‘ ._—— .,,_____—_—___ v —.—_. 7
. . ........-.n-‘--n....-.u.a....¢n .-...~~o<-.~‘~ _,._‘ --~ ......_........ -<——~ ~~ --.-~-~---«--~>----—-A »* '~"--‘-'- -’-"' q“
0 "

 

 

 

 

mi; Illlllllllllllllllllllllllllllllllllllllllllll L

3 1293 01055

This is to certify that the

thesis entitled

A STUDY OF THE DISCOURSE MARKER
NANKA IN JAPANESE

presented by

Makiko Séito

has been accepted towards fulfillment
of the requirements for

Ma.s_t_e_r.h__degree in Linguistics

. ,- I ’1 ‘
Y’k-vL-1v/w é-erC. 4,4 [4—m67—I\. "I

Major professor

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

 

PLACE IN RETURN BOX to romovo this chookout from your record.
TO AVOID FINES Mum on or baton on. due.

———————_—_——————l
DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l lT—lf

usu lo An Affirmative AotloNEqual Opponunlty Institution
cha-M

_ _ 7 _———

 

A STUDY OF THE DISCOURSE MARKER MANKA IN JAPANESE

BY

Makiko Saito

A THESIS

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

MASTER OF ARTS

Department of Linguistics and German,
Slavic, African, and Asian Languages

1992

ABSTRACT

ASHDYOF'IHEDISCDURSEMARKERMMGINJAPANESE

By

Makiko Saito

‘Ihe function of Japanese discourse markers, especially nanka 'liJce,
scmehow' was studied in informal ccnversaticn with four male and four
fetale native speakers of Japanese. This study was mostly based on
Schiffrin's study (1987) in which she analyzed eleven disccnrse narkers
in English. Based cn my chta, five words were ocnsidered as Japanese
disocnrse narkers in my study. ‘Ihey are nanla 'sanehcw', chlara 'so',
darn 'but', datte 'but, because', andsozede 'and, thm'. Each of than
hasits literalmeaningaswellasmeothern'eaningthat is closely
related to its discourse ocntext. mag these discourse markers, mania
ocmrrednorethansixtypercertmtofalltheoomrrmcesofdisomrse
narkers. 'Ihisstudysrmsthattheocwrrenceofthenarkernankais
sensitive to age, rank (i.e. power) differences, and the degree of
relaxation, but not gender. Moreover, producticn of nanka is closely
related to individual differences. The functions of name are l)

softener, 2) turn initiator, and 3) filler.

'Ibuyparents

iii

I mild like to express my Wt gratitude and debt first to
Professor Mutsuko Ends Simm, try chair thesis advisor. I greatly
respect her as a linguist, a teacher of Japanese, and a considerate
perscn. Since I started a master's program at Michigan State
University, she has been a wcnderful adviser who taught me both the
sternness and the pleasure of learning and teaching. Her ccntinums
academic interest strongly inspired me to understand nore about Japanese
language. It ishardtodescribeinmrdshcwfortimteIwastohave
met her. Without her guidance and patience, this work could not have
heal dcne better.

I also wand like to express my amreciaticn to my thesis
ccnmittee marbers, Professor Susan M. (ass and Professor Dennis R.
Preston. 'Ihey kindly spent nany precious hours giving me significant
Garments and discussing various points of this work.

My sincere thanks go to the seven subjects who participated in the
interviews and to my friaids in lingusitics who offered me valuable
advice and suggestions.

Finally, I would like to thank my pams for their endless

enoouragaient .

iv

 

TABLE<M?CIIHEEHS

LJSH'OF'TAELES . . . ...... . . . . . . . . . . . . . .

LISTOFF'IGEFSW. .............. viii

LISTOFWQ‘ISW .......

2.1. Studies inEnglish discoursenarkers . . . . . .
2.1.1. Studies inthe 1970's . . . ......
2.1.2. Studiesinthe1980's ...........
2.1.3. Schiffrin (1987) ......... . . . .

2.2. StudiesinJapanesedisooirsenarkers......

2.3. Definiticn of disooursercarkers inJapanese . . .

3.1. Procedure . . . . . . ..............

3.2. Participants ....... . . . . . . .

3.3. Fragmentaticnofdata..............

4.1. Japanesediscair‘senarkers ......
4.2.'Iheresultsofallgroups............
4.3.Faialegroup .......
4.4.Malegroup .............
4.5.Mixedgroup... ....... ......

5. DISGJSSICN AND GNCUJSICN .......

6. WCNSANDSWCNSMFUHEERESEARCH. .

APPEI‘DIX..................

53

60

63

rIable
Table
Table
Table

Table

Table

LIST OF'IAELIES

List of subjects .................. 16
Simrary of Japanese discourse markers ....... 31
Nurrber of nanka by three groups .......... 35

Smuaryofmmbersofclauses,nanla,and
percentageofnaniausedbyallsubjects . . . . . . 36

Occurrence of mania according to the positicn
of the clause ................... 37

Summary of characteristic features of nanka
by female, male, and mixed groups ......... 52

 

LISI‘OFFIGEES

viii

 

LISTOFABHiEVIATIO‘JS

final particle

negative

direct object marker

past tense

quasticn marker

subject marker

tag-questicn-like auxiliary verb forms

m1

Discourse nerkers play one of the inportant roles in oonversaticn,
sincetheynotmlysupportthe structureof theomversatim, butthey
most of the time also express critical infoniaticn of their own.
Discourse markers are "sequentially dqaendent elements which bracket
units of talk" (Schiffrin, 1987: 31) . They are strongly oamected with
their discourse content and, therefore, they camot be explained only by
the syntactic characteristics of smtences.

meof thenostrecentiaiorksmdisowrsemarkersis Schiffn’n
(1987), in which she discusses several English discourse markers--ah,
well, now, then, y'knav, Imean, so, because, and, but, and or. In her
study, she illustrates anrng other things whether these markers have
meanings, whethertheyhave functicns, andwhat theyaddto coherence.

In this thesis, I analyze discourse riarkers in Japanese, focusing
partimlarly m the discourse marker nanka '1ike', as it occurs
pram'nertly in Japanese discourse. I also investigate whether disocnrse
markers inJapanese fmicticnthesanewayasthoseinfinglishdoand, if
theydonot, thereascns. meofthereasmswhyl chosethis tcpic is
that there has bee: little research in this field. Another reasm stalls
franmy pedagogical interest. As Stubbs (1984: 203) points out, a nodel

oflanguageshaildfomsmlanguageinuseinoamecteddisomrse.

2
Discourse markers are important, as well as syntax or phanology, for

secmd language learners to naster in order to achieve native- like
carpetance. .

Schiffr-in (1987) does not discuss gender differences in the use of
discourse narkers. In Japanese, however, gender is one of the inportant
factorsinanalyzingdisoanrse, asit isnarkedeveninashortpnrase
or in choice of words (Kindaichi, 1957). Therefore, the use of
discourse markers might also show some differences between men's and
mnen's speech. libreover, since indirectness is greatly favored in
Japanese (Kindaichi, 1975; Itasaka, 1978), theremight be sore discourse
markers which indicate indirectness and vagueness.

In this study, I take both a quamzitative and qualitative
approach. The quantitative approach is objective,- since it deals with
frequency, itcansrmthetendencyofthespeakers' preferencesin
words. With this amroach me can also generalize certain prqnerties of
discourse narkers such as their position in sentences and types. In
contrast, the qualitative approach is more subjective in that the
interpretation of the observer plays an inportant role. With this
approachanecanexaminetheoantent of eachdisooursenarker. Because
of the two characteristics of these arproaches, I feel that adqating
anly ane of then is insufficient. Therefore, both approaches are taken
in this study.

Mystudy isuostlybasedantheseninalworkandisowrsenarkers,
Schiffrin (1987) , which clearly presents English discourse markers. I
will atteipt to demonstrate: 1) what types of discourse markers are in
Japanese, ii) how they functian and whether they contribute to making

canversatianodnerent, arxiiii)whetherthereareanydifferencesinthe

3
useofdisoonrsenarkersinmen'sandwonen's speech, andif thereare,

how they are different.
Based on the points mentioned above, the following hypotheses are
considered:

A. Japanesedisoonrseuarkershavealnostthesamefnmctionsas
those of English discourse markers.

B. 'Iherearesonedisoonrsenarkers inJapanesewhich indicate
indirectnessandvagueness.

C. Therearedifferencesintheuseofdisconrsenarkersbetween

menandwonen.

The fornat of the present thesis is as follows. In chapter 2, I
reviewsoneof theprevious studiesondisooursemarkers. Inchapter3,
Idisonssthenethodandprocednresackptedinrrystudy. Chapter4
ontlinestheresults of thestudy. Inchapters, Ipresentriyanalysis
of these results, along with my conclusion. Chapter 6 discusses the
limitations of the present study and sons suggestions for future

research.

M2
REVIEOFPREVIGISM

2.1. Bunio- 1n Enlist: discourse miners

In the literature, adverbs, conjunctions, and interjections seen
to constitute the bulk of discourse markers. Although there have been
nuinerous analysee of the syntactic and serantic characteristics of these
categories, Qnirk (1955) was the first to conduct discoursal studies on
then. He drew attention to the inportance of "reonrrent nodifiers" such
as you know, you see, and well in colloquial English. Stubbs (1983)
alsopointscuttlet studies of these categories indisccurse sequences
are of great interest to sociolinguists, since their functions greatly
varydependingonthepositionofagivendisconrse. Beloerevier
sons of the najor works on the elenents that can be regarded as

discourse markers.

2.1.1. Studios in the 1970's

AfterQnirk (1955) therevasnotirudnstudyondisconrsenerkers
in the 1950's or 1960's. Hmrever, there ameared an increasing mmber
of studies in the 1970's.

Schegloff and Sacks (1973: 303-4) dnserves that Wis-ell, O.K, and
90-00 are named as a kind of pre-clcsing device, or more precisely,
“possible pro-closing". mat is, these words are often used to close a

4

5
discourse on a given topic. Not only that, they can also function to

openadisccurseonanewtqnic.

Sacks, Schegloff, and Jefferson (1974) states that in taking one's
turn, especiallywhen it is overlapped, there are sore comm turn-entry
wordswhichbeginaturn. 'Iheyare, for example, well, but, and, andso
. Accordingto then, anoverlapwiththeprevionsturnwillnothannthe
constructional development which trese turn-entry words begin.

mlliday and Hasan (1976: 269) discusses sons conjunctive iters as
follows. When now is stressed, for ecanple, it is deictic and not
cohesive. If it is not stressed it opens a new stage in the
comunication, such as a new incident in the story, a new point in the
argument, or a new attitude of the speaker. Also, well typically
oconre at the beginning of a response in dialogue. "Response" differs
from an "answer" in that "response" is to make a return by sons action,
whereas "answer" is to coiply with a question based on one's lonowlerhe.
When the addressee emphasizes well, not only does it indicate an
adonouledgenent of the question, but an answer can be expected to
follow. In contrast, when well is reduced, it indicates that what
follows is a response to what has preceded, and well is sonstimes
corbined with the respondent's claim showing a reluctance.

Labov and Fanshel (1977: 182) agrees with Hallichy and Hasan's
analysisbysayingthatwellecpressesthecomectionbetweenwhat
precedes it and what follows it. In addition to that function, they add
that well is also used to shift topics.

chmanystudiesweredoneondiscoirsem'arkersotherthanwell
in the 70's eocqot for Von Glaserfeld (1974) and Polanyi (1979). The
formeranalysedthewordbecause fmmaphilosqnhicalandlogical view,

6
and concluded that because has something to do with causation. He

mtegorizes because into four gI'OUpS: 'matern’al' muses, 'formal'
muses, 'efficient' muses, and 'final' muses, according to its musal
relations. This study, however, is not fully based on natural
conversation. Polanyi (1979) explains that with you know people
structure the stories they tell. She states that the marker you know
givestheaudienceastrogimpressionthatwhatthespeakeris saying
is the truth, and therefore that speaker's implimtion must be
understood.

2.1.2. Studios in the 1980's

In the 1980's, the study of discourse markers developed more fully
than in the 1970's.

Wootton (1981) Shows that well can mark "non-grantings" in a
conversation between parents and children. According to him, well is
used before "rnon-grantirgs", but not before grantings. In child to
parentrequest sequences, wellarpearsintheparent'sresponseto
indimte that the child's request will not be granted. Wootton argues
that this initial non-granting well is used to minimize the rejection
soietimes arnd to maximize the possibility of granting at other times. I
speculate that it is also true in adnlt-adult conversation.

Owen (1983: 44) suggests that well might be interpreted as a
hesitation marker, that is, a speaker uses well when he/she mmnot cote
up with expected responses to the offered assessments, questions,
requests, andoffers. Healsoaddsthatwellmnprecedearesponse
whenthepresuppositionofaquestion itselfneedstobemnceled. In

7
other words, well expresses the speaker's difficulty in not being able

to answer.

As was pointed ont in section 2.1., Stubbs (1983) eiplnasizes the
importance of analyzing discourse markers in conversation. Since the
markers make no syntactic predictions, syntax has little to say abont
them. Neither does seventies have much to say abont the markers,
bemusetheyarenotusedintheirliteral sense. Stulzbs (1983: 68)
pointsoutthatoneofthemainfactsabout suchnarkers is thatthey
are essentially interactive,- they relate an utterance to another. hell
used in utterance-initial position in spoken language is especially very
diffionlt to translate into other languages, because it has no
propositional content, and the meanings given in dictionaries are based
only on written data.

Beanan (1984: 60-61) presents an analysis of the coordinating
conjunctions: and, but, and or in the spoken and written languages. The
resultofhisecperimentvasthattheactual frequencyof coordinating
conjunctions suchasand, but, so, or, andandso, wastwiceashighin
the spoken language as that in the written language. In addition, his
study shows that and in the spoken data mostly functions as a "filler
word"; i.e., it is used to avoid awloiard silence.

2.1.3. Sdniffrin (1987)

Schiffrin (1987) treats eleven English disconnse markers, namely,
oh, well, and, but, or, so, bemuse, now, then, y'know, and Imean. She
describes the signifimnnce of analyzing discourse markers as follows:
"an analysis of disconrse markers is an analysis of disconrse coherence"
(1987: 40). Inotherwords, eventhe sanedisconrsemarkerhas several

8
different meanings denending on the situation,- therefore, it is

importanttlatthespeakerandthehearerunderstandthefomsarxi
neanirgs of what is said in that particular situation.

Whereheranalysis standsontm'ost fronallothers isherdetailed
annd close examination of the functions of each marker. She adopted the
fonnnat of group sociolinguistic interviews in which she herself acted as
the interviewer and also shifted her role to a participant. The
sociolinguistic interviews revealed information on the participants '
varions situations annd speech style, which contributed to the aconracy
of her analysis.

In her analysis, Schiffrin (1987) used examples drawn from
narratives and arguments only, bemuse telling stories annd arguing were
two speech activities frequently egaged in by the participants.
Regarding well, in addition to the funnctions fonnnd inn earlier studies,
Schiffrin adds another tnat it narks the "participation framework" of
discourse (P.102) . The term "participation framework" is originally
introdnced in Goffnran (1981). It is said to mpture speaker/bearer
relations and speaker/utterance relations. Schiffrin also states tlat
"well anchors the speaker innto a conversation precisely at those points
where upcoming coherence is not guaranteed" (p.126) . ‘Ihis statenent is
generally supported by her data. For example, well is used mudn less
before answers to yes-no questions tIan before answers to win-questions.
‘Ihis is because, according to her, yes-no questions strictly limit
respondent's options to make answering easier, whereas wh-questions do
not provide such clear- ont mtions.

Sdniffrin (p. 328) sumarizes suggested conditions for discourse
narkers as follows:

9
(1) 1. synntactimlly detachable front a sentence

2. Cmmrnly used in initial position of an utterance
3. Have a range of promic contours
e.g. tonic stress annd followed by a pause, phonologiml
reduction
4. onerateatbothlomlandglobal levelsanndseveralplanesof
discourse, i.e., it must be vague or have no meaning at all
These suggested conditions of English discourse markers are useful
indeterminingwhatcanberegardedas discoursemarkers inJapanese,
which I will discuss later in this thesis.

2.2. sunflo- in awe dim mariner-l

rIherehasbeenalmrnstnnoresearchfocusingonJapanesedisconrse
narkers. There are, however, many studies on particles: for earple,
Uyeno, 1971; Maynard, 1989; Codi, 1990; and Ishikawa, 1988, 1992.
Japaneseis richinparticlesandsoneof thencontn‘bute tomakingthe
conversation coherent. While disconrse narkers are independent words
(sore linguists mll then "fillers") , particles are usually affixed to
noun phrases annd sentences. However, some linguists (e.g. Uyeno, 1971)
mtegorize certain interjections urnder "particl “ as will be shown
directly below.

Uyeno (1971) presented the first conprehensive work on sentence
particles, in which she proposed two groups of particles in Japanese.
The first group of particles occurs at the initial position of a
sentence. 'Iheyare a 'oh', aa air, are 'meI', saa 'well', e 'huh?', so
'yes', anndmaa 'myi'. For instance, in the stateients such as, a (cbo

shi-yoo) 'Oh, my goodness (what shall I do?)', and saa (tabe-yoo) 'well

IO
(let's eat!)', sentences in the parentheses maybe omitted bemuse the

particles at the beginning of the sentences convey sufficiem:
information in the context: according to her, a inndimtes 'serionsnness'
annd saa indimtes 'let's do...'. The second gronp of particles mlled
"sentence particles" are we, 20, 2e, sa, yo, ne, and ma, which are used
at the end of a sentence. Uyeno states that "these reflect the
speaker's attitude in conveying a piece of information to the
addressee.” (p. 5) Weno's analysis, however, is based on the (hta of
constructedindenendentsentencesandshortdisconrses, andshedaesnot
ecplainwhysentenceparticlesaremainlyusedinthe spokenlanguage.

Maynard (1989) also mtegorizes Japannese particles into two gronps
as in Uyeno (1971). One gronp consists of the particles which are
directly related to expressing a speaker's jungment when uttering a
sentence, annd the other gronp consists of the particles whose functions
are primarily grammatiml. She further mtegorizes the particles of the
first gronp inn-.0 two subtypes,- particles trat oconr in utterance initial
position and those in final positions. Utterance-initial particles are
the same as those of Uyeno (1971) annd the utterance-final particles are
the sentence-final particles which Uyeno calls 'sentence particles'.
These particles are generally attached to sentences as in, kyoo wa atsui
cbsu ne 'It's hot today, isn't it?‘ and bore itadalon m 'I will take
this'. BothWenoandMaynardsuggest that frequent insertion of such
particles in Japannese intensifies speaker's "level of involvenent" in
conversation.

Chikamatsu (1979: cited in Shibamrnto 1985: 61) suggests tIat there
areparticles whichareusedeconsivelybyonegenderandalso those
used by both. She mtegorizes sentence-final particles into three

ll
gronpsaccording totheusagebygender: (l)wa, nno, teusedbywoien,

(2) ze, yo, nousedbymen, annd (3) yo andnneusedbybothmenand
women.

Amog these particles me is used most frequently in discourse.
According to Uyeno (1971), Mizutani (1985), annd Kindaichi (1975), me is
usedwhenthespeakeragrees withthehearer's asemption orwhenthe
speakerexpectsthehearertoagree. Inotherwords, neexpresses
enpathy. Cook (1990: 35) addsanothermeaningof neas soiethinglike
'yon know' in English; it establishes a cooperative relationship between
conversation participants. It shonld be nnoted trat, while the English
discoursemarker 'yonknnow' isanindependentphrase, it corresponds to
the particle ne in Japanese.

Ishikavia (1988, 1992) analyzes the sentence particle as in
disconrse annd suggests that there are three fuctions of the particle ne.
First, it enacts shared information which belogs to the hearer.
Second, it elicits an acknowledgment or agreenent. Third, it
contennalize a coqnerative footing in the " participation framework".
AccordingtoIshikawa, nnemnservethesethreefmnctionsatthesame
time.

Sawyer (1992) analyzes the aonuisition of ne by L2 learners of
Japanese. He finds tlat learnners tend to acquire ne first within a
chunk like soc desu ne 'Ttat's right' since it is menorizable annd
useful. Susan Gass (persoal coimnnimtion) poinnts out that this is
actually a frequent phenomenon of early language acquisition, i.e.
things are learned first as a chunk.

In addition to sentence-final particles, Maynard (1989) describes

"fillers" in Japanese discourse as follows:

l 2
The term "filler" refers to a broad range of utterances (inncluding

independently appearing particles) trat do not mrry identifiable
or relevam: propositional meanings. Sore fillers are phrases or
brief comments, as exeiplifiedby name 'well'; sore are mere
aipcy utterables such as warm "uhh (1989: 30)

Fillers ensure that "the channrnel of coimmnimtion retains men" annd also
"createanatmosphereinwhichthespeakersrnws sonehesitancyandless
certainty abont his/her message content". Maynard, 1989: 3 0)
Therefore, like sentence particles, fillers also contribute to making
the conversation coherent. Two types of fillers are introdnced: (l)
language-prediction-based fillers annd (2) socially motivated fillers.
According to Daynard, language-prodnction-based fillers are used when
smooth speech is interrupted, these fillers are usually marked with what
Maynard mlls “meta-communimtioal conments" such as, uuunnto, hora,
are, 'uhh...yon knnow, that' annd socially motivated fillers, which are
used more frequently, occur to fill a potential silence to avoid
potential enbarraseiernt. In addition, Ohta (1991) states that hedges
such as, name 'kind of/sonething', ano 'uh', and ma 'well' are quite
comm as markers of hesitation and they are used to reduce speaker
remonsibility for that utterance.

It is clear that the definition of fillers given above overlaps
with Schiffrn'n's definition of disconrse markers in part. 01 the other
land, Schiffrin's suggested conditions for discourse markers might
eoclude Japanese sentence-final particles, bemuse dovionsly they do not
occur in sentence-initial position. In the next section I will discuss

and define discourse markers in Japanese.

)3
2.3. Detinitionotdmmmw

As stated in chapter 1, Schiffrin (1987) defines discourse markers
as "sequentially denendent elements which bracket units of talk" . What
this means is that synntactic, prosodic, annd innformation units of talk
arebomndeddenendingonadisconrsecontent. Therefore, wemnnnotjust
takeontadisconrsemarkerfromasentencetocbserveitsmeaning; it
has to be doserved in relation to the discourse content.

mynard (1989) uses a term filler, annd it has two functions as
mentioned above. I speculate that what Maynard calls "language-
prodnction-based fillers" are mere eipty utterances, which have no
particular meaning besides filling a pause to avoid awkwardness in
conversation. On the other hand, what Naynard mlls socially-motivated
fillers create an atmosphere in which the speaker shows some hesitancy
annd less certainnty abont his/her message content. We cannot predict
this from the literal sense of the word; it is closely connected to the
discourse contect. We can see that Schiffrin's definition of discourse
markers and Maynard's definition of fillers overlap in part. Daynard's
definition of fillers can be divided into two types. That is, socially
motivated fillers conld be discourse markers in Japanese, which is also
consistent with Schiffrn‘n's view, bemuse both linguists suggest that an
eleventhasadifferentmeaningdependingonthecontect. Inthis
study, IregardJapanesedisconrsemarkersastmsewordstlatIave
separatemeaningsthantl'nseecpressedinadisconrse content. Tnat is,
Japanese discourse markers are indenendent words which have literal
meaningsaswellasothermeanirgstratarecloselyrelatedtotheir

discourse content .

 

3.1. Procedure

The format I erploy in the present study is gronp interviews,
following Schiffrin (1987) . Here the ternn interview is used though it
may actually reserble r egular c onversation. I n s ociolinguistic
interviews, wnat isthemostimportantis tominimnizetheeffectof the
Cbserver's Paraonx (Labov, 1972), that is, howtoobservethewaypeople
talk when they are not being systenatimlly observed. In order to
minimize this paradox, varions innterviewing skills annd strategies were
used.

Therearethreegrows: onewithall fenalespeakers, onewithall
malespeakers, anxionewithtwomaleandtwofeiale speakers. Oneof
the fonr fenale participants is myself. In the all fenale annd minced
grows, Iactedastheinterviewerandalsoasaparticipant. Inthe
allmalegrow, Iaskedonespeakertoactastheinntervieweraswellas
aparticipannt. Thethreegrowswillbecorparedintermsofthetypes
andusageofthedisconrsemarkers. Moreover, oneofthepurposesof
formingtlnreedifferentgrowsis tocbservehowgenderaffectsthe
occurrence of discourse markers. Within each grow, one individual is
designated as an "interviewer" wlno more or less leads the conversation

asnecessary. However, theroleofaninnterviewerwaskepttoaminimnm

i4

IS
inordertocreateanatural settirg. 'I'hereasoiwhyIchcsesucha

format is that, since a owe-to-oie interview toids to make participants
nervous, authentic datamaynotbeobtainedby it. In coitrast, during
groip interviews, people can address and soretimes reinforce each other
atanytimeandtheyfeelmuchless pressurethanwheitheyareformally
interviewed. mreover, since the participants are acquainted with each
other, theyaremorerelaxed. Ibelieve that thevernaoflarprovidesa
furiameital baseline of larguage analysis.

'Iheinterviews, eachofwhichlastedoiehour, tookplaceinone
of the participants' apartment and they were recorded on audio tape.
Each group interview was held separately. The same gender groxp
imerviews (fetale/male) were performed before mixed groxp interview.
Durirg the interview, the participants were basically free to talk about
anythirg they chose. However, to make the corversatioi flow, the
"interviewer" first asked other participants in each gron Amerila ni
kiteichibanocbroitakctowanandxu ka? (What surprisedyoithemnst
when yo1 came to the United States first?) and Amerika ni kite kara
ichiban taihezdatta}mtomnandesu1a?(mtvasthehardestthixg(s)
you experienced since you came to the U.S.? ) Also, the interviewer
added such topics as 'students at Michigan State University' , 'Japan
bashirg', 'future career', and so o1. After the interview, participants
were asked to fill out an 'informatim sheet' which includes sex, age,
studeit status, hometown in Japan, years of residezce in the U.S., and
major. (See Appendix) This informatio: will help us to see the

participants' different styles of ooiversatioi.

 

16
3.2. Participants

Foirmeiandfoirwmeiparticipatedin this study. 'Iheyare all
native speakers of Japanese and are omen-.ly studeits at Michigan State
University. 'Iheir ages rarge fron 21-31 years old. The participants
allhavelcmeachotherforapprocimatelytwoyears, andtherefore,
they usually talk in colloquial (informal) style, not in polite style.

Table 1. List of subjects

 

Subject Sex Age Student Honetown Years of. residence Major
status inJapan intheU.S.

 

 

A F 27 G Osaka 4 years Sports psych.
B F 29 G chohama 1.8 years Educatioial psych.
C F 28 LE Kanagawa 4 .7 years Lirguistics
D F 29 G 'Ibkyo 3.8 years Lirguistics
E M 21 u; Ibaraki 4.5 years Physical edicatioi
F M 27 G Fukuoka 3 years Agricultural eooi.
G M 31 G Saitama 3 years Physics
H M 22 In 'Ibkyo 3 years mblic administratioi
Wm
C=graduate

3.3. mum of data

Japanese selteices are usually broken into a nurber of smaller
units in infomal corversatioi. Chafe (1980:14) suggests that "clause-
final risirg or falling pitch is the sitgle most ooisistent signal" of
what he calls an "idea unit" and each unit usually carries one

informatim piece. The "Lmit of talk" in Schiffrin's (1987) definitio1

l 7
of discourse markers is similar to Chafe's "idea unit". This notioi of
"unit" is inportant inanalyzirgthedata, becauseaserteice isnotthe
most appropriate unit to understand language use and social interaction.
Discourse markers occur quite freely in sentelces and are difficult to
define syntactically. Also, sexteices are oftei enzrenely difficult to
idertify in informal speech. I therefore adopt this "idea unit" for the
analysis ofmydataandregardanideaunitas aclause forthe sakeof

corveiieice.

m4

Inthis section, Iwillproceedinthefcllowirg manner. First, I
presetJapanesedisconsemarkersappearirginmydataalogwiththeir
frequency. Second, I list each marker in its literal sense and also as
adiscoirsemarker. 'Ihird,Ifooiso1themarkernanJe,asthisisthe
most frequeitly used marker. I will detmstrate its occurrence as a
discoursemarker in detail, used in each of the three grons: female,
male,andmixedgro1ps.

4.1. manor-em

Basedoimydata, I coisider the following fivewords as discozrse
markers in Japanese: nanka 'sonehow', dakara 'so', dam 'but', datte
'but, me, and sorede 'and, thea‘. 'Iheperceltages of occurreice
cfthesemarkersclassifiedbygroipsaresl'nminFigurel.

AsisclearinFigurel., nankaccoipiesmorethanahalfcfthe
occurreice of all the discourse markers (66%) and datte follows it
(19%). Noticetratthediffereiceintheperceltagebetweelthetwois
large.

Below I list each of the Japanese discoirse markers who:

expressing their original (literal) meanings as well as functioning as

18

19

Figurel. Perceitagesofccwrreiceofthediscoirsemarkers

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

inJapanese
66
6O '
50
as
% 30
25
23 M
20 F 18 19
Mix Total
10 8 8
M Mix 6 .
3 2 2 2 L
IF FFJ M War
126 137 98 361 16 43 44 103 13 9 13 25
nanla dakara den)
60
50 Heralegroip
M=malegro1p
2:5 Mioemixedgrozp
% 3O
20
10
6
F a Max M 1m“ ML
14 4 14 32 2 5 7 l4

(htte screde

20
discoirse markers. I will disoiss the most frequeltly used word nanka

more in detail in the sectiois followirg it.
Wisamrdderivedfronnanflcaviaeiplxnyacccrdirgtothe

Japanese dictioary called Koojiei. Nanilemeans 'sone, any, sorethixg,

andanythirg'. 'Iheoriginaluseofnankaisshowninexample (2).

(2) F: m toraburu toka nai? scoiu no.
any troible such have-NE; thatkind

'Doi't yoi have any trozble of that kind?‘

Nankain (2) is usedas 'any' tomodify toraburu 'troible'. Inthenext
exanple (3), nanka is also used to modify the followirg word warui koto
'wrogthing'.

(3) D: Atashi,nanh warui kctc shita kana to otnttamm.
I sone(thirg) wrog thirg did maderthatt‘rnght

'IthoghtthatIdidsorethirgwrog.‘

In my data, nanka is more oftei used as a softeier to avoid directness
of the speech. In (4), for example, the speaker talks aboit people's
large payments to stores resultirg from the frequent use of their credit

cards.

(4) G:Maruinomaitsukisaharai gasa,nanh gotan
Nar'ui's everynmth payment S sotethirglike 50,000
m rckman mo natte sa.
evei 60,000 eve: become FP

'mithly paymeat to Marui will be even sotething like 50,000
or 60,000 ye1.‘

21
bbtlcnwirgtheexactamommcfthepaymert,nanlaisusedtoecpressit

vaguely. Ifthespeakerradnotusednamaintheabcveecample (4), it
woild somd as though he were certain aboit the amomt (i.e. either
50,000 ye: or 60,000 yei). Similarly in (5), speaker F gives his

(5) A: m biiru to piza tte awanai ki ga oJru.
sorehow beer and pizza match-NEG feelirg S (b

'SorehcwI dol't thinkbeerandpizzagotogether.’

'IhespeakerFusednankatomakehis statenert lessassertive.
Therefore, not oilydoes nanka express its original meanirg 'sotething;
anythirg' but itfurtherseerstcfmictioiasabuffer, sotcspeak, to
show less certainty or directness. I will return to this marker in
sectiois 4.3, 4.4, and 4.5.

'Ihe secoid frequeitly used word is dam 'so; therefore'. The
origiral functioiof thismrdistocoljoinrasoiandresult. (6) and
(7) illustrate the use of dakara in its typical conjunctive function B
ecplainsherdisadvantageinhercareercausedbystayirgintheU.S.

fortwoyears.

(6) B:Hokano daisctsuno ho ycri ninei dairectau,dnkara
other 1miv.grad.iateperso1t1an2yearsbehind so

meritto janakute dereritto.
merit be-NEB-and deterit

'Iwillbezyearsbe‘nindtlanctherrmiversitygraoiates, so
it'snctameritbitadamerit.‘

22
SpeakerBstatesherreasoitratshendllbetmyearsbehindbefcrethe

vnrdchlaraandtheithereoilt "it'sademeri ". Inexample (7) below,
Gecplainsthatthereisanindividual telqzhole foreachpersoiina
hospital. While in (6) dam coijoins the same speaker's clauses, G
says dakara followirg another weaker's utterance.

(7) D: Hitcri hitotoi zutsu?
a perso1 oie each

'019 (telephoie) each persm?‘

G:Un, amasdmikakerebachdqmetsutsmagaruno.
Yeah,so therecall ifdirectly comect E'P

'Yeah, so if you call there, it will be directly comracted.‘

Speaker D's utterance is a reasoi and a result follows after dakara in
speaker G's utterance. In both examples (6) and (7), the word daJara
coijoins reaso1 and result, which is its original functioi. In
cortrast, the followirg example (8) arms the use of dalera as a
disconse marker. Speaker H eithusiastically tells what surprised him

mmhefirstcanetotheUs.

(8)H:Kitatd<inikitatdcini,dnhraanflnoorurui ni
camewheiatcanewheiat aalcoholkindm

seigei ga atta.
limitatioi S there was

'WhenI came, wheaIcame, therewasalimitatiol (of age) o1
alcdmols.‘

23
In speaker H's utterance, there is no reasoi-result relationship.

Neither was there an explicit statenert trat can be interpreted as
equressirgreasoiintheutterancepriortoit. migrainthiscase
seems sinply to erp‘rasize his statenert ("there was a limitatim of age
or alcohols"). In the next example (9), there is also no reasol-result
relationship; speaker G begins his utterance with dakara right after F's

questioi.

(9) F:Soredejochco oetedoosurunc?
thei informatioiOgethow do Q

"men, what do you do after gettirg the infomatim?‘

Gznahra coipusaabu tteic wa iroma saabisu gaaruwake.
corpu.servicecalled S variois service S thereare

"Iherearevariois (kinds of) service incmputerservice.‘

Again, in (9), dakara seems oily to erphasize the followirg staterezt.
Daleminthiscasecculdmeanttat “wtatIvanttosayis...” or"1et
meexplainthis...“ Inmydata,themalespeakersusedda1aramcretlan
the ferale speakers (see Figure 1.). This might be a result of this
characteristiccf dakara. Ancverusecf dakaraseersveryargmeutative
because expressirg oie's cpinio1 elphatically usirg dalara gives the
impressio: that the speaker is very aggressive. I speculate that women
trytoavcidtocmud1useofdakarasinceinasocietylikeJapan,worei
aresuppcsedtobelessagressivethanmei.

As a coijunctioi, demo 'hlt' indicates an omositioi to the
previois statement. In example (10), speaker G cmoses speaker F's
minim usingdematthebeginnirgofhis utterance.

24
(10) F: Nihonjin dattara ikoo tte iu kedo ne.
Japanese if go say flngh F9
'1: theyareJapanese,theywo11dgo, thongh.‘
G: Dem amerikajin wa ikanai yon.
but American S go-NEG FP

'But Americans woildn't go.‘

Demisalsousedtocorbinetmclausessaidbythesamespeaker,as
shownin(11). H—describeswlathedrirncsmorecfsincehecametothe

U.S.

(11) H:Ki1u1tekini gyurmnyrm shikatoranai yo, duo kocchi kite
basically milk oily take-NEG FP but here since

kara pqnpuwaodcunoru.
sincepcpSmuchdrink

'BasicallyItakeonlymdlk, hitsinceIcamehereImimmuch
qu-'

Dam in its conjunctive functioi indicates an qnposition like in ample
(10) and (11) above. onthectherhand, as adiscoirsemarker, demis
usedtocortrastanupcotdrgmnitwithapriornmit, as Schiffrin (1987)
aggested regarding the English discoirse marker 1111:. Example (12)
slowsthatspeakeereginshisutterancewith darn. JustbeforeF's
utterance, they were saying that there is a mutual friend who makes

terrible noise when he eats.

(12)G:Zenzen ranka sazenze1 mawari kinishinai no.zuu zuu zuu
atall atallaromdmind-NEIG FP

F:

2 S
tte koo yatte nondete sugee oto tatete sa.
like this do drink-and terrible noise make FP

'He (hasn't minnd (the peqnle) aronndhim. Henakes a terrible
noisewhenhedn'nks (sonp) likeznnnzuuzuu.‘

M, 8m.

oh yeah
'Oh yeah?‘

:Soredenankasa,ochamottekoaitosa,ctamottekoi

and FPtea bring-NEG FPteabringcone

ya. (LALE-I)
FP

'And when (a waiter) doesn't bring tea, he shonts "bring me
tea!“

:Duo nanka anzuki-sannornnmrreitogakono ue ni

Mr. Suzuki's rcomate S this urper in

omdenriakedo,raamentabemtdcinimoocdoroitatte.
live but noodle eat When very surprised said

'Mr. Suzuki's roomate is living in the ugner floor, but he
said that hemas so surprisedwhenhis rcomate ate noodles.‘

SpeakerF'sstateremismthnosedtoanytIdng,bntcontrastswiththe
prior subject. The nact acample (13) incidates that, in its usage, dam
returnstocontrasttheoriginalsubjectinstadcfthepricrsubject.
Thespeakersaretalkingabontanesthesiaatthedemist's office.

(13)C:Nanikco mr'ikakerudakedeldhnwake?

what likethis sprinkle only by work
'It works only by sprinkling it?‘

B:Un nankanenanlehxianosakikaradetelmn no.

yah FP tubeoftipfronconeontFP

26
'Yeah, itcomesontfronthetipofatube.‘

D:Daotashikanisoochyone,masui gaitaindamon.
certainlytruebeFPanesthesiaShurt

'It's certainly true, sinnce it is anesthesia (injection) which
hurts.’

Firsttheytalkedabontananesthesia injection (main tmic), then
speakerBtalkedabontagastype anesthesia (sub-tmic), andfinally
speaker D goes back to the main tmic (anestmsia injection),
contrastingtheformertmicbegimigwith dam. Thedisconrsemarker
denomight retains the literalmeaning 'but'. However, it is attended
toadifferentlacicalunit. mendenoisusedasadisconrsemarker,
thespeaker's thonghtsarestill connected totheprior tmic (notthe
presentone).

Datte 'because, but' is usuallyusedto indicatea reasmandto
cgposesonethingatthesametime. Inaample(14),Esaysthatspeed
skatirgshoesarenotforrentanrihegivesareasmaftertheword

datte.

(14) E: Aaiu rinku dewa supiido sukeeto Iasanai yo, datte
thatkindrink at speed skating rent-NESFPbecause

kutsu yorikonna nagaindamon.
shoes than thismuch log

'Atthatkindofrink, speedskatingshoesarenctforrent,
because (theedge is) thismuchlogerthantheshoes.‘

ample (15) showsthat datte isusedtoqposeC's minim. SpeakersC
andGaretalkingabontalocalareaiantyo.

27
(15)C:Arewaeki nogyakuldcavadelaishanoarurno
that S statim of other side on conpany S be side

janai jan.
Int

'That ismthectherside of’the stationandnotmtheside
wheretheconpanyis.‘

G:Sco? dattemise wamukoonohoo niooi jan.
so but storesSotherofsidemmany

'Isthatright? Buttherearemanystoresmtheotherside.‘

Dattein(15),inaway,hasalmostthesameusageasthatofdem,
sincetheybothareusedtoqposeaprevionsstaterent. However-,ina
partionlar disconrse content, datte is used to assert or express
speaker's claim or minim. Example (16) is a cmversatim between
speakersHandE'sabontpizzainJapan. SpeakerEusesdattetoassert
Mathelonwscverspeakern'sstatenert.

(16) H: Nisen ka sanzenen surun desho are.
2,000 or 3,000yen cost TAG that

'That costs 2,000 or 3,000 yen, right?’

E: Datto ima noniyasan de dasun desho, piza tte.
now bar at serve TAG pizza

'They serve pizza at the bar now, on't they?”

Speaker E's datte in (16) does not indicate a reason or cppositim to
H's statement,- it simply atpresses his observation. In other words, he
vents to enphasize or justify himself by adding more informatim. In
the nact ample (17), speakers C and D disonss Kamikaze (suicide)

28
corps. natte occur at the beginning of D's utterance to declare her

(17) G:Soo da yo. Are muchakucha ch yone.
rightbeFPtheycrazy beFP

'That's right. They are crazy.‘

D: Kamikaze desho? natte are we sugoi meeyo dato
Kamikaze 'mo that S very hmor be
onotten demon ne.
think he FP

'Kamikaze, right? They think that it is such an honor.‘

Speaker D adds more informatim (it is such an honor) to speaker C's
utterance following the marker datte.

Sorede 'and, then' is used to coordinate sentences annd also to
indicate a successim of time in its literal sense. For instance, in

(18) , sorede is used to carbine two features of speed skating shoes.

(18) E: Hontono yatsu wa kmnani nagain ch yo, accede sugoi
real things ikethis long beFP and very

toidearukaraabunainchyo
sharpenedsincedargeronsbeFP

'Arealthirg (emeoftheshoes) islikethislog, andsince
it issharpenedsomuch, it is dangerons.‘

The nact acample shows a successim of time. B is telling a story that
one of 8‘s friends was involved in a suspicions hypnotic seminar and

becamealeaderoftheseninar.

29
(19) B: Shidoosha ni made nacchatte nomads soko de kekkm aite
leader even became then there marriage partner

made mitsuke chatta no.
even find PAST FP

'(She) evenbecamealeader, then (she) evenfonndamarriage
partnerthere.’

Inample (19) tmevents, beconingaleaderandfindingamarriage
partner, happened at two different times. Sorede tells us the order of
these events. When sorede oconrs utterance-initially, it does not
acpress the literal meaning; rather, it inndicates a result. In (20),

for acample, a result is given after the sorede.

(20) G: Pmerikajin wa co gata gaooin dayo ichiban.
American S O typeSnanybe most

'BloodtypeOisthemostcomonmeamogkrericans.‘

F: A, honto?
oh really

'Oh, really?‘

D: Eebii mo ooinn da yone, igaito.
AB also many he FP unexpectedly

'Unacpectedly, there are also nanny type AB (pemle) .'

F: Boreas so, sociu seikaku shiteru nda.
th, thatkinnd character have FP

'That's why they have th, that kind of character.‘

Thenact ample also shows that sorede is used to indicatea result.
SpeakersFandGbegintheirutteranceswithsoredeafterspeakerHsays
ttatheuassurprisedbecansetlnerevasanagelithinsellirgalcdnol.

3O
(21) H: Nenree seegen ga atta koto kocchi kite kara odoroite,
age limit S be-PASI' thing here cone since enrprised

de kafe ni itta toki recto biaa tteno ga atta.
arndcafetowentwhenroot beercalledS be-PASI'

'I was surprised at the limitatim of agewhen I camehere, annd
wheanenttothecafe, therewasadrinkcalled"rcotbeer."'

F:8<nradaanoareyaneanoikkairymngaku o
wellthat wellmcestudyingabroado

jitai shiyoo to.
mcline try to do

'That's why well, that well, yon mce tried to declinne to go on
a study abroad program.‘

G:Sor'edobureikunotabi nikanadaniikutoflAIH-I)
break everyinCanadatogo

'That'swryyongotoCanarhduringeverybreak.‘

SoredeinbcthspeakerFandG, canbetranslatedas 'that'swhy', and
it inndicates results of what speaker H did. Table 2. is a surmary of
Japanesedisconrsemarkers in their literal meanings andtheirmeanings
as disconrse markers.

While the words discussed above are all disconrse markers, several
words can be regarded as fillers. They are: am 'well', sono 'well',
use 'well, ah', Iona 'well', iya 'well', andmoo 'well'. Bren though I
categorize then as fillers, there is still a fuzzy area between
disconrse markers annd fillers. I will present acamples of sore fillers
to show what fillers are like. Among the fillers, ano oconrs most
frequently (35%) and iya follows it (25%). They are so-called
'1anguage-productim-based fillers' (Maynard, 1989) and they usually
oconrwhenaspeakerstunblesorcamntconeupwithanintendedword

3 I
Table 2. Summary of Japanese disconrse markers

 

 

Literal Disconrse marker

name 'sone, any, . uncertainty, softener
sonething, anything' inndirectness

dakara 'so, therefore' erphasizer
conjoins reason and result

dam 'but' contrasts with the prior
oppositim subject

datte 'but, because' asserts or claims one's

cppcsitim, inndicates reason minim

sorede 'and, then' result
coordinates sentences

 

immediately or to gain time while grminng for a word. For ample, (22)

inndicates a typical use of am as a filler.

(22)F:Netetaraikinnari ano, daredakkeanoIndiananiitasa
sleeping suddenlywellwhobeQ wellIndianainwasFP

nantoka-san.

something NB.

'When I was sleminng, suddenly, well, who was it? well, Ms.
something, who was in Indiana.‘

Nattampleshowstheuseofsonoasafiller.

(23) F:Sorekara ato solo, minnna no kurashi no yccsu we?
then well everyone of living of situatim S

'Then, well, (how is) everyone's living situatim?‘

32
Sonoandanofnmctimalmostthesameasapause filler. Thewcrdhora

has two usages: one is an interjectim to get pemle's attentim and the
other is a mere filler. Ebmple (24) shows that hora is used to get

attentim.

(24) A: m, mite.
hey look

'Hey, look!‘

Horainmydata ismostlyusedtofillapausewhenaspeakerisgrming

for a word. In the nact acample, hora occurs with am in successim.

(25)D:Yoku minnna Mannidesahcraanosuutsu toka tsuknnru
often everybody Maruiat FPwell well suit like make

nda yone. gem: de.
be FP monthlypayment by

'Enrerybodycftenhaslike, well, well, asuitmadeatMaruim
amonthlypaymentplan.‘

bbo is used to state enphatically. This is shown in ample (26) .

(26) G:Ibo cfrnuugaldcoo gurai no toki kara, mo zenzen
Jr. high school abont of when since at all
suugaku dekinnaktta.
math. can-NEB-PASI‘
'Sinnce about Junior high school, I conldn't o3 matheratics at
all.’

33
The first moo conld mean 'already' and enphasizes chuugakkoo 'Jr. high

school' annd the second one enphasizes zenzen 'at all'.
Maa is also used to fill a pause. E><annple (27) inndicates that
speaker E seeks what to say.

(27) F: Inforeeshon konittii?

informatim conmittee
' Informatim committee? '

E:Un m dakarayoo wadakaranandeno shiritai
yeahwell points everythingknowwant
koto bokuniittemitaina.
thingme toask like

'Yeah, well the piont is, like “ask me everything that yon want
to know.“

Finnally, themrdiyahastwomeanings: meisaninterjectimmeaning
'no' annd the other is as a filler meaning 'well'. In ample (28) iya

is used to mean 'no'.

(28) F: Nande? yasui kara?
why cheap because

'Wlny? Because it's cheap?’

G: Iya chigau, gmpu dakara.
nno notright monthlypayment because

'No, it is nnot, because it is a monthly payment.‘

The nact acanrple indicates the use of iya as a filler. After a long
talk speaker F concludes the tmic starting with iya.

34
(29) F: Yooko-chan no koto mo aru kedo are hanasu to nagaku
Yooko ofstoryalsobebutthattalkiflog

naru Jena. (mm)
become sinnce

"There is also a story about Yooko, but if I tell that it's
going to be log.’

H:Tsugi ikocze tsugi!
nact let'sgo nact

'Nact, let's go on to the nact(tmic)!'

F:Iya moonanka iroiro arukedono.
well varioisbe but

'Well, there are varions stories. '

Speakerannadefuncf what speakeeridbeforeandspeakeeras
enbarrassed (he wants to charge the tmic). Then finnally speaker F
finishes the tmic about speaker H starting with iya. Interestingly,
there were many mere filler words in informal speech than I had
acpected.

AsIhaveshownabcve, thereareseveraldiscoirsemarkersand
fillers in Japanese, annd it is interesting to note that only one
disconrsemarker, nanka, accoints forcversixtypercent of the entire
ocoirrence of disconrse markers. In the following sectims, I will

thereforefooismthemarkernanle.

4.2. mmltsctallgrops
oneofthepirposesoffcrmtingthreedifferentgroipswasto

mservehowgerieraffectstheocoirrenceofdiscoirsemarkers. Table

3. srnwsthattherevasvirtuallynodifferenceintheocoirrenceof

35
nanlaamogthethreegrows. Ineachgroip,thetotalnmberthat

nanJaocoirredvasdividedbythetotalnnimberofclauses. Inthemixed
grow, thepercentage (7%) is slightly lcwerthanthoseoftheothertm
grows. Inthisgrow,nanlaoconrred96timeswithmalespeakersand
mly twicewith fenale speakers. Theresult of fenalespeakers (number
of nanka =2, number of clauses=379) in the mixed grow distorts the
resultofthemixedgrow.

Table3. Numberofnankabythreegrows

 

Fenalegrow Malegronp Mixedgroip

 

Number of
nanka 126/1107 (11%) 137/1096 (13%) 98/1271 (8%)

 

Without the result of ferale speakers (i.e. the result of male
speakers),themrnberoftinosnankaoconrredis96whilethemmberof
clauses is 892. Therefore, the percentage is 11%, which is almost the
sameresultasthoseoftheferaleandmalegrows. Again, accordinngtc
Table3., thedensityofthemarkernankashowsalmostnodifferewe
amogthethreegrows.

Table4srowsthatthereisalargeindividialdifferenceinthe
umber of claises. Subjects C, D (ferale) annd F, G (male) participated
intwodifferent interviews (ferale/male grow andmixedgmup). The
enallestmrnberofclauses (84) isfomdwiththemalespeakerEinthe
male grow interviewand the largest number of clauses (516) is found
withthemalespeakerFinthemalegrowimzerview.

36

Table4. Smmaryofnnmbersofclanises, nanka,
andperoentangeofnankausedbyall subjects

 

 

Subject Sac Number of Number of %
clause nanka
A F 276(f) 54(f) 0.20 (f)
F 197 (f) 25 f) 0.13(f)

C F 217 (f)+171(mx) 8(f)+0(mx) 0.04(f), 0(mx)
=388 =8 0.02

D F 417 (f)+208(mx) 39(f)+2(mx) 0.09(f), 0.01(mx)
=625 =41 0.07

E M 84 (m) l (m) 0.01 (m)

F M 516 (m) +461 (mx) 50 (m) +47 (mx) 0.10(m) , 0.10(mx)
=977 =97 0.10

G M 332 (m) +431 (mx) 72(m) +49 (mm) 0.22 (m) , 0.11 (mx)
=763 =121 0.16

H M 164 (m) 14 (m) 0 09 (m)

Fetale total 1486 128 0 09

Male total 1988 233 0.12

 

37
These nunbers indicate that speaker E did not participate in the

cmversatim so much and that speaker F spoke and dominated the
conversatim most of the time. The largest percentage of nanka (0.22)
isprodicedbythemalespeakerG, followedbythe fenale speakerA
(0.20). The smallest percentage of nanka is observed in the male
speakerE (0.01). The total of the tmpercentages of fenaleandmale
(0.09, 0.12 respectively) show that there is not a large gender
difference in the frequency of name,- however, there is a drannatic
differenceinthemixedgrow. Thenumberofnankauseddecreases
awectinthecaseofmemalesubjectF (wherethepercentageisthe
same). Thesemmnberssuggestthatthepresencecf theqwcsite sacmay
influencetheoconrrenceofnanka. Areasmforthismightbethe
speakers' hesitatim in speaking to the opposite sac and they coild nnot
speakasfreelyas inasingle sacgrow.

Nanka ocoirs not mly clause initially, but also medially and

finally. Table 5. shows the distributim of ccoirrence of nanka in

Table 5. Occurrence of name according to
the positim of the clause

 

Fenalegrow Malegrow Mixedgroip 'Ibtal

 

Total 126 137 98 361
Initially 50 (40%) 47 (34%) 31 (33%) 128 (35%)
Medially 70 (56%) 74 (54%) 55 (56%) 199 (55%)

Finally 6 (5%) 16 (16%) 12 (13%) 34 (10%)

 

38
three different grows. Name occurred most clause-medially (55%) arid

oconrred least in final positim (10%). The distributim of name is
almostthesameallthroighthethree grows. Clannse-finalnankain
the male and mixed grows (16%, 13% respectively) oconrred slightly
higher than that in the fenale grow. This is related to the fact that
thefrequencyof clause-initialnankainthemaleandmixedgrows is
slightly lower than that in the fenale grow. The above quantitative
results show that there is little differences according to gender in
general; however, both fenale and male speakers use nanka less
frequently when masking in the mixed grow. In the following sections,
I will show some qualitative results which do indicate gender
differencesandalsothesametendencies of nanka. Iamgoirgtoshcw
these results by the grows: fetale, male, andmixed grows.

4.3. Fonlogrow

Inthefenalegrowcmversatim, nanlewasmostlyusedtomakea
staterent vague or uncertain, and also to avoid an assertim or a strog
claim. Below are some amples which inndicate this.

Narnia often oconrred utterance-initially, it also oconrred with
malespeakers,:hmever, itseensthat fenalespeakerspreferredtouse
this initial male than male speakers. when nanle occurs initially, it
indicatesthatthespeakerisnotsureaboitwhatsheisgoingtosay.
In (30), for example, speaker B is telling the gronp that she is still
afraidof talkingtopmericansbecauseofherpoorfinglish, andthat
even after she goes back to Japan, she will still be uncomfortable with

foreigners.

39
(30) 8:“ ima no hoo ga hidoi kamoshirenai (LAIEH)
now S worse perhaps
'PehapsIamworsencw.‘

(D: Fuuun)

mmnmn
'Mrmm'
Nanhminna atashigashaberernnnjanaikato
everybodyI Sspeakcan that
onotterunjanaikana
thinnk wonder

'Iwonderif everybodywillthinktnat Icanspeak (English).'

Theseoconrrencesofnankain(30) canbeinterpretedas"Iamnotsure
bit..." or "I don't knnow well bit...". Moreover, in (31) speaker A
acplainsaboitanincidentsurromndinganewreligimwhichhappeneda

cowle of years ago in Japan.

(31) A: m sono musumetachi we jihatsutekini atsumattan
those girls S spontaneonsly gatlnered

mtte, kyooso no uchi ni
Ihear foinderofreligim of house to

'I hear that those girls gathered spontaneonsly at the fomder
(of the religim) '8 house.‘

Aankain (31) also indicatesthespeaker'suncertaintyaboitvmatsheis
goingtosay. Byusingnankaatthebegimningoftheutterance, the
speakersignals that "Iamnotsureaboitthisbut letme say....". In
away, aspeakerdefendshimv‘herself firstbyshowinghis/her lackof
unnconfidenceoruncertainty.

4O
Nanka soretimes occurs with other disconrse markers arnd fillers.

Emecially in the fenele grow cmversatim, nanka often oconrred with
thedisconrsemarkerchro. Itoccurred six times with fenaleandtwice
with male meakers. In ample (32), speaker D shares her acperience
that atarmtaurantpemlewere talking aboit Japanese after the Japan
bashing, annd she was uncomfortable in that situatim. Dem annd name

are used in successim.

(32) C: Mite hanashiteta kamoshirenai yo.
lod< talked perhaps FP

'(Those pemle) perhaps talked looking at yon.‘

D:Iya, sonnnakoto naikawakannaikedo,dmnanka
nno that thingNEG mm but

sociu hanashi shiteta

that kinnd talk did

'No, I dm't know that it isn't the case, but theywere talking
aboit that kind (of thing).'

Demisamarkertoacpressanqwosing minim. Thereforenanleafter
demo here suggests that the speaker used it to softens her omositim.
Alsoinample (33), nankaisusedwithdero. SpeakerD says that one
of her friends changed after participating a hypnotic seminar.

(33) C: Sore hontmi kiku nno kana.
that really effective wonder

'I wonder if that is really effective. '

D: Duo mks, "atashi akaruku natta mitai" toka itteta mm.
but I cheerful became seen that said

'But she said, "I have become cheerful.“

4)
Again, in (33) above, nanka following demo softens speaker D's

statement. Withoit name, speaker D's minnim can be simple opositim
against speaker C's assumptim that the hypnnotic seminar may not have
been effective, but nanka is inserted to avoid it. Therefore, nanka can
be interpreted as functiming as a buffer.

Repetitim of nanka often hamens with fenale speakers. It
oconrredserentimes with fenaleanndmlytwicewithmale speakers.
Shibamoto (1985) states that rmetitim of word is one of the
characteristics of woman's meech. For instannce, (34) illustrates B's
acplanatim aboit anesthesia at the dentist's office. Nanla is repeated

twice.

(34)B:Un, mhnenanlalondanosakileradetehnruno.
Year FP tube of tip from coneoit FP

'Yeah, it conesoit fromatipcfatube.‘

Sinmilarly, the nact ample shows the use of name twice in successim.
The speaker is also fenale.

(35) A: Sooiu ne shiriai nno okusan toka sa, koto gaatta

that acquaintances wife like FP thing S happened
kara sa, hmto nanh mule yoku wakara nai na
since FP really well understand NBS FP

toka onoin da kedo ne.

thinnk be FP
'Since the thing like that happened to acquaintance's wife, I
on't think that I really understand (articles in a magazine)
well.‘

42
It seems that these repetitims of nanka not only indicate more

mcertainmy,butalsothefactthatthemeakeris lookingforawordas
inthecasewiththefiller ano. Inadditim to nanka, otherdiscoirse

marker also were repeated. One of the acamples is with datte.

(36) D: natta datte haratta bun, datte betsunni ooen mitaini

paid part smarately Owen like
ne, maikai haratteru wake janai kara ne.
FP every- time paying case NEE bmuse FP

'Sinceapartyoipaid, smarately likeme, it isnotthe
casethatyoiarepayingeverytime.'

natte is rmeated twice in successim annd used again later.

When an utterance ends with nanka in fenale speech, it often
indicatesthatthemeakerconldcmtinnuem, butshedoesnotwantto
acpress the rest,- she lets the hearer(s) guess what she wannts to say.
Theuseofutterance-finalnanlainmalemeechisdifferent intlatthe
sentence is conplete in itself and name simply happens to be postposed
from its original positim. (For disonssim on Japanese postposirg, see
Simon, 1989.) In acample (37), C tells how thunder is frightening annd
dangeronsandmeakerArespominaninconpletesentenceendingwith
name.

(37) C:Shdd<ushi kananka shiran kedo,sannin
deathfronshockorsonethingdm'tlcnow but 3 person
gurai.
aboit

'onn'tknowifitwasadeathfrontheshockorsonething,
butaboit 3pemle (died).'

43

SpeakerAcoildhavefinishedherutterencebysayingsamesuchthingas
"(myn that is) too bad" or "(myl that is) terrible," but she did not.
However,wecaneasilyguesswhatmeakerAvantedtosayfrmthe
contact. In the following ample speaker A also implies sonething
afterthemarkernanm.SpeakerDconfessestlatshegainedmudnweignm
inthefirstyearintheU.S.becauseinthedomnitorysheconldeatas
muchasshewanted. SpeakerCstmsatnanJaanddoesnotconpleteher

sentence.

(38) D: Atashi we are c yatte ichinnenne sngoi futtocchattan

I S that 0 do first-year very gainweight
dakedo sa.
be FF

'1 didthat firstyearanndgainedsomuchweight.‘

C: S00 m...(mm)
yeah

'Yeah? (That's). . . . '

Speaker C coild have cominnued m by saying something like, "(Yeah?
That's) awful! I feel sorry for yon." or "(Yeah? That's) too bad, yon
shonldhavebeenmorecareful," bit shedidnot sayanything. The

44
inconplete (syntactically loose) sentence is also one of the

characteristics of mnen's speech (Shibamoto, 1985).
AsIhaveshown, nankaintheferalegronpis generallyusedto
soften the tone annd to inndicate indirectness.

4.4. hologram

In the male grow, two subjects (F and G) produced cmmiononsly
large number of clauses (see Table 3.), whichmeans the they dominated
the cmversatimandtheothertwo subjects listenedmost of the time.
Therefore, the tmics of the cmversatim were mainly F and G's
acperiences, minims, annd acplanatim abont sore events.

Themalespeakersalsousednankatomakea statement vague. In
the following ample (39), F tells what he read in a magazine abont
gradiate studentsinJapan. Therearethreeocoirrencesofmnka.

(39) G: Nihmno gakusee tte min are dayo ne, knnrai shi,
Japanese studem: suchbe FP notcheerfuland
min daigaku, daigaknninsee toka tokuni mi.chi

gradu, graduate student suchas especially street
no manmnaka aniite toorenai tte iu kara ne. un
of middle walk throgh-cnan-NEG say FP oh

konoaida,mnhAerakanankashinbmiyazasshi

recently Area sonething newspaper no magazinne
miteta nchkecb, ano nihm no daigakuin wa suramn
looted but well Japan in grad. school S slum

ka shiteru tte.
like become they say

'Japanesestudentsarenotcneerful. Iheardthatgradiate
studentscannotwalkinthemiddleofthestreet. Oh, recently

45

IlodtedatAreaorsonething, newspaper7nomagazine, well
theysaythatgraduateschoolsinJapanarebecom’nglikea
slum.’

Meemmoencescfnanlasinmmmdicatemncertaintyofthemeaker.
Thefirstnankamnldfmnctimasafiller'well',thesecondonecanbe
interpretedas "Iamnnot surebut...", anndthe thirdme enptasizes his
lackofcmfidenceinwhathewaslookingat (newspaperormagazine).
Thisusageofnankaisthesameasthatofthefenalemeakers.
Inthemalegrow,nankaisalsousedwithothercmjunctivesand
disconremarkers. Especially,themalemeakersoftenusednankawith
dalara. Dakana oconrred frequently with male speakers, but not with
fenalemeakers(seeFigure1.). Isumectthatitisbecausecveruse
of dakana soinds argumentative. Go-oconrrence of dalera with name
oconrredseventimeswithmalespeakersandmlymcewithfenale
speakers. In ample (40), dalar and nanka are used in successim.
SpeakerGacplainswhyhedoesnotlikeLmflergraoates'parties.

(40) G:mkara manila hanashi teteno tsuzukanai nda yo. Nanka
talk when continue-DEB FP

hanashitsuukakyoominotaishoogachigann kara.
talk or interestofsubjectSdifferentbecause

'EVenwhenI meakwiththem, it doesn't continue. Becausethe
subject or onr innterests is different.’

F: Un
yeah
'Yeah'

G:Da)narananla sa, ipp,i;wantekinna,mejaa nani toka sa
FP ge general major what like FP

46
(LAKE) sorede owacchau nnda yo ne.
then and FF

'(I ask abont) ge, general (subjects) like theirnajors, then
that's the end of the cmversatim.‘

Asadiscoirse marker, dakara tends to erphasize (assert) the following
statement. Thereforetheoconrrenceofnanlawithdaleraslowsthata
speakertriestoavoidastrogstatementorbeingassertive. Thenact
amplealsoshowscooconrrenceofdakaraandnanle. Ganplainswhythe

oil leaked oit from the pump of his car.

(41) G: Sono, sooiufuuni naru ichijikan ka nijikan maeni tashita
well likethatbecomelhonr or 2hours beforeadded

mayomnnnhsmotdcinikymininankahibiga
FP thattimeatsuddenly cracks

ookiku nattan ka shiranai kedo, waaatto moreta njanaino
big became or lam-NEG but a lot leaked

tabun.
maybe

'Well, Iaddedciloneortvol'oirsbeforeithappened. I
dm't krnow, maybe the cradcbecamebigger sudienly at that
time, but it leaked oita lot, maybe.'

SpeakerGactuallydoes notknowthe realreasonwhyoil leaked,
therefore, nanlaoconrredafterchlanatonakehiscomentless certain.

In additim, nanha is sometimes used to get me's turn in the male
grow conversatim. In ample (42), right after G's utterance, F
starts his turn with nanka to get other pemle's attentim, arnd he

continuestotalk.

47
(42) G: Kocchi wa ii yma, sutaffushigpu tte iumon gaaru kara.
here 8 good FP staffship called S be since

'Here (the U.S.) is good, becausethere is a thing called
staffship.'

F: Dhaka, ore mo daigakuinsee to tsukiatte, Kangaku no
I too grad.studem: with associate Kangakni at

koro, sono uchino kyooju ni isshookemee ato kuttsuiteitta
when well onr prof . demerately after followed

nnhyone.
FP

'I too associated with graduate students, when I was at Kangaknn
well, (they) desperately followed onr professor.

Speakechonldnavebegmnhistumwitlnntusingranka,bithedidnot.
Imeoilatethat, inthiscase, meakerFusednanletoavoidtooabrnwt
aturnshift. Forecverinthefollowing ample (43), speakeeregins
histurninthemtioileofH'sutterance, whentheyaredisonssingthat

anAmericanconpanyis easiertovorkforthanaJapaneseconpany.

(43) H: Ii imi. de yurniiku to in,
good sense in unique say
'It isuniqueinagoodsense...‘
G: m, unto sugoi (higakn made
really very university till

we ammari kanjinakatta chkede. Kaisha tte sa rnanka
S notnuch realize-NEB-PASI‘ mly conpany be FP
ojisan ippai' ' iru jan.

oldmen many be FF

'1 mly really didn't realize somuchuