrates an exchange frame
with dual functions. Other examples of exchange frames are found in

Nodes 3, 6, 12, 14, 15, and 17.

4.2. Exchange Linkage

Longacre's discussion of dialog types mentions compound dialogs
which are formed by a series of linked exchanges (1976). The nature of
linkage is not discussed by Longacre, but his examples show us exchanges
which are placed side by side with no device other than position acting
to link them. Linkage of this sort, which I shall call conjunctive link—
age because the linkage is realized by placement, is seen in Node 4 of
the folktale.

The structural and functional detail made possible by the analytic
changes proposed in the previous chapter allow us to observe other types
of linkage between exchanges. These additional means of linkage occur
when a quoted utterance (or part of one) functions in two exchanges
rather than only one. For example, it is possible for an utterance to
simultaneously resolve one exchange and initiate another, as the
fragment of Node 6 diagrammed in Figure 9 (page 57) shows. Notice that
sentence 6.11. resolves the previous exchange by answering the shark's

question (6.10.), but also stimulates the shark to pursue the topic by

 

57

 

 

 

 

 

 

6.9.Monkey: Do you take me i.s.rt. ques.
for a washerman's donkey?
6.10.Shark: What's this about i.s.rt. c.ques. nr.ex. ~—
a washerman's donkey? ans.
6.11.Monkey: It's the one that d.s.st.< rem.
has neither a heart nor
ears.
6.12.Shark: What's the story i.s.rt. ues. s. c. ues.
about a washerman's donkey? clstr. nr.ex. .—
Tell me so I may know. sb.s.st. ro.
7.1.-17.9.(Monkey tells story.) mono. resp.

FIGURE 9: Shared Linkage

asking another question (6.12.). Sentence 6.11. therefore has a dual
function in the discourse, and links two exchanges together. A situa—
tion like this, in which the construction is shared in its entirety be-
tween two exchanges, will be called linkage via a shared constituent,
or shared linkage.

Another type of linkage found in this etic analysis is similar but
not identical to shared linkage. An example found in Node 4 is present—
ed in translation Figure 10A (page 58). In this case only part of a
construction (sentence cluster 4.2.) is shared between exchanges, so
this type of linkage will be termed linkage via a partially shared con—
stituent, or partially shared linkage. There are three instances of
partially shared linkage in the folktale, and the constituent which is
partially shared is always a sentence cluster (sentence clusters 4.2.,
4.7., and 17.4.). Sentence clusters and other sentential types may
participate in shared linkage also. The three types of linkage are
presented schematically in Figure 11 (page 59).

The effect of linkage is to make the linked exchanges appear to
overlap each other. Overlap of this sort is tolerated and expected by

tagmemicists, since it does not effect the emic analysis. This overlap

 

58

4.1. Shark: You have become d.cl.rt. rem. ues.
Si1ent' DO you speak? 'l.Cl.l"t. gues- I s.c15tr. :|
ans. nr.ex.-—
ns

4.2a.Monkey: I have nothing d.s.rt.

to say;

.Monkey: If you had told cmplx. “

me back there, I would lk.s.st.

have brought my heart.

4.3. Shark: Do you have your i.s.st.
heart here?

4.4. Monkey: Don't you know i.s.st. ues.
about us? When we roam cmplx. ans s.clstr. —Efl§4—
about we leave our hearts lk. '

in the trees, and go s.st.
about with only our
bodies.

s.clstr. a

     

U'

rem.
c.ques.

 

nr.ex. —

 

FIGURE 10A: Partially Shared Linkage

4.1 IU
4.2 CU
complex exchange
4.3 CU
4.4. R”
FIGURE 108: Exchange 4.1.—4. Surface Structure (Longacre's Model)
4.1 c. ues.

simple repartee

compound dialog

complex repartee

 

FIGURE 10C: Exchange 4.1.—4. Deep Structure (Longacre's Analysis)

is accounted for as a manifestation of the wave—like properties of lang-
uage (Pike, 1967, 1977).
Much of what Longacre would call out-of—phaseness can be accounted

for here by overlap due to the types of linkage mentioned. For

59

 

turn ————7 turn turn

turn ———-a exch. 1 turn exch. 1 turn exch. 1
cnstr.

turn —————4 turn turn
cnstr.

turn —~——— turn exch. 2 turn exch. 2

turn —————-exch. 2 turn turn

turn ———-

CONJOINED LINKAGE SHARED LINKAGE PARTIALLY SHARED LINKAGE

FIGURE 11: Types of Linkage

example, exchange 4.1.-4. (Figure 10A, page 58) would have to exhibit
out-of—phaseness in Longacre's model. This out-of—phaseness can be seen
in Figures 108 and 10C (page 58), which present the surface and deep
structures of this exchange as Longacre would analyze them. The deep
and surface structures are not identical in this example, and are there—
fore out-of—phase. (Note also that the deep structure analysis of
Figure 10C, page 58, fails to treat sentence 4.2. as a whole entity.)

In contrast to this analysis, the analysis which produced Figure
10A (page 58) used the revised scheme in which function may be associat-
ed with structure at any level. (Recall that Pike's revised tagmemics
requires structure and function to be bound together.) We see that
sentence 4.2. can stand alone as an answer to the initiating question,
but that part of it, complex linked sentence stem 4.2b., can be viewed
as acting to stimulate the next utterance, and hence functions as an in-
itiating remark for a new exchange. Thus, using the revised analytical
method we obtain two exchanges which are linked together by a shared
constituent, and which have their structures and functions totally in

phase. Increased analytic detail allows us to view the exchange linkage

 

6O

   
  
  

6.4.Shark: Let's go to my place. sb.cl.rt.
6.5.Monkey: Let's go where? i.cl.rt. nr.ex. ———
6.6.Shark: Let's go to my place. sb.s.st.
6.7.Monkey: You're crazy! d.s.rt. nr.ex. _T_
6.8.Shark: How's that? i.s.rt.

6.9.Monkey: You must take me d.s.st.
for a washerman's donkey.

nr.ex. —

FIGURE 12: Structural Out—of—Phaseness

clearly, without the need to call on out-of—phaseness.

4.3. Structural Out-of—Phaseness

 

The apparent structural overlap caused by shared linkage gives rise
to an interesting juxtaposition of hierarchical structures which can be
called structural out-of—phaseness. This type of out-of—phaseness is
quite different from Longacre's out-of—phase deep and surface structures
because it does not result from discrepancies between functions and
associated structures. Instead, structural out-of—phaseness results
when all the constituents of one exchange are constituents of other ex-
changes as well. This can happen when the constituents of a simple (two
turn) exchange are shared with the preceding and following exchanges, as
happens with exchange 6.6-7., shown in Figure 12 above. The sharing of
fillers between exchanges causes what can be described as out-of—phase—
ness between the exchanges and their fillers. This situation is illus—

trated schematically in the following example:

Example 4.3-1. turn
exchange

turn
exchange

turn
exchange

turn

5. AN EMIC ANALYSIS OF EXCHANGES IN THE FOLKTALE

5.1. Contrasting Exchange Types

An emic analysis is essentially a process in which constructions of
a given level are grouped together according to similarities in their
functional and structural configuration. Groups of constructions which
differ with respect to two or more structural or functional features are
said to contrast with each other, and every contrasting group is called
an emic class. Table 1 (page 63) presents the results of such an analy-
sis for the exchanges in the folktale. The groupings shown in Table 1
are based on the function of particular exchange slots. Exchanges with
the same number of turns and the same function for each turn were placed
in the same class.

This table shows that the twenty-one exchanges found in the folk-
tale may be grouped into eleven contrasting classes. 0f the eleven con-
trastive classes, six were represented by only one example. This is ex-
pected with a small corpus. The other classes had multiple examples,
the maximum number being five for the normal proposal exchange.

The functional relations between constituents of constructions with-
in a particular emic class are unique for that class (Pike and Pike,
1977). Examination of the classes proposed in Table 1 bears this out.
In the three simplest classes, the normal remark (1), normal proposal
(5), and normal question (11), there are three unique pairs of function,
remark—evaluation, proposal-response, and question-answer. In each case
the last utterance serves to resolve the first.

Classes seven and eight represent complex exchanges in which the
last—resolving-first functional pattern also holds. In these cases the
response to an initial proposal is given only after some clarifying

61

62
discussion. Example 5.1-1. illustrates this pattern clearly for Node 2
(paragraph 2). The monkey's final utterance, “Let's go,” functions as a
response to the shark's original proposal. The intermediary turns serve
to clarify the proposal.

Example 5.1-1.

2.1.Shark: Your kindness is great. I want you to go to my
place so that I may repay your kindness.

2.2.Monkey: How will I go? We animals of the land don't go
into the sea.

2.3.Shark: I will take you. Not a drop of water will touch
you.

2.4.Monkey: Let's go.

Exchanges in which the first utterance is resolved will be called
resolved exchanges. Actually, all exchanges (except class four which is
a special case) have resolving utterances of some sort, but it is not
always the initiating utterances which are resolved. Classes 1, 5, 7, 8,
and 11 contain resolved exchanges.

Exchange classes 2, 3, 6, 9, and 10 represent unresolved exchanges
because their initial utterances are not resolved. The resolving utter—
ances in these exchanges relate to the second or third turns. Example
5.1-2. illustrates this point for exchange 4.2.-4. The initiating re—
mark made by the monkey about bringing his heart is never evaluated by
the shark. This evaluation might have said something like, ”Yes indeed,
I'm sorry I forgot to tell you." Instead, the shark presents a counter-
question, ”Do you have your heart here?” which is resolved in the final

turn by the monkey's answer.

63

 

 

No. Description Configuration Examples
1 sim.rslvd.rem. rem. eval. 3.4.-5.
2 exp.unrslvd.rem. rem. ques. ans. (eval.) 4.2.-4., 6.7.-9.,
6.9.-11., 17.4.-9
3 exp.unrslvd.rem. rem. pro. resp. 6.11.-M.
4 special case rem. ques. eval. 3.5.—7.
5 sim.rslvd.pro. pro. resp. (eval.) 6.6.-7., 4.7.-9.,
13.2 —3., 15.3.-4.
6 exp.unrslvd.pro. pro. pro. resp. 3.2.-4., 4.5.-7.
7 exp.rslvd.pro. pro. ques. ans. resp. 2.1.-4., 9.3.—7.
8 exp.rslvd.pro. pro. rem. eval. resp. 14.3.-6.
9 exp.unrslvd.pro. pro. ques. ans. 6. .-6.
10 exp.unrslvd.pro. pro. ques. ques. ans. 17.1.-4.
11 sim.rslvd.ques. ques. ans. (eval.) 4.1.—2., 12.1.-3.

 

 

 

 

 

 

 

TABLE 1: Emic Exchange Classes

simple expanded

 

resolved 1, 5, 10 7, 8

 

unresolved 2, 3, 6,
9, 10

 

 

 

 

TABLE 2: Classification of Emic Exchange Classes

Example 5.1-2.

4.2.Monkey: Because you didn't tell me back there I didn't
get to take my heart.

4.3.Shark: Do you have your heart here?
4.4.Monkey: Don't you know about us? When we roam about we
leave our hearts in the trees, and roam with only
our bodies.
Even though exchange 4.2.-4. has been labeled unresolved, sentence
4.4. is an answer which resolves the question put forth in sentence 4.3.
Sentences 4.3.-4. appear to represent a resolved question exchange that
is embedded within exchange 4.2.-4. Example 5.1—1. displays a similar

situation because sentences 2.2.-3. are seen to form a resolved question

exchange inside of the resolved proposal. The data therefore indicate

 

 

64
that exchanges may be embedded within exchanges in this instance of
Swahili narrative. Embedding of exchanges within exchanges was briefly
discussed by Longacre in connection with clarification dialogs (1976).
Exchanges that exhibit embedding will be called expanded exchanges.
Those which are not expanded will be termed simple exchanges.

Table 2 (page 63) presents a matrix showing the assignment of these
newly coined exchange features to the emic classes derived from the data.
No exchange is assigned to the simple unresolved quadrant because such
an exchange would not be an exchange at all; it would be a single utter—
ance. Emic class four was not included in the matrix because its resolv-
ing turn (3.7.) is not spoken but thought by the monkey, and functions as
a link between two conversational sequences.

The prospect that exchanges in this folktale allow other exchanges
to be embedded in them considerably alters our view of functional rela-
tions within expanded exchanges. Counter-functions as proposed by
Longacre are no longer necessary in this analysis because the entire em-
bedded exchange functions to counter the initiating utterance. In re-
solved exchanges, the relation between the initiating utterance and its
resolving utterance will be much more explicit, since they will be clear-
ly paired at the same structural level in the diagram. The general pat-
tern of exchanges becomes initiation, optional expansion, and resolution,
in which the initiation and resolution have their normally paired func-
tions, and the embedding functions to counter the initiating utterance.

The diagram of Node 2 in Appendix 4 presents an emic reanalysis of
paragraph two as an example of this altered view of functional relation-
ships. Because sentences 2.2.—3. are interpreted as forming an embedded

exchange, the diagram appears more complex than its etic counterpart

 

65
(Node 2, Appendix 3). Node 3 is classified as an expanded resolved pro-
posal exchange (class seven) while its embedded exchange is classified
as a simple question exchange (class eleven). The function of the em-
bedded exchange is countering, but within this exchange the normal
question-answer function sequence is observed. The matrix proposal ex-

change maintains the normal proposal-response function sequence.

5.2. Emic Exchange Classes e§_Forming e_Hyperclass

 

 

The embedding of exchanges within other exchanges suggests that the
contrastive exchange classes may be interrelated in some systematic way.
It seems that any of the three basic exchange types (rem., pro., and
ques.) are expandable by exchanges of any type, and if an exchange is
expanded, it may be resolved or unresolved. The folktale's proposal ex-
changes illustrate this principle, although incompletely. We observe
that a proposal expanded by a question may be resolved or unresolved
(classes seven and nine respectively). If we posit that the second slot
in an exchange is fillable by an optional exchange, we can propose the

following tagmemic formula:

 

‘ etc. I etc.
= init. s.rt. + exp. exch. + reso. s.rt.
Exchange + rem. l cntr.l ' eval. >l.
pro. resp.
ques. ans.

 

This formula is generalized, because three are actually necessary,
one for each proposal, remark, and question exchange. A cohesion rule
is necessary to insure that a resolving utterance is present if the
expansion option is not taken.

Table 3 (page 66) lists the possible expansions of the three basic

exchange types down to one level of embedding. 0f twenty—one possible

66

 

 

 

 

sim.rslvd. exp.unrslvd. exp.rslvd.

rem1 evall(1) remZ-eval2 rem2—eval2—eval1
pro-resp(3) pro-resp—eval1
ques—ans(2) ques-ans-eval1

pro1 resp1(5) rem—eval rem—eval—resp1(8)
proZ—resp2(6) proZ—respZ—resp1
ques-ans(9) ques—ans—resp1(7)

ques1 ansl(11) rem—eval rem—eval-ans1
pro—resp pro-resp-ans1
quesZ-an52(10) quesZ—ansz-ans1

 

 

 

 

 

 

TABLE 3: Possible Expansion Sets of Exchange Hyperclasses
(parenthetical numbers indicate emic class in data)

expansions, ten are represented in the folktale. Absence of certain
expansions from the data may be due to the limited corpus or some re—
striction in the language, but the cause cannot be determined from the
present data. However, the data suggest the conclusion that the con-
trasting exchange classes represent members of the expansion set of
three basic exchange types. Such an expansion set would be similar to a
grammatical paradigm, and is called a hyperclass in tagmemics. The evi—
dence is strong enough to suggest the organization of exchanges into a
hyperclass, but not plentiful enough to set limits on expansion with
confidence.

For example, the hyperclass of possible proposal exchanges is re-
presented by the middle row of Table 3 above (to one level of embedding).
The simplest proposal exchange is prol-respl. A more structurally com-
plex group of proposal exchanges is the unresolved expanded group. Here

a proposal is followed by an embedded exchange which may be of any type,

 

67

remark, proposal, or question. The possible patterns are: prol-rem—
eval, prol-proz-respz, and prol-ques-ans. The latter two patterns are
found in the folktale, but the first is not. The most structurally com-
plex proposal exchanges are the resolved expanded type. These have a
proposal, embedded exchange, and a response. The possible expansions
are: prol-rem-eval-respl, prol-proz-respZ—respl, and prol-ques-ans-
respl. The first and last of these patterns are represented in the data.

As a final note, class ten, represented by sentences 17.1.-4. of
the folktale, show two levels of embedding. The initial proposal is
countered by a question, thereby introducing an embedded exchange. This
embedded exchange is countered by another question exchange, which re-
presents the second level of embedding. This lowest-level exchange is
resolved by an answer, leaving the upper two unresolved. (See Node 17,

Appendix IV.)

5.3. Emic Exchanges in the Structure of the Folktale

Appendix IV presents each exchange of the folktale diagrammed accord—
int to its emic structure. Because the emic analysis presented in the
previous two sections altered the view of functional relationships with-
in exchanges, the emic diagrams significantly differ in appearance from
the etic diagrams presented in Appendix III. The emic diagrams appear
more complex with more structural levels because of frequent embedding.

The analysis presented in Appendix IV permits the total of function
labels for exchange roles to be reduced in number from nine to seven,
because counter-remarks, etc., are no longer needed. This is not to
say, however, that counter-functions are of no use in etic analysis or

that they will never be found in an emic analysis of exchanges in other

68
texts or other languages. The analysis of this data simply allows these

terms to be set aside.

5.4. Exchange Structure epg_Linkage Type

Given that many exchanges in the folktale are unresolved, and that
many of the exchanges were linked together in a variety of ways to form
complex conversational structures, the question of what relation ex-
change structure bears to linkage type naturally arises. For example,
we might suspect that linkage is what causes exchanges to be unresolved,
and thereby hypothesize that unresolved exchanges will always be linked
to other exchanges.

Table 4 (page 69) throws some light on this question. The evidence
which it presents leads us to reject the hypothesis that unresolved ex—
changes are always linked, since there are two instances of independent
unresolved exchanges. Data in the table indicate that all structural
exchange types may occur independently (unlinked). However, it is clear
that not all exchange types are linked to other exchanges, for there are
no instances of resolved expanded exchanges being linked to other ex-
changes.

It is difficult to explain this observation, other than to suggest
that resolved complex exchanges are large paragraph-like structures that
are complete in themselves. As such, they are perfectly suited to stand
independently. However, there is no good reason why they could not be
linked. All of the construction classes which fill the resolving slots
of these exchanges are capable of being shared, and conjoined linkage
places no demands on internal structure.

An interesting pattern between type of linkage and type of structure

69

 

indp. cnjd.lk. p.s.lk. sh.lk.

 

rslvd.sim. 5 0 1 2
rslvd.exp. 3 O 0
unrslvd.exp. 2 1 2 4

 

 

 

 

 

 

TABLE 4: Exchange Structure vs. Linkage Type

 

s.rt. s.st. s.clstr.

 

shared 3 3 0
partially shared 0 O 3

 

 

 

 

 

 

TABLE 5: Type of Linkage vs. Structure Linked

shared is revealed in Table 5 above. It is observed that shared linkage
always involves sentence stems or sentence roots, but never clusters,
while partially shared linkage always involves sentence clusters, but
never stems or roots. The data is too limited to suggest anything but

a trend in this direction, and it is difficult to explain why this pat-
tern should occur. The pattern of partial sharing is perhaps the easier
to explain, because we may expect the more tightly bound constituents of
sentence stems and roots to be available for sharing on an all-or—nothing
basis. The constituents of a sentence cluster are likely to be able to
stand independently, so the cluster might be "broken“ more easily. It
is more difficult to explain why sentence clusters are not shared in
their entirety. Perhaps their large size has somethigg to do with it,
but it would not be surprising to find a sentence cluster being shared

in some other text.

 

70

5.5. Normal ye, Non-normal Class/Function Relationships

 

 

In language, we normally expect units which perform a given function
to take a particular form. In English, for example, sentences which
express questions are expected to contain interrogative clauses, while
commands take imperative clauses. When a function is performed by the
type of structure we expect to do that job, a normal class-function re—
lationship is said to exist. Occasionally, a function is performed by a
structure which does not normally do that job. For example, the inter-
rogative clause "Could you close the window?" performs the function of
the imperative "Close the window, please.” In such cases a non-normal
class-function relation exists.

The correlation between function and class is important to tagmemi-
cists, and Pike and others have given the topic a good deal of attention
(Pike and Pike, 1977). Very little work of this sort has been done at
the higher levels of the grammatical hierarchy, although some off-norm
variation between class and function is expected. Table 6 (page 71) was
constructed to provide a brief view of normal versus non-normal relations
at the exchange level of this folktale.

Table 6 (page 71) is a matrix which shows the frequency with which
each of the six exchange functions is associated with the three basic
sentence classes used in the etic analysis. Because these sentence
classes are not emically refined any inferences drawn from the table
must be viewed as tentative. Most assignments of normal class—function
relations were modeled after what might be expected for English. All of
the exchange functions had the declarative assigned as their normal

form except for question, which was assigned the interrogative. In

71

 

dec. sb. int.

 

 

eval. 5 — —

 

pro. 14 6 1

 

 

 

resp. 5 4 -
ques. — - 17
ans. 7 1 1

 

 

 

 

 

 

TABLE 6: Normal vs. Non-normal Class/Function Relations

addition, the subjunctive was allowed for proposal because the subjunc-
tive is often used for making requests, suggestions, and imperatives in
Swahili.

Table 6 above shows that the normal class-functions relations with—
in exchanges turn out to be largely as they were expected. Only four-
teen percent of the utterances were off-norm. The largest proportion of
off-norm utterances involved responses in the subjunctive. Examination
of response turns in the folktale reveals that many proposal exchanges
were resolved by a subjunctive clause root. Node 2 is typical of this
because the shark's invitation is finally resolved by a twende (let's
go), a clause root with a subjunctive suffix. Perhaps the subjunctive
should also be considered a normal form for response turns in Swahili

exchanges.

6. SUMMARY AND CONCLUSIONS

6.1. Summar

This thesis has performed two tasks, the first being to account for
the structure of conversational exchanges in a Swahili folktale using
Pike and Pike's revised tagmemic theory (1977), and the second to show
that complex exchange structure can be analyzed without recourse to the
notion of out-of—phaseness between structure and function.

The first two chapters provided background for this study: over-
views of discourse analysis, tagmemic theory, and Swahili literature and
culture. Chapter Three discussed some problems that occurred in the in-
itial analysis of the folktale. These problems involved the terminology
used to describe tagmemic function and the use of narrative quote sen-
tences to act as turn fillers. Arguments were presented in favor of
abandoning the narrative quote sentences and placing quote frames direct-
ly within exchange structure. Longacre's (1976) treatment of exchange
structures and functions was described and criticized on the grounds
that structural detail had been lost in his analysis, and that his
treatment of out-of—phase deep and surface structure at the discourse
level was inconsistent with the general congruity of these structures
at lower levels. It was proposed that Longacre's deep structure roles
be used as functions (roles) within Pikean tagmemes describing exchanges.
Chapter Four discussed the etic description of exchanges in the folktale.
Following the proposals for the revision of Longacre's scheme produced a
description which allowed several types of linkage between exchanges to
be discovered, and which allowed structural out-of—phaseness to be ob-
served. The emic analysis discussed in Chapter Five showed that the
folktale's exchanges could be contrasted to form emic classes, and that

72

 

73
exchanges appeared to be embedded in other exchanges.1 The systematic
nature of this embedding led to the formation of a hyperclass of poss—
ible exchanges. It was shown that some classes of exchanges do not re—
solve the initiating utterance, and that exchanges which contain an em-
bedding and a resolved initial exchange are not prone to linkage with

other exchanges.

6.2. Findings Concerning Swahili

Although this study has focused on theoretical issues, a number of
findings about Swahili should be accented here. The most basic of these
findings is that the narrated exchange exists as a unit of discourse in
Swahili. Other authors (Longacre, 1968; Waltz, 1977; Kerr, 1977; and
Koontz, 1977) have considered narrated exchanges to be types of para—
graphs, but no effort to show this for Swahili has been made here. Some
narrated exchanges observed here serve as constituents of paragraphs
(e.g. Nodes 9 and 13), while others, perhaps, may be considered to fill
paragraph slots (e.g. Node 2). Structurally, Swahili narrated exchanges
may have exchanges embedded within them, so that 'complex' exchanges are
possible (cf. section 5.1.). Also, it was found that constituents of
turns may be shared between exchanges, so that Swahili exchanges may be
linked together. Analyzing the exchanges of the folktale in terms of
embedding and linkage has clarified the relationships between the con—
trastive exchange types, so that a hyperclass of Swahili exchanges
could be posited (cf. section 5.2.). Finally, because many of the ex-
changes in the folktale could be seen as having an embedding, the total
number of functions for turns within exchanges could be reduced in

number (cf. section 5.3.). Swahili, therefore, can be considered to

74
possess a simpler set of exchange functions that those posited for
Philippine and Central American languages by Longacre and his colleagues

(1968, 1977).

6.3. Conclusions

The major theoretical finding of this study was that it was not
necessary to call on the phenomenon of out-of—phaseness to explain the
relations between grammatical structures and functions at the exchange
level. It was demonstrated that functions and structures could be seen
as congruent if the turns of exchanges were analyzed in sufficient de-
tail to show functions associated with substructures of individual turns
and linkage by means of shared constituents. This demonstration of link-
age between exchanges is a finding which extends previous work of ex—
change structure, for analysis of linkage via sharing has been non-
existent until now. Much of what has been accounted for by out-of—
phaseness in Longacre's work is analyzed here as dual function in situa—
tions of shared linkage between exchanges.

The analysis of exchanges embedded within exchanges represents an—
other important finding of this study. Previous work such as Longacre's
(1976) has suggested that such embedding may occur, but no previous work
has shown that a regular expansion set, or hyperclass, of complex ex-
changes may be formed. The existence of an exchange hyperclass is im—
portant because it parallels the existence of hyperclasses at the clause
and sentence levels.

The emic analysis in Chapter Five uses a system of function labels
which is even simpler than the already elegant system proposed by

Longacre. In the analysis presented here, counter—functions were

75
eliminated by including utterances carrying such features into embedded
exchanges, the function of which was countering. The combined use of
embedding and counter-exchanges (rather than counter—turns) enabled the
analysis to more accurately show the functional relations of an ex-

change's turns to each other.

6.4. Significance pf §pe_Study

Two sets of problems within tagmemic theory have been approached in
this study. The primary set had to do with Pike and Pike's revision of
tagmemics in 1977. An analysis of exchange structure had never been
done within the revised theory. Also, an analysis using the revised
theory could not permit out-of—phase relations between structures and
functions. Another set of problems concerned Longacre's treatment of
conversational exchanges. One problem here was that his analysis was
not detailed enough in that the constituents of a turn were not always
directly associated with their function in the exchange. A second
problem was that his use of out—of—phaseness at the discourse level was
inconsistent with the overall "in phaseness" of the lower levels.

This study has demonstrated that complex exchange-level structures
can be analyzed without recourse to out—of—phaseness. This in effect,
confirms the ability of the Pikes' revised tagmemics to account for ex-
change—level structure. Beyond confirming the Pikes' theory, this demon—
stration also has serious consequences for Longacre's treatment of ex-
changes and his general approach to tagmemics, for, by showing that
structure and function may be analyzed as congruent at all levels, it
removes part of his justification for treating deep and surface struc—

tures separately. Given the questions raised about the accuracy and

 

76
consistency of Longacre's analysis, and findings of an emic exchange
hyperclass with a simplified set of exchange roles, this study implies
that the newer, revised Pikean theory is preferable to Longacre's

earlier system.

NOTES

The participants of the folktale were of more or less the same social
status. It is possible that the analysis presented here would be
slightly different were the participants on different levels of the
social scale.

 

 

 

 

APPENDICES

 

APPENDIX I

Text ef_ppe_Folktale

Below is a copy of Kisa cha Punda wa Dobi as Reverend Steere

transcribed and translated it. Some of his spellings of Swahili words
do not follow modern usage. The paragraph divisions of Steere's text
have been altered in accordance with the analysis presented in
Appendices II and III. Each new paragraph is one of the numbered

terminal nodes on Chart 1 (page 86).

Kisa cha Punda wa Dobi

ghe_Sto§y efiphe_washerman’s Donkey

Node 1: Aliondokea kima akafanya urafiki na papa. Pana mti mkubwa,
jina lake mkuyu, umeota katika kilindi, matawi yake nussu yako
mjini, na nussu yako baharini. Yule kima kulla siku kwenda
akila kuyu, na yule rafiki yake papa huwapo chini ya mti.
Humwambia, utupie nami rafiki yangu vykula; humtupia siku
nyingi na miezi mingi.

There once was a monkey which made friends with a shark. There
was a great tree, of the sort called mkuyu, which grew near the
deep water; half its branches were over the town and half over
the sea. The monkey used to go every day and eat the kuyu
fruit, and his friend the shark was there under the tree. He
used to say, 'Throw me some foodg my friendk" and he used to
throw to him, many days and many months.

Node 2: Hatta siku hiyo papa akamwambia kima, fathili zako nyingi,
nataka twende kwetu nikakulipe fathili. Kima akamjibu,
ntakwendaje, nasi hatuingii majini, nyama wa barra. Akamwambia,
ntakuchukua, tone la maji lisikupate. Akamwambia, twende.

Till one day the shark said to the monkey, "You have dbne me
many kindnesses, I should like for us to go to my home, that I
may repay you for your kindness." The monkey answered him,
”How shall I go? we don’t go into the water, we beasts of the
land." And he said, ”I will carry you; not a drop of'water
shall get to you." And he saiah 'Zet us go."

77

 

Node 3:

Node 4:

78

Wakaenda zao hatta nussu ya njia. Papa akamwambia, rafiki
yangu weye, ntakwambia kweli. Akamwambia, niambie.
Akamwambia, huko kwetu tunakokwenda, Sultani wetu hawezi sana,
na dawa tumeambiwa ni moyo wa kima. Kima akamjibu, hukufanya
vema usiniambie kulekule. Papa akamwuliza, ginsi gani?

Akafikiri kima akaona, nimekwisha kufa; sasa ntanena uwongo,
labuda utanifaa.

They went half the way. And the shark said, "You are my friendL
and I will tell you the truth." He saidL 'Tell me." He saidL
’flhere, at home, where we are going, our Sultan is very ill,
and we have been told that the mendine for him is a monkey's
heart." The monkey replied to him, "You did not db well not to
tell me there on the spot.” The shark saidL ’hbw so?"

The monkey consideredg and felt, 'wy life is gone already; now
I will tell him a lie, perhaps that may save me."

Papa akamwuliza, umenyamaza huneni? Akamwambia, sina la kunena,
kwani usiniambie kulekule, nikapata kuchukua moyo wangu. Papa
akamwuliza, hapa, kunao moyo wako?

Huna khabari yetu? Sisi tukitembea mioyo yetu huacha mitini
tukatembea viwiliwili tu, wallakini hutanisadiki, utaniambia
nimeogopa, sasa twende zetu hatta huko kwenu, ukanichinje kama
utauona moyo wangu.

Papa akasadiki, akamwambia kima, turudi sasa, ukatwae moyo
wako. Kima akamwambia, sikubali, ela twende kwenu. Akamwambia,
turudi kwanza ukatwae moyo wako, tupate kuenenda.

Kima akawaza--ni heri kumfuata hatta mtini, akili nnayo
mwenyewe nikiisha fika.

The shark asked him, "You have become silent; don’t you speak?”
He saiaL ’Q'have nothing to say, because of your not telling me
there on the spot, and I might have brought my heart." The
shark askeaL ’Wave you your heart here?"

’Tbn't you know about us? When we go out we leave our hearts
in the trees, and we go about with only our bodies; but you
won’t believe me, you will tell me I am afraidk let us go on
now to your home there, and kill me if you find my heart."

The shark believed it, and said to the monkey, ’Zet us go back
now and you get your heart." The monkey saidL ’Q‘don’t agree
to that, but let us go to your place.” And he said, ”Let us go
back first and take your heart, that we may go on."

The monkey considered--I had better consent to him as far as to
the tree, I know what to do when I have got there.

79

Node 5: Wakaenda wakarudi hatta mtini, akapanda juu yule kima
akamwambia, ningoje hapa, papa, naenda twaa moyo wangu, tupate
kwenda zetu.

Akapanda mtini akakaa kitako kimya.

They went and returned to the tree, and the monkey climbed up,
and said, "Wait for me here, shark, I am going to get my heart,
that we may be off. ”

He climbed into the tree and sat down quite still.

Node 6: Papa akamwita. Akanyamaza. Akamwita tena. Akamwambia, twende
zetu. Kima akamjibu, una wazimo? Papa akamwuliza, ginsi gani?
Kima akamwambia, umenifanya punda wa dobi? Papa akamwuliza
kima, ginsi gani punda we dobi? Akamwambia, Ndiye hana moyo,
wala hana mashikio. Papa akamwuliza, ginsi gani kisa cha
punda wa dobi? Nambie, rafiki yangu, nipate kujua maana.

The shark called him. He held his tongue. He called him again
and said, "Let us be going. " The monkey answered him, "Let us
go where?” He said, ”Let us go to our home. ” He said, ”Are
you mad? " The shark said, "How so?” The monkey said to him,
”Do you take me for a washerman's donkey?” The shark said to
the monkey, ”What about a washerman's donkey?" He said, ”That’s
what has neither heart nor ears. " The shark said, ”What is the
story of the washerman's donkey? Tell me, my friend, that I
may know what it means. "

Node 7: Akamwambia, Dobi alikuwa na punda wake, akimpenda sana
mwenyewe. Akakimbia punda akaingia mwituni siku nyingi, hatta
akamsahao mwenyewe dobi. Akanenepa sana kule mwituni.

And he said, "A washerman had a donkey, and its owner was very
fond of it. And the donkey ran away and went into the forest
many days, till its owner the washerman forgot it. And it got
very fat there in the forest. ”

Node 8: Akapita sungura, akamwona yule punda, mate yaka mtoka, akanena,
nyama imenona hii. Akaenda akamwambia simba. Na simba atoka
ugonjwani, amekonda sana. Sungura akamwambia, ntakuleta nyama
kesho, tuje tule. Akamwambia, vema.

”And the hare went by and saw the donkey, and foam coming from
its mouth, and he said, ’This beast is fat. ’ And he went and
told the lion. [Vow the lion was recovering from an illness; he
was very weakly. The hare said to him, "I will bring you some
meat tomorrow, that we may come and eat. ” The lion said, "Very
good. ”

 

Node 9:

Node 10:

Node 11:

Node 12:

Node 13:

80

Sungura akaondoka, akaenda mwituni, akamwona punda, na yule
mke. Akamwambia, nimetumwa kuja kukuposa. Na nani?
akamwuliza. Akamwambia, na simba. Akakubali, akafurahi sana
punda. Akamwambia, Twende zetu, bass.

The hare arose and went into the forest, and found the donkey;
now that donkey was a she. And he said to her, "I am sent to
come and ask you in marriage.” "By whom?" she asked. And he
saiaL 'By the lion." And the donkey consented and was very
glad. And she said, "Let us go, that will do."

Wakaenda zao, hatta wakafika kwa simba. Akawakaribisha simba.
Wakakaa kitako. Sungura akamkonyeza simba, akamwambia, nyama
yako hiyo imekwisha kuja, nami naondoka. Akamwambia punda,
nnakwenda chooni mimi, zumgumzeni hapo na mumeo.

And they went, till they arrived at the lion’s. And the lion
invited them in, and they sat down. The hare gave the lion a
sign with his eyebrow, telling him,'This is your meat, it has
come with me already; I am going out.’ And he said to the
donkey, ”I am going on private business, converse here with
your husband."

Simba akamrukia, wakapigana, akapigwa sana simba kwa mateke,
naye akampiga makucha mengi. Akaangusha simba akakimbia
punda, akenda zake mwituni.

The lion sprang upon her, and they fbught: the lion was kicked
very hard; and he struck hard with his claws. And the donkey
threw the lion down and ran away, and went off into the forest.

Akaja sungura, akamwambia, Je! simba, umempata? Akamwambia,
sikumpata, amenipiga kwa mateke amekwenda zake, na mimi
nimemtia madonda mengi, sababu sina nguvu. Sungura akamwambia
simba, tulia we.

The hare came and saidh ’hullo! lion, have you got it?" He
saiaL ’Q'have not got it; she kicked me and went off; though I
have made her many sore places, because I am not strong.”

The hare said to the lion, "Don't put yourself out of the way.”

Wakakaa siku nyingi, hatta punda akapona madonda yale, na

simba akapata nguvu sana. Akaenda sungura kwa simba, akamwambia,
waonaje sasa, nikuletee nyama yako? Akamwambia, kaniletea
ntaikata vipande viwili.

They stayed many days, till the donkey was well of her wounds,
and the lion had got very strong. And the hare went to the
lion and saiak 'hhat do you think now, shall I bring you your
meat?" He said, ”Bring it me, I will tear it into two pieces."

 

 

Node 14:

Node 15:

Node 16:

Node 17:

81

Akaenda sungura mwituni. Punda akamkaribisha sungura,
akamwuliza khabari. Akamwambia, na mchumba wako anakwita.
Punda akamwambia, siku ile umenipeleka, amenipiga sana kwa
makucha, naogopa sasa. Akamwambia, hapana neno yalio ndio
mazumgumzo ya simba. Twende zetu, bass.

The hare went into the forest; the donkey welcomed the hare
and asked the news. He saidL ’Tou are invited by your lover."
The dbnkey saidL ’That day you took me, he scratched me very
much, and now I am afraid." And he saidL ’Hhis is nothing, it
is only the lion’s way of conversing." She said, “Let us go,
then."

Wakaenda hatta wakafika. Simba alipomwona tu, akamrukia
vipande viwili.

Hatta sungura alipokuja, akamwambia, chukua nyama hiyo ukaoke,
wallakini sitaki kitu mimi, ela moyo na mashikio ya punda.
Sungura akamwambia, marahaba.

They went till they arrived. The lion, when he had only
caught sight of her, sprang upon her and tore her in two
pieces.

When the hare came, he said to him, "Take this meat and roast
it; but myself I want nothing except the donkey’s heart and
ears. ” The hare said, ”Thanks. ”

Akaenda akaoka nyama mahala mbali, simba hamwoni. Akatwaa
moyo ule na mashikio akala yeye sungura, hatta akashiba. Na
nyama ngine akaziweka.

And he went and roasted the meat in a place apart, where the
lion did not see him. And the hare took the heart and ears,
and went on eating himself; till he had had enough. And the
rest of the meat he put away.

Akaja simba, akamwambia, niletee moyo na mashikio.
Akamwambia, yako wapi? Simba akamwuliza, kwa nini?
Akamwambia, huyu punda wa dobi, huna khabari? Akamwambia,
ginsi gani kutoa kuwa na moyo na mashikio? Akamwambia, wewe
simba mtu mzima hayakuelei? Kama ana moyo huyu na mashikio,
angalikuja tena hapa? Kwani marra ya kwanza amekuja akaona
atakuuawa, akakimbia, marra ya pili amekuja tena, bassi kama
ana moyo angalikuja? Simba akamwambia, kweli maneno yako.

And the lion came and saiaL ”Bring me the heart and ears.” He
said, "Where are they?" The lion asked him, ”What does this
mean?” He said, "This was a washerman's donkey, did not you
know?" And he said, ”What about there being no heart and ears?"
He said, "You lion, a grown-up person, and is it not clear to
you? If this animal had had heart and ears, would it have come
here a second time? For the first time it came, it saw it would
be killed, and ran away; and yet it came again the second time.

 

Node 18:

82

Now, if it had any heart, would it have come?" The lion said,
"There is truth in what you say. "

Bassi kima akamwambia papa, nawe wataka unifanye mimi punda wa
dobi, shika njia wende zako kwenu, mimi hunipati tena, na
urafiki wetu umekwisha. Kua heri.

So the monkey said to the shark, "And you want to make a
washerman's donkey of me. Take your way and be off home, you
are not going to get me again, and our friendship is ended.
Good-bye. ”

 

 

APPENDIX II

Abbreviations

act.
adj.
adv.p.
agr.
ans.
ant.act.
c.pro.
c.ques.
cl.
clar.
cmmt.

cmplx.lk.s.st.

cmplx.s.st.
cncl.
cnjd.s.st.
cnsq.
cntr.
cnv.fr.
compl.
conv.
crcm.
crd.act.
crd.s.clstr.
CU
d.cl.rt.
d.s.rt.
d.s.st.
denou.
dev.
dev.act.
dis.
eval.
ex.fr.
fr.
i.act.
i.cl.rt.
i.s.rt.
i.s.st.
imp.s.st.
indp.
init.
invt.

IU
lk.s.st.
mono.
n.q.s.
nr.ex.
par.

action

adjective

adverbial phrase
agreement

answer

antecedent action
counter-proposal
counter-question

clause

clarification

comment

complex linked sentence stem
complex sentence stem
conclusion

conjoined sentence stem
consequence

counter

conversation frame
complement

conversation

circumstance

coordinate action
coordinate sentence cluster
continuing utterance
declarative clause root
declarative sentence root
declarative sentence stem
denouement

development

development action
discussion

evaluation

exchange frame

frame

initiating action
interrogative clause root
interrogative sentence root
interrogative sentence stem
imperative sentence stem
independent

initiation

invitation

initiating utterance
linked sentence stem
monolog

narrative quote sentence
narrative exchange
paragraph

83

84

par.cmplx. paragraph complex

prem. premise

pro. proposal

prob. problem

prom. promise

p.s.lk. partially shared linkage
q.f. quote frame

qt. quote

ques question

rem remark

reso resolution

resp response

rslvd resolved

rsn. reason

RU resolving utterance

s. sentence

s.clstr. sentence cluster
sb.cl.rt. subjunctive clause root
sb.s.rt. subjunctive sentence root
sb.s.st. subjunctive sentence stem
sbs.act. subsequent action

sh.lk. shared linkage

st state

sub-denouement
terminating utterance
unrslvd. unresolved

sub-denou.
TU

APPENDIX III
Etic Analysis
The following diagrams represent the etic analysis of all nodes in
the folktale which contain conversational exchanges. Narrative para-

graphs were not included. The etic analysis is discussed in Chapter

Four.

85

86

ocos
Neoum

wnOQ m3 MUCDQ MSU wwwx $0 mLzuuzLum ELOHW ”H Hm<Iu

 

.aocmn

 

 

Acoum

.OCOE

 

.zocmu

.xmm .xFQEu.Lma

.>mn .xpasu.emn

.OCOE

.>:OU

.cma

 

.Fmsou.xmmm

.cma

 

.xommunzm

.Lwa

 

 

.uoo.>wu

.ch

 

.uum.u:m

.zocmvlnsm

.xw.&:

.Lma

 

.xmmaunsm

.cma

 

 

.pum.>me

.Lma

 

.uum.u:m

.cma

 

 

mc_mwum

 

mc_uuwm

 

rummmmlll .xpan.>coo .Fasou.xmma

 

.cma
.>:oo

.cwa

 

.xmma.uow

 

.uom.>mu

 

.x—QEu.>cou

 

.>oe

 

.nocm

 

mcwmmum

 

.>:OU

.>:OU

.xw.L:

.cea

 

mcwuumm

xmxcow m.:mscm;mm3
m an u.:oz w; mxem xmxcoz.wfi

Amxcoo usone
wewe—axe mew: new mxmm cow4.NH

meow use
Heme; mwcw; can mumwmm mLm:.©H

umvw>wc pews
.cm—wa mama .mcczpmc xmxcoo.mfi

eczema
o» mecoo moo:_>cou mcm:.¢~

c_mme :e_a .pmmc Amne.mfi
wewe—axe cow; .mcczamc memz.NH
mmamomm Amxcoo .pcmwe xmge.HH
m.:o_4 pm Home __<.o~

mecoo m:o_u_moaocg mcm:.m

:o_4
new; mcmPa .zmxcoo mmmm mcmz.w

aoxcoo to ccaoemxowm.x
xcmcm mmmsemc xmxcoz.o
mmaeomo moxcoz.m

mccsemc
.xcmnm mpooe .mmwp xoxcoz.e

m:m_a zmxcoz .mmmmmwcoo xcmzm.m
xmxcoz mmuw>cw xcmcm.N

eczocmxoem.H

 

87

.xw.L:

 

 

 

 

 

 

 

 

 

 

 

.mmmg .pc._o.nm
.e.¢ .pe._o.e
.mce .pc.m.nm

. .me_u.m . . .
mam .oea p; F8 u
.e.8 .uc._c.e
. m.m .Em; .u;.m.u
.mmzm.o .pncu . L. u.e
v. .mwzc.o p F .
.m.c .ue.m.c
. .pe.m.nm

.oeq .pm.m.u_ Em;

.me o.m .oca p: m e

.0; P . c.m.
.58; p e

. L.m.
. .8 .pm.m.e .w.c p e
m weep .a.>ee

 

.Lm.xm

mema_ee< seem ”N mooz

.om panda
.wuchH.n
ass: @Nop 8: w:<
.mvnEwZme<.mv.N
.xom :8:Op
NNes :8p83 k8 mo:fi 8 no:
.mpmazxwva whee m— w:ow.o
«80m mx8e NNeS M
.m:x:;o:xmuz.n
«Se: on fie8m 8: fi:<
.mwnEwBme<.mm.N
.mN8Ee:8 w:8m 83
«sop83 8&p op:e om p.:8w 83
.mcsmn m3
use»: .wcwwes wwmcwzpe; eme:.o
«om H NNee 38m
.mhmucmmeuz.n
.fimamem:8 m8«:82
.znwhscxm wew¥.mN.N
.mmmefl:ex exom :sxpos S85 N on
.e_eeeae mae_:xaxie.a
888mg :88 on 8E08 89 88h 9:83 N
:pmzx much¥ mxmpmc.w
«p8msm we mmm:fi:e« 988M
.eeeexe oxaN e_eeeea.c
ammx:oé on fie8m «:8:m
.mswx 8Nw_:3sexe eqca.n
memo 8:8 :33.
axe: seem eea:.aH.N

 

88

mem>_e=< seem ”m mooz

.ms amp:
ease as maeeaaa .me: a Name MN.~ ace

.oc .me_o.m .emwwcmpw munm_ .omcoZ: mcmcmp: mmmm.o
.me—o.m‘ w «w8mw mw8mhm8 E_H

 

 

 

 

 

 

_e>a .Fe>a .em.a.e .aczx eem_2¥aeez.e
.wmmwN88x w:8 p:mxo:p m8«:oE
.mnc .pm.m.x— .mcowxw mewx wcwxrwmx<.mm.m
IJmMMMIII..xm.Lc mom 38m
.mcze.c .ec.m.e edema emeee.n
new: moxm8 «s8:m w::
. .8 .pL.m.v .me—szmem mawa.mo.m
.mac v .ae eeeNNae ace Name 88 a.eeee so:
. m>w .pm.m.v .m—sx wpsx mwnEmwcwm: me> m>:m+:x::.n
— new: wm&838:8 mox:n¥
. .c .pL.m.U .znwhmew me¥.mm.m
.mpu .xm.;: w .ps88: m.m8«:oe 8 we wagesommsm

. m:eoefioe 8&p w:8 «Nose mama we m:e«
.>coo $88 am:eom 8&.83 8:8:3 888mm s80 sq
.mva 83 0505 w: mzwnEm853p

 

 

 

 

 

 

 

 

.Ew; v.wwcw came a: .mcmm wNmzm; 3pm:
.QmmL . weep_:m .evcmzxoxmczp :mex ox::.n
«Se: 8% @e8m 8: 8:<
.e.c .ps._o.e .mwnEeZEex<.e¢.m
.85 Nmoa
. . .pg._o.m .mwnEmwz.n
.na: .xm.Lc egg 8 «Se: amen 8: m:w
. .8 .wL.—u.u .menEmzsmx<.em.m
L .eease see ace Name 8883 N
.0; .pe.m.e .wrwzx Manamzxeec.o
.o; .pm.m.u w .8:8e&% ms 8:8 88m
.58; .pc.m.e .mzmz smcmz _xeeem.a
«Se: 8% 888m «a8:m
.w.c .p;.m.u .ewnEmZEexm maea.eN.m

.m83%N8: p:8s mm:&
.awh: m» :mzc mum: .om~ «eccexez H.m

 

 

0L$0>CU OHWIWOU

 

N H
0' W n U D!

O

4: (a)
U‘ N U’ n!

U1
D

U’

01
U" O!

O

\l
W

n

(1)
0|

0'

L0
DD

.Papa akamwuliaa.

im,

.Umenyamaza

You ’ve become quiet,

.hunen

u speak?

do yo
.Akamwambia,

And he said to him,

.Sina la kunena a,

I have nothingw

.kwani usiniambie kule kule

nikapata kuchukua moyo wangu.

because of your not telling me

back there I might have gotten
bring my heart.

.Papa akamwuliza,

Shark asked him,

.Hapa kunao moyo wako?

Do you have your heart here?

.Huna habari yetu?

Don’t you know about us?

.Sisi tukitembea mioyo yetu huacha

mtini, tukatembea viwiliwili tu.

en we roam about we leave our
hearts in the trees and go about
with only our bodies

.Walakini hutanisadiki,

utaniambia ninaogopa.
But you won’t believe me, you’ll
tell me when I' m scare

.Sasa twende zetu hata huko kwenu,l
9U

ukanichinje kama utaona moyo wan
Now let's go to your place and
kill me to find my heart

.Papa akasadiki,

Shark believed him,

.akamwambia k1ma

and to ldM

Mk
.Turudi sasa, ukatwae moyo wako.

Let ’3 return now, so you may
get your heart

.Kima akamwambia,

Monkey told him,

.S1kubali,

’t

agree,
.ela twende kwenu.

except to go to your place.
miba

.Akamwa

And he said to

.Turudi kwanza ukatwae moyo wako,

Let' 3 return to get your heart,

.mmmkmem

so we may get to go.

.Kima akawaza.t

Monkeyt

.Ni heri kumfuata hate mtini,

akili nnayo mwenyewe nikiisha
fika.

It’ s best to go along with him
until the tree, I'll know what
to do after we arrive.

MODE 4: Etic Analysis

 

 

 

 

 

 

89
d.s.rt. q.f.
d.cl.rt. rem. ues
ues s.clstr. '
i.cl.rt. '
d.cl.rt. “-
d. 5. st.
s.clstr. ans.
H;m rem.
d. s. rt. q’f‘
1 S rt c.ques.
i. s. rt.
s.clstr. ————————3§§4—
cmplx.
lk. 5. st.
crd. s.
clstr.
s.clstr.

 

   

 

 

 

 

d.s.rt.
d.s.rt.
d.s.st.
d.s.rt.
res .
d.cl.rt. r0 d.s.st.
d.s.rt. ro.
d.cl.rt. ‘f'
d.s.st. ro. es
res cmplx.s.st. ——-—lL—Eé—
sb.s.st.
d.s.rt. q.f.
s.clstr. eval.

nr.ex. _lfll£;_
nr.ex. —2599L—
nr.ex. ——91§4—
nr.ex. —52§94—
M, Jk.

 

COHV .

 

90

.>:ou

 

.OCOE

.ammg

.x0.L:

 

.OmmL

 

.L.u.
dillupu
.xw.L:

 

.uL.m.w

 

.X .L: . L.m.F
.mg m .muzcd u .

. . L.m.
1JuJaWIII111IIII|IIII1IIII1III u v

.Eog .ug.m.u
. mm;

 

 

 

.uL._u.u
.uL.w.nm
.uL._u.u

.u;.m.w

.ug.m.u
.uL.m.nm
.uL.Fu.v

.uL.m.u

 

. .tfie
. . mm;
Jail E: ”H €13

.uum

mwmxpuc< u_um no maoz

.xgoum mppmu amxco: m.mH - H.m

.33 has H on «wxumfik as «us Ema
.msnzx macaw: .zmcm» +x_wcg .w_nEm_z.u
«$833» FEumemua 93 .x0 ESQ man mLSS
~wnou a: aucsa mcu om_x _:mm _m:_n.a
.Em: waxes «aq:m
.u~_—=zeaxm mama.mmfl.o
.mLUm x0 «saw: u «soagwa uxo mxh m.mm
.ovx_wme ecu; are: ozoe can; o»_uz.n
.5: 3 3% me wé
.cwnEm35a¥<.uHH.o
«mmxrnfi w.:§:§mmmu3 u uxcad mwxu m.uu:k
~_nou m: mucza ago umwx wcmm wmcwa.n
ammxroz ~5me «LEE
.ms_x m~wpzzsmxm mama.uofi.o
.3.»ng Fragmxmca 3 x3. ms 33» so»
._aou a; auczn excuawcman.n
.53 R. NXEQ mu«:92
.mwnEszmxm mE_¥.om.m
wow 30:
t:% .55:
‘Ehx memu «imam
.MN_—:3Eaxn aqm¢.~m.o
.398 we}:
.oE_~a3 m::.n
«Em: 0“ WwUm m: fire
.~_nEm35wx<.o~.o
.mogm ms 3 cm 3qu
.=umzx wvcwzh.n
«Ev: ou fimcm m: fixw
.mwnEszmx<.ue.o
«.233 cm Pug
wine: wu:w3~.n
.EHE E93323 b.3232
.znwhsmxu mE_¥.mm.m
$.33 .30 to ma 3&3
.=um~ mu:w2w.n
~53 cu 33% w: HQ
.M+aEmzem¥<.uv.m
.rgcmd meNuo w:
.ncmu mu_:Em¥< m.m
.uwamm memuSNL u: w2<
.m~mEmAcax< N.o
.Ew: umNNcu «Luxm
.wuwzecxu can; H.o

 

91

memx_mc< atom "w maoz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.woow
. .hum .mEo>.n
gmmg "Ewe eh wwsm we wxw
. .c .pL.Fu.v .mwnsszmx<.mm.w

.x . $ .udm has m3 huah ow

.pum.mnm m a: .Ew; .pm.m.nm m meax mnzu.o
1J1!Mfillll..pm.m.u 30&&osou Ndswis :6 :0 :@&Q NN_H

o; . .pL.m.v ocmwx mam»: mum—sampz.n
0; "she ENQh mesa

. .v .p;.m.u .mwnEMZwam m;:m::m.ov.w

m .«dms mama mua ma firs

. .pg.m.u .mcmm muc0¥wEm.n
.me—u.m Umco «wmeNNw :6
.pEEo Eofik mrwgmaooma mug nova 30%

.Lwa . .pL.m.u .wcmzhcoms mxopw anavm mz.mm.w

pm .fiowq wNou EEG kima mm
. . .pm.m.x_ .mnevm Minamzemxm mucmmx< N.w
pow mom .pnx we Naswss megs
.ug.m.u .wwc mcocmew m5w>z.m
.1w11141111 .m.c.: JawUm ma
gum mam .uL._u.u .mcmcmxm.v
am:wmoo&fi firs
. . .ae.m.v .mxouaxmx mume.o
I4MWM4flIII .pm.m.x— pow mam .mmxfiow hdau 36w wsd
. . . .ug.m.u .cncsa ops» mcozmem.n
pom mam nag Emmwsm mgsm
.pg.m.u .mgzmcsm opwqu<.cfi.w

.pom._

 

92

mwmA—mc< Ume ”m mooz

.muB gxo :0 m@ m.umq

 

 

. .p;.m.nm .mmn .zpr wus¢3p.n
mm: .EE 8 3% ea
.ldldmllll .pL.~u.v .wwnEmZme<.mn.m
a .mmmd£ mama mUB fiia Emmxmd mmxrsm
1J11111111 .pm.m.x— .mucsq mcmm wcmaswmxm WFwnzxwx< o.m
pom mnm .sowq mm
. .p;.m.v .mnewm mz.n
mcm «ama wNou mm
uJilgWIIII .pg._o.v .mwnEmzsmx<.mm.m
.lublmlfimlfl .xm.Lc L. .ENNK muQVNmG womm.
1 Idldmllll .uL.—o.u .me_:3mem.n
w «ang mm
Idlllmqlll .p;.m.w «wcwc mz.mq.m
mm: 0 .30m 0% mmomo&&
.Lma ow mEom 0h Emmx mtwmg 56 H
,IWIJWIIIII .um.m.v .mmoazxsx mwzx mZEzgosz.n
o; «mma 0% Even ma wxw
,IWIJWIIII .pL._o.u .wwnEmzemx<.mm.m
w .mNBEmK c was mm«:ow usau 302
. .p;.m.u .me mucza w_:z wz N.m
uEEo .ummxpk ma“ 0% «smg wxc fi%m~ madm

 

 

.pom.~ .pm.m.x~ .wczpwzs mucmmxm mxoucoaxm mgzmczm H.m

 

93

.meE

 

 

 

 

.—w>w .yL.m.U
.%.U .uL.m.U
.EwL .pm.m.u

gal .LHm—U.m
.Eme .me_u.m
.mcm .HL.—U.U
.$.c .pL._u.c
.meU .HL.m._.
.w. .p;.~u.v

erJmlll..pm.m.x—
.wum .pL.m.U

 

 

.L$.xm

mtmz_mc< atom ”NH mooz

.maaem
.mz mw_:p.p
nxowq END“ mgdm
.mnewm mwnszmxw mgzmczm.mm.mfl
.thtm&hm k0 «USN ms %0 mmxcoma
mm£0g8&om bros :mugom maca H w§6
.:>:m: mcwm snmnmm
wmcwe mucoume mwpwstc WE_E m:.u
mhko 26%
fiic .m%00£ &m& ath mg flea mam
.mme
muchmeo .mxmpme mzx mmwawcosa.o
«awe hwm h.=www H
.mpmqssxwm.n
«Eva wash we Esq
.mwnEozamx<.mN.NH
mama hmm 30m fiwfi «Sowq NNmk
mmmeEmE: .wnswm mw.u
qswa 0» Beam wsc
.mrpEmzsmxm.n
msdo mean
mgsmcsm mwmx<.mH.NH

94

.uma

 

 

 

 

 

. .u;.m.u
. . oL
Jamal E: m
. mm; .pg._o.nm
.w.c .pL.—u.v
.p;.m.P
.o; .
,1411141111 .xm.;:7 . .meFo.w_
o L
p a mom o .mmzc .pg.m.w
. . .pa._o.u
. . . i
1J114wnun pm m JF
a .pum .pL.m.n
.La.xm
.pm.m .pum.c;u .pm.m.ugo
.pom.u:m .xFQEu flHHHHHHHHHHII .u;.m.u

.pum

$39: 0.8 ”2 ES.

.mmomwm 039 092% he hxu NN.H
.__w3w> wucmaw> mpmxwmpc.u
qms 0“ he miwgm
.mwmewcm¥.n
”Eva wNoh we use
.mrnszsmx<.mm.mH
«NGENSU sxom 30m mxwae N wNzoam
woxmx mew»: mmpm_:xwz.u
«30: «swap 30m 0% weak
Nammm wwmcom3.u
"Eva 0% when firs
.mwnsmzsmxc.n
fiowq 09 firms mgum
mnswm mzx «Langsm wucwmx<.wm.mH
.mEQSnm mama msdoma sown fits
JnmN6m£ mESSOB w.mm«:om Nwh::
.mcmm :>:m: mpmamxm mpswm a:
.mxmz encoume mcoamxm muczq upm:.n
«mmsw miss wmhmma mm&&
.wmcwx: :xvm mmxcxm3.mfi.mfi

 

95

.Lma

 

 

 

 

 

 

 

 

 

 

 

 

.pom.acm

.QmmL .ws.m.nm
.w.c .pL.Fu.w
. . . .pL.m.u

.omm; xm L: ._m>m .me_o.m Em;
._m>o .u;.m.u
.$.c .uL.Fo.u

r111
.Em; 1v.mePu.m
.ammL

.$.c .pL.m.u
.oga .pe.m.u
.nog .xw.g: .a.c .pg._u.v
.L$.xm .pm.m.x_
.pg.m.u

mwm>_mc< owpm uqfl maoz

.om m.umm
.mmn napr mu:m3%.a
«Eva ow wwsm mam fifiw
.mtnamzec¥<.mo.efi

.mmamanoo mnowm mss mae umxh m.pd&&
.mnE_m a» o~E=mEszE own: or_w>.o

«mxwahor m.h~
.o:m: m:mam:.n
asma wNou ma w:w
.mwnsmzemx<.am.¢fi

.sdmk H 30: mmscmu mwa auws
ms ova ma ms pimm 30m swag haw udafl
.mmmm maomomc mwngums a:

mcmm mquwcwsm mwawawcmE: my? :x_m.n

«Ewe wmoh mmxmom
.mwnEmBmem mvcza.m¢.¢H

.som mNNSQ amaoN 930%
.wpwzxmcm oxmz anaggue mz.n

q&m£ @NQQ ma wxv
.mwnamzemx<.mm.vfl
.mam: ma“ Eva

wmxmu firs m&dm wmsommm3 mmxroq

._Lmnm; «NW—zzstm

.masmcsm mcmwnwgmemxm muss; N.¢H
.umm&p% mag ou:w h2m3 mach
.wczptza wgsmczm mucmmx< H.¢H

 

96

.Lma

.pom.mnm

.xw.Lc

 

.Qme .230:

 

.inw .pL.m.u

. .pm.m.u
.o; .me_o.m Em;
.0» p; m new

.2406
IAHJWIII..pm.m.n
.Lm.xw

 

mEru .FU.>UM

.pmao:

 

 

.puw

.xFQEu

 

.pum.p:m

55.3:

mwm>_mc< owwm ”mH mooz

.mxnuxs
.mnmcmgmz.n
«Eva ENQh m&dm
.mwnsmzsmxm ma:m:=m.m
.masm @26 assma m.mm«:ow mag 99%
ummoam quwah G «:63 «_:ofi M «3m
.mcczq a» owxwmms a: ozoE
«Fm ,vaE :pwx wxmpwm wctmew3.u
mp@ «000 fixs poms mwau mde
mmxomx: oat; mama: m:x:;u.u
«Eva 0» wean ma firs
.mwnEwBmem.n
nussu mg6m :m&&

v.mH

.mnszQVFa agsmczm mpm:.am.mfi

.mmmmwa

03% ou:« mma mash firs gmx tom:
hasmN ma «ama 36m :owu mu teem aw
.v_wzw> mvcmqw> muwxawxm
mwngamxm asp mcozaoQWFQ onevm
.fimnwaxd mmaw quxx hxma mmgm
.mxwimxmz mpm: mucwmxwz

m.mH
H.mH

v fl

 

97

.>:OU

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

._m>o .ug.m.u
.$.m .ug.m.n
.pm.m.v
.ug.m.n
.Emg
.gm.m.x_
.ocoe

.mcm

.mco .pm.m.w
. .xm.gc . . .
omwg .mmzc.u pm m _
.m.m .uL.Fu.n
.mmac.u .um.m.w
.u.mi .¢L._u.o
. .uL.m.w
.Eog .Lompu.m mm:

.Emg .ug.m.u

.mcm
.w.m .pL._u.n
.mmzm.o .pg._u._
. . .ug.m.n

.na; .xo.gc w a
.mm:c.u .ug._u._
.%.m .ug._u.v
.ogm .pm.m.nm
.u.c .ug._u.n
.ug.m.n

.Lw.>:ou

mmmAch< owum ”Na mmoz

.mza» mad nukes axe»
.oxox ocmcme ._pwzx.n
.Ewa was» town
.mwnsmzemxm caspm.cm.m~
«mace mama um
waxes .uhdma u wua Eda a» kw .NNma
waw3¥__mmco oxos can Manx .wmma.u
«mswa waoomw d :wcmu msdo «N
.mcmu ansxmsm _pwa a» mgmz.o
Jnmmuumm and vaNma wa waxes 9»
36m 9w mesa aw mEmn gmmmk ma“ mom
.m_newxmxm .mzmzzxmum mcomxm
.mnaxmao n~cmzx a» meme _:n3x.mm.afi
mxwumd mama wmxxxuma mama
a» uNxoa mmdm axe «Lama 6 Eva a» kH
moan; new»
mfl:x__umcm o_xwmms a: ozoe new «aux N.NH
matchmamwzz
32% :3. 38 £03 gigs U :8»
www_m:xmez mE_~E :uE onswm m3m3.n
.Ewa was» ma Ema
.mwnEmzemx<.mm.NH
«madm firu
uscma U maoca a» “Ga“ um aw 30a
wowxwmme
m: oxoe a: mzzx mouzx _:mm wmc_w.n
.Ema 0“ Exam ma wrw
.o_asmzsmx<.mm.m~
«362a 30m m.afima
~_;mnmc m:=:.u
ammarnfi m.:dsxmam63 6 mm mma&
.wnou m3 nussu _: :xzz.n
«Ema mach ma wzw
.awasmzamx<.cq.ma
ES
«w:_c m3¥.n
.s»: wmxma nova
.mN__:3meo answm.mm.afi
mama» mad mmmaa
~_sz exm>.n
«Ema was» ma nzv
.mwnsmzsmx<.mm.aa
.mkcm was “Lama map ms mamkm
.owx_mmE m3 ozos mwum~_z.o
.Sua 0» awum mzw
.mwnEmzeax<.n
«msdn mama

.moe_m mfimx<.cc.ac

 

APPENDIX IV

Emic Analysis

The following diagrams represent the emic analysis of all nodes
diagrammed in Appendix III. The emic analysis is discussed in Chapter

Five.

98

 

99

 

mamapmc< ovEM ”N mooz

 

 

 

 

 

 

 

 

 

 

 

 

.om m.umq
. .pL.—o.nm .ov:w3~.n
game «Ewe esoa we ere
. .U .pL._o.c .mwnEmzsmx<.m¢.m
m .xom £030e
NNwB smnes k0 mogfi 6 no:
.mcm .pL._o.nm .mpmasxwmwfi wwme mF m:op.o
.me_o.m_ .xom mmuu was H
.mcm .ogq .wL.Fu.u .wsxzcusxmaz.n
«E®£ 0% wmcm m£ wsw
. .U .uL.Fo.v .mwnawzsmx<.wm.m
m .mNcswxa fitfim m3
«amass m£u oumw om p.sgn m3
% .X0.&E .MLLQD 03
.xm.L: .Emg .pm.m.v Manx: .wcwmms wwmcwspc; wmmc.u
.pm.m_ 5cm H NNes 30m
.mwsm .uwcu .mmzc .pL.—o.v amflwccmmepz.n
dnmamamxd mmxioé
. .c .pL.m.U «:nwnsmxa wEw¥.mN.m
w .mmmefiswx xsom :axnma mus N om
.Em; .pg.m.nm .w—wsumm mQ__:xmch.w
. . . mode ass 0“ msom on son “:63 H
.0; pm m u .0; .pL.m.v :um3x mecmzx mxmpmc.v
.oga .me—u.m «udmxm we mmmstwx s30»
.Ewg .pL.m.U nwmcwxc ome FFTLumm.u
«mmxrnz on wwcm xsd£m
. . .ug.m.v .mswx me_:3mem mama.n
. .U .pm.m.u w mfiw wSQ SN&&
w warp .Q.>vm oaw; 3xwm mpmz.mH.N

 

 

.g$.xm

 

100

 

mrm>_mc< owe“ ”m mooz

.mE mmm£
NNeB he mmcxamm qmvN 6 Nmmn NN_H 30%

.og .me_u.m .mmewcmpw mvnm_ .omcoz: newccp: wmcm.o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.56 (EB 68w @8ch 5L.
.m ._m>w .pm.m.u «mesx mcmwzstwz.n
Jewmemuma fins u£®3©£p mm«:s€
.w.c .pm.w.x— .wcomxm wEFx wewxwmmx<.wu.m
IJdeml .xw.L: wow 30%
.mmzc .pa.m.w Macao Tween.n
Ill .xw.LC «ES m m6 2
.omm; .£ w x «&d&m w w
.$.U .u;.m.v .wprzzsmxm mamm.mm.m
as SENR so: 3% % regs so»
.Ew; v.pm.m.v .w—ax w_:x mwnEmeWm: wEo> m»:m%:x::.n
._w>w «Ew£ wmmmsmxs mmNEQE
.m.c .pg.m.u .znwhmem me¥.wm.m
.xm.L .psdmx m.mw«:os d we @maeaemmsm
l. c . .
mwn mswuwwms m&« ESQ «Knew msma we mxwx
$30 queom maama mam£3 mede &:o “V
.20; v.p.m.m .mex m3 omoe w: warnEwm53p
.QmmL .vnco mzmn m: mcwm wNmZm; 3pm;
.>:ou . wcmp—sm .mucmZxoxmcsp :pm3x 0x:I.n
«se£ 0% wean m& wiv
.m.c .pL._u.v .wwnEmzamx<.mq.m
IJMMQMI xm L: .mE NNu&
.ogm .pL.Fu.m .mwnawwz.n
«Ee£ fimou m£ w:w
.m.c .pL._o.U .mwnEm3me<.mm.m
.nog .xm.Lc .§u3&n m£p 30m NNme NNwB H
.0; .pL.m.u .Tszx ornamzxmp:.u
.OLQ .um.m.u_ w desmv&% m5 mad 30M
.50; .p;.m.u .mzmz :mcmx wxwemm.n
«Ew£ 0% wedm «&d&m
.m.c .pL.m.u .mvnEmZmew mama.mm.m
.mdskN6£ nxms mmgm
.pm.m.u .mwn: m8 :m2: mpm: .omN mucwmxm3 H.m

 

 

.Lm.>cu

.5

.2a

.3a.

.4a.

.5a.

.6a.

.7a.

.83.

.9a.

.1a.Papa akamwuliza,

Shark asked him,
b.Umenyamaza,

You've become quiet,
c.huneni?
db you speak?
.Akamwambia,
And he said to him,
b.Sina la kunena,

I have nothing to say,
c.kwani usiniambie kulekule
nikapata kuchukua moyo wangu.
because of your not telling me
back there I might have gotten
to bring my heart.
Papa akamwuliza,
Shark asked him,
.Hapa kunao moyo wako?
Do you have your heart here?
Huna habari yetu?
Don't you know about us?
.Sisi tukitembea mioyo yetu huacha
mtini, tukatembea viwiliwili tu.
When we roam about we leave our
hearts in the trees and go about
with only our bodies.
Nalakini hutanisadiki,
utaniambia ninaogopa.
But you won’t believe me, you’ll
tell me I’m scared.
.Sasa twende zetu hata huko kwenu,
ukanichinje kama utaona moyo wangu.
wa let’s go to your place and
kill me to find my heart.
Papa akasadiki,
Shark believed him,
.akamwambia kima,
and told Monkey,
c.Turudi sasa, ukatwae moyo wako.
Let's return now, so you may
get your heart.
Kima akamwambia,
Mbnkey told him,
.Sikubali,
I dbn’t agree,
c.ela twende kwenu.
except to go to your place.
Akamwambia,
And he said to him,
.Turudi kwanza ukatwae moyo wako.
Let's return to get your heart,
.tupate kuenenda.
so we may get to go.
Kima akawaza,
Monkey thought,
.Ni heri kumfuata hata mtini,
akili nnayo mwenyewe nikiisha fika.
It's best to go along with him
until the tree, I'll know what to
do after we arrive.

0‘

CT

0’

U'

U'

U‘

0

O"

NODE 4: Emic Analysis

101

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.f.
d.s.rt. q
rem.
d.cl.rt. ques.
. ues s.clstr. ' _
i.cl.rt. ' nr ex init.
d.cl.rt. q.f.
d.s.st. ans. ans
rem s.clstr. '
cmplx. rem.
lk.s.st.
prob.
d.s.rt q.f. nr.ex.
1 s.rt ques.
1 s.rt gues. ‘ nr.ex. ans.
cmplx. ans. s.clstr. ans.
lk.s.st.
crd.s. rem.
clstr.
s.clstr. pro.
s.clstr. -—2594—
d.s.rt. sbs.act. nr.ex.-——gl§L—
d.s.rt. q.f.
d.s.st. pro
nr.ex. -——£fl££4—
d.s.rt. q.f
res .
d.C1.Y‘t.<E d.s.st. P852.
d.s.rt. pro.
d.cl.rt. q‘f' nr ex reso.
ro.
d.s.st. cmplx. resp.
sb.s.st. res ' s St.
d.s.rt. q'f' 1. k
n.q.s. -——lfl——
eval.
s.clstr.

 

 

7\ ' ”Rte.

COHV.

 

 

95.22... PEG 6 mac:

 

. 8.. .82. :23” 2.3 3.52 9: - ..N

.395. has H on first.“ 9: .uE Ema
.9. 436.3 .na“... 32:: 59:; ECE .aEEEzd

35:0 55...: .o; .Lumpud «mmxé PruEmsmus mfi .bu PS“; 33 3.5.3
.mm: #59. ~52. a: 3:2. Eu 3.5. .23 3:21.
|l J84: .5: 33.3 «53%
.omug

|.I&|I| .u...m._u .muwpzzsnxm 33.3.6
w .mhum .3 “rug: U uxgug 28 mxu Fem
.29. v.um.m.v 62.3.2. man; in: 935 can; abuzA.
.5»: on use” we wee
JHJ|II .ugéufi .mEEmESEéZd

.L cu J33... mmuvtgw Snumemca a «Scam RE» 3233
u I4mmmmrlllllllllllillllln .ug.m.w ~_nou a: mucza ago emu. .cmm wmcva.n
.xs.g=

$328: ~3me «53%
11‘ Jude 65.; 31:358.... 83.8.6
w $82.3» 9.35%me u .Sk ms 8.3 :9"
.59. I'll'v.t.m.u .52. a: 2.5.. 55:552.".
.25 “SE 3 3% $8282
fill' .266 6.55359: 2.5.590

. .xw..:._ new 30:
Loco I4lummlllllllllllllllnuil. .ue.m.. s.cam .mcwq.n
mm «5.9: fimxmu «scam.
1% J83»:

.mzu

 

- 1(12

 

imi'll Judi“. £53.58: 85.86
w .3.»an 933$
. .Emg v.u...m.u .9533 25A.
>=ou a :33 3 New» m... fix...
. $8.... . . .ugéué .oEEmESEdTw
m—v _ up a .mp3.» 9: 3 am mlmu
.ogm .u...m.n.m .326. 355...
«SE. 3 35m m: «2%
.ugéué .m..n=.m;=.§<.mo.w
«98.3 cm mlg
.26...— wwamz 2:655

 

 

.Lacu

 

..L.W.u .sn.n2m.m me_¥.sm.m

5:: .xmd: .9... 4.193 .33 2:85;.

 

 

. . .f.—co .Enamzano.<.mv.w
w a .rwumc @330 mm
. . .se.m.u .scms au_zsc.< m.©
.C x0 .urqum fimxwgmx m: was».

. 3.. .u... 36 .m~uE§:m.< N6
. . .Lum—u.m_ :55 meN60 «gm.
.C 2.3 . .uuw .tdfi .ufizefim 32 To

 

 

 

 

 

103

mwmz_m:< 0.5. "w mooz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.fioom
. .wuw .me>.n
amm. .EQx 0. wficm m. wmw
.$.c .pL.—o.v .awnsszmx<.mm.m
.xw..: .ucm mus m3 “6&9 ow
.poo.mnm .EwL .wm.m.nm .m—sx mmsp.u
.o; .pm.m.u 30&&osou msswxb :6 30m mxwma NN.H
.o. .pL.m.v 05mm. can»: mum—:xapz.n
«Ew£ wNOQ mash
. .c .pL.m.v .mwnEszwxm m;:mc:m.m¢.m
w .«Uma m&mn mss m£ wsd
. .pg.m.u .mcmm mucoxmsm.n
.L m u.m Umco «mmmrme :6
.pEEo p F 303% mrwmmaoumg mus :owu 30E
.Lwa . m .p..m.v .wchnCOm: axoum «germ mz.mm.m
p .xovq wNon @26 g:m3 mm
. . .pm.m.x— .mnewm wwnsmzemxm wvcmmx< N.w
pom mam .95. m. NUwas mw£g
.ug.m.v .mw; wcocmev mEm%z.m
a .38 g
.pL._u.n .mcmcmxm.u
qmfiwNooxw w:u
. . .u;.m.v .mxowwxmx mumE.u
lqmmmqmn. .pm.m... pow mam .mmxxow “a.” saw wgu
. . . .pg.m.u ,mucza wpsz mcozmem.n
wow mam «mg wmmmsm mgdm
.p..m.u .mgsmczm muramx<.mfi.w

:5:

 

104

.Loa

 

 

 

 

 

 

 

 

 

.ammg .pL.m.nm
.w.v .pg._o.u
.puw.mnm .pm.m.x~
.p;.m.u
.g;.~u.w
Iqmmmqmmmnlc .xm.L:
.p;._o.u
.pg.m.w
.OLQ .pm.m.u
.m.c .pL.Fu.u
.pEEu .pg.m.v
.pm.m.x_

 

.uum.w

mwma_wc< qum "m mooz

.mds x30 :0 mQ m‘umq

.mmn “upmN wu:m3&.n

«s«£ a“ wvum mgm

.mwnEmZme<.mm.m

.mmm6& mhwa was EEG Emmsmc mmxxoa

.wvczq mcmm wcmgz$mxm Wansxmx< o.m

.Eowq mm

.mnewm “2.3

«Aux wmow mm

.meEmZme<.mm.m

.Ew£ fimxma m£m

.Mvaszsmxa.n

mED£3 mm

wrcwc mz.m¢.m

.30m 09 wmomoga
0% qum on Emmx mnwmg Ed H

.mmoasxsx wnzx wzszpmawz.n

.&m£ 0% wwdm m£ wtw

.m?DEmzawx<.cm.m

.mNdEmk d mda mmxnofi 96x“ 30%

.me mucza m_3z m2

.kmw&9% m£u cu uxms wxd fikmm mash
.w::p_3E mucwcxm mxoucomxw mgzmczm

N.@
H.@

 

105

.waE

 

 

 

 

 

.Fm>m .p;.m.v
.$.c .p;.m.u
.Em; .um.m.u
. .me_o.m
mam . .
.Ew; me_u m
.mcm .pL._o.u
.w.d .uL._o.u
.mmsc .p;.m.w
.w. .pL.Fo.u
:JIJHHII. .wm.m.x_
m .uum .ug.m.u

 

 

.L$.xw

mwmx—mc< owEm ”NH mooz

.mxsom
.mz m_pzh.n
£03 38 2%
.mnswm mwnEszmxm w;:m::m.mm.NH
.&“m:m&um k0 «DUN m5 k0 mmxdoma
mm£oaa§nm mst :muuom madx H find
.=>:m: mcwm :nwnmm
wmcws mvcoume mwgmaw: wave mc.v
mkxo :da
@:6 «M%oo£ am£ £uw3 mE uw£ m£m
.mme
mucmzxmsm .mxmpma mzx mmwawcmEm.u
n§ Sm Pfimo a
.myng:xwm.n
”fig 38 g Bw
.mwnszemx<.wm.NH
m§®£ «mm 30m fiww arovq Nmmk
«mpmaEmE: .anawm mn.o
«Ew& 0» wwum @26
.m_nEm35mxm.n
mace mmdm
mgzmcsm anmx<.mH.NH

 

106

mvma_cc< QwEm umH mooz

.mmomwu 03h ouxw «w «:0 NN.H

 

 

 

 

 

.o; .p;.m.u .Pszw> ancmaw> mpmxwmp:.o
IdmmMHII .mepo.m «ms 0% «w mfiwam
. mm; .p;._o.nm . .mmpm_wcm¥.n
E®£ wNog m£ wsw
.w.c .pg._u.u .mVnEmBme<.mm.mH
«Naswmc axon 30m wxwka H fiNxo£m
.oL .pg.m.w moxmz can»: wmpw_:xwz.u
Idmdewmml .xw.g: .og .mepo.m «so: «thu 30m 0% gd§k
.mw: .pL.m.w «mmmm wnmcom3.u
«wa on wficm @26
.$. .pL.Fu.u .mwnEmzemxm.n
. .IfllJWII..um.m.x Sowq 0% ufim3 madm
Log m _ .pom .p;.m.u mas_m mzx mLzmcsm mn:mmx<.m~.mH
.Lw.xm .mxosgw mam: dequ towm @26
mandm£ mfissoa m~mm«:oQ N@#:3
.wcmm :>:m: mpmamxm mnEwm m:
.pom. Lu .pm.m.vgu .mxmx museums mcoawxm mvcsa upm:.n
.pm.m_ U «mmnw miss wmpmm& mm§m
.pom.pcw .xFan .pL.m.v .meTAC :xwm mmxmxm2.mfl.mfi

.vom

Agj— 44‘

 

107

.Lma

 

 

 

 

 

 

 

 

 

 

 

.QMwL .HL.W.Dm
u; 2.66
. .256

.LPmPU E0.»
.—M>w .m ._.m>0 .HL.W.U
.0; .266

JPIUMII .X0.LC a .Xw.LC

.EOL .LflmFU.m
.$.U .HL.m.v
.OLQ .wL.m.U
.%.U .u.L.—U._U
.22.xm .pm.m.¥_
.PL.W.U

 

.pom.pcm

mwmx_mc< omEm ”vfi mooz

.mcs kso :0 ma m~umq
.mmn .zwwN wucwzk.n
.Ew£ 0% wwcm mxm wnw
.awnsmzsmx<.mo.vfi
.mmSmanoo msowm mfis m£u pmxm m~ud&&
.mnswm m» oNssmsszE own: ow_m>.o
«m:®£po: m.g~
.ocm: mcmamz.n
«mmx wNou mm wsw
.mwnsm35w¥<.mm.vfi
.&Gm% H 30: mmsfimo mw£ £«m3
NE pwx mx ms «2mm 30m :m£3 msv «6&s
.mmmm maomom: mmngome w:
wcmm mmwawcwsm wmewacmE: m—w :xvm.n
«wa wNou mwm2om
.mwaEmzemxm mucza.m¢.¢fi
.303 mmmdn amaom $30»
.wpw3xmcw oxmz waszcua mz.n
«&m£ 2N0“ m£ wxw
.mwnsszmx<.mm.¢H
.mam: m&u
Ew& mem6 228 mmwm @mEQUNmS mm«:oQ
.wgmnm; me_:zsmxm
.mgsmczm agmvaw2m¥5mxm wvcza N.¢H
.ummka m£u 09:» 92mg madm
.wcspwzs mgsmczm cccmmx< H.¢H

108

.Lma

 

 

 

 

 

 

.awmg .czo:
.%.c .p2.m.v
. .pm.m.u
_ Em:
:JIIIIWIIII .xm.2: . .me_o.m

pom mam o: .ogq .p2.m.aE_

. . u.
. . .wm.m.u .0. p2 _ u
0 m «amp .Fo.>um

.Lw.xm

.pm.m.x_
.pom .XFQEo
.pm.m.x~

 

.pom.u:w

mvmz_mc< UFEM umH mooz

.mx:0:&
.mnmgmgmz.n
.50: 0N0: 0&0:
.mwnsmzsmxw m::m::m.m¢.mfi
.ms00 0:0 0:00: m.mm«:0fi 0:: &D%
«@0080 «@:w:: 0 #:03 9.:00 M :3m
.mucza ox owxwmma m: oon
m—m “were :pwx wxmpwm wcwmem2.u
m9: «000 0:0 #005 00:: 0x00
mmxomx: ozwc mam»: m:x:;u.u
«Ev: 0: 000m 0:0
.mwnEmzsmxm.n
«0300 080m :0&&
.mnsxoawfim «Lsmczm wum:.wm.mH
.mmomwm
03% 0::0 a0: 0&0: 0:0 :0: :00:
0000N 0: «R0: 300 :0wu m0 :00m 0:
.wpv3w> mucmafi> mpmxsmxm
mwszmem “:p m:o:anw_a mnEwm N.mH
.wmmwma0 m0:u pr:: 0:03 m0:&
.mx_mmxmz mp0: mucmmxmz H.mH

 

109

.>COU

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.20.):ou

.—u>0 .u:.m.0
.w.a .u:.m.u
._0:0 .um.m.u
.Emgm .u:.m.u
.20: .000
.Emgm .um.m.x~
Ia .X0.LC all .OCOE
.mcm .um.m._
.mmmm .um.m._
. . .uL.Fu.0
0 a
fildmmmulll .x0.::
.mmsm .Hm.m._
.w.m. .u2.~u.0
. .u:.m._
.50: .gumpu.m m0:
.502 .u:.m.c
.mca
. . .u:._u.0
all sxm-Lc $[m
.mozc .HL._0.W
Idummm .xw.:: .mwm .u:.m.0
.mmac .u:._0._
.na: .xo.:: .0um .uL._u.v
.020. .um.m.nm
.0.c .uL._u.0
.u:.m.0

mwm»_m=< 0:50 ":0 mooz

.0300 0&0 00:03 :30»
.oxmx caucus .w_0zx.g
as»: 0N00 :000
.ownsmxamxm mnE_m.mm.-
«0500 000: um 0N303
.0:00: a 00: 00: 00 :0 .0003
mmhsxppmmcm ozoe mam use: .wman.u
~0E00 0:0000 0 :0000 0:00 0N
.mcmu anaxmeo __wa ax cgmz.n
‘0000000
0:0 00-0x 0: 0N003 0» 300
um 0500 a» 05m» umxwk 0:0 :00
.mwnewxmxm .mzdszxmum mcomxm
.wnzxwam mNcmzx a» 0:05 _cmzx.cm.mH
0:0000
0:0: 00::300L 000: 00 0N303
0:00 0:0 0:00: 0 00: um kw
Noam; 0:00 mn=x__omco
o_x_mme a: oon ace mme m.-
«0:000:00:: 0.:00
000 0:0 «:00N 0&0005 0 a30»
~_0_w:xm>m;
wE_~E :HE mnE_m 0303.2
«8»: 0N0» 0: 0::
.m_nsmzscx<.mo.mfi
«0:00 0:0 0:00:
0 0x00~ 00 00:0 00 00 300
mowx_mms a: oxoe
a: 033x movax Paco wmc_0.n
«Em: 00 0000 0: 0::
.m_nsmzsmx<.mm.m~
«30:: :00 0.:0wm
Nvgcnm: m:=:.u
.m0«:00 0.:0EN00003 0 00 00:0
._non a: cvczq _: :A::.n
«E0: 0N0: 0: 0::
.a_anzamx<.ov.NH
0.03
~_:_: 03x.n
«Eu: 00x00 :0wq
.o~w_:3acxm one_m.mm.mfl
0.00:0 0.0 0.8:}
~_Qo3 oxa>.n
.50: 0000 0: 0:0
.mwnsmzsox<.om.NH
.0:00 0:0 0x00: 0:: 0E 0:0:0
.owx_mma a: 0x05 00u0__z.0
.00: 00 0000 0: 0::
.m_3503E0x<.n
.0500 :00q
.anE_m 0n0x<.mfi.~d

 

LIST OF REFERENCES

 

 

 

LIST OF REFERENCES

Bolinger, Dwight (1979) ”Pronouns in Discourse” in T. Givon (ed.)
Syntax and_$emantics, Yol: 12, New York: Academic Press.

 

Chafe, Wallace (1979) "The Flow of Thought and the Flow of Language” in
T. Givon (ed.) Syntax ang_Semantics, Moi, 12, New York: Academic
Press.

Crawford, John C. (1963) Totontegec Mixe Phonotagmemics. Santa Ana:
SIL Publications in Linguistics 8.

van Dijk, T. A. (1972) Some Asgects of Text Grammars. The Hague:
Mouton. "'

Erteschik-Shir, Nomi (1979) "Discourse Constraints on Dative Movement"
in T. Givon (ed.) Syntax and_$emantics, le, 12, New York:
Academic Press.

 

Fillmore, Charles (1968) "The Case for Case” in Bach and Harms (eds.)
Universals in_Linguistic Theory. New York: Holt, Rinehart, and
Winston.

 

Fries, Peter H. (1970) ”On Double Function in Tagmemic Analysis.
Anthropological Linguistics, 12:122-135.

 

Hale, Austin (1973) Clause, Sentence ang_Discourse Patterns jfl_Selected
Languages of_India ang_NeEal. SIL Publications in Linguistics 40.

 

 

Hasan, Ruqaiya (1978) "Text in the Systemic-Functional Model“ in N. U.
Dressler (ed.) Current Trends jn_Textlinguistics. Berlin: deGruyter.

 

Heanley, Robert M. (1909) A_Memoir of_BishoE Steere. London: Office
of the Universities’ Mission to Central Africa.

 

Isenberg, H. (1971) "Uberlegung zur Texttheorie" in J. Ihwe (ed.)
Literaturwissenschaft und_Linguistik, Moi, 1, Frankfurt: Athenaum.

 

Ismail, H. and Peter Lienhardt (1968) Ihe_Medicine Egg; swi a 5g
Nguvumali. London: Oxford University Press.

Johnson, Frederick (1938) A_Standard English-Swahili Dictionary. London:
Oxford University Press.

 

Kerr, Isabel (1977) ”The Centrality of Dialogue in Cuiva Discourse
Structure” in R. Longacre and F. Woods (eds.) Discourse jfl_Grammar:
Studies in_Indigenous Languages of Columbia, Panama, and Ecuador,
Part g. SIL Publications in Linguistics 52.

 

 

Klammer, Thomas P. (1971) Ihg_Structure of_Dialogue Paragraphs ig_Written

 

English Dramatic ang_Narrative Discourse. Ph.D. Dissertation,
University of Michigan.

 

110

 

111

Knappert, Jan (1967) Traditional Swahili Poetry. Leiden: E. J. Brill.

 

(1971) Swahili Islamic Poetry, 191, 1, Leiden: E. J. Brill.

 

Koontz, Carol (1977) ”Features of Dialogue within Narrative Discourse in
Teribe" in R. Longacre and F. Woods (eds. ) Discourse Grammar: Studies
in Indigenous Languages of Columbia, Panama, and Ecuador, Part 3.

SIL Publications in Linguistics 52.

 

 

Kuno, Susumo (1972) "Pronominalization, Reflexivization, and Direct
Discourse." Linguistic Inguiry 3:2.

(1978) "Generative Discourse Analysis in America” in N. U. Dressler
(ed.) Current Trends 13_Textligguistics. Berlin: deGruyter.

 

Longacre, Robert E. (1968) Discourse, Paragraph, gflg_Sentence Structure
1g_Selected Philippine Languages. Santa Ana: SIL Publications in
Linguistics 21.

 

 

(1976) Ag_Anatomy gf_Speech Notions. Lisse: deRidder.

Loriot, James P. (1958) ”Shipibo Paragraph Structure" (Later published
with Barbra Hollenbach in Foundations gf_Language g, 1970).

 

Mathesius, V. (1939) "On Information Bearing Structure of the Sentence”
(translated by T. 0. Yokohama in Harvard Studies 13_Syntax Eng
Semantics 1,).

 

Palkova, Z., and B. Palek (1978) "Functional Sentence Perspective and
Textlinguistics" in u. U. Dressler (ed. ) Current Trends in
Textlinguistics. Berlin. deGruyter.

Petofi, J. (1973) ”Towards an Empirically Motivated Grammatical Theory
of Verbal Texts” in Petofi and Rieser (eds.) Studies jg_Text Grammar.
Dordecht, Holland: Reidel.

Pike, K. L. (1954, 1967) Language in Relation to a Unified Theory of the

Structure 91 Human Behavior. The Hague: Mouton.

 

(1966) Tagmemic aflg_Matrix Linguistics Applied 19_Selected African
Languages. Santa Ana: SIL Publications in Linguistics 23.

 

 

 

(1976) "Toward the Development of Tagmemic Postulates” in Brend
and Pike (eds.) Tagmemics 191, 1; Aspects gf_1he Field. The Hague:
Mouton.

Pike, K. L. and Evelyn Pike (1977) Grammatical Analysis. Dallas: SIL
Publications in Linguistics 53.

 

Rieser, H. (1978) ”On the Development of Text Grammar” in N. U. Dressler
(ed.) Current Trends jn_Textlinguistics. Berlin: deGruyter.

 

 

112

Schegloff, E. (1979) "The Relevance of Repair to Syntax-for—Conversation”
in T. Givon (ed.) Syntax Egg_Semantics, 191, 11, New York: Academic
Press.

 

Steere, Edward (1870) Swahili Tales.

Sterner, Suharno, and Pike (1976) "Experimental Syntax Applied to Relations
Between Sentence and Sentence Cluster" in Suharno and Pike (eds.)
From Baudi 19_Indonesian. Jayapura: Cenderawasih University.

Waltz, Carolyn H. (1977) "Some Observations on Guanano Dialogue" in R.
Longacre and F. Woods (eds.) Discourse Grammar: Studies lg
Indigenous Languages gf_Columbia, Panama, ggg Ecuador. SIL Publications
in Linguistics 52.

Whiteley, Wilfred (1969) Swahili: 1ߤ_Rise gf_g_National Language.
London: Methuen.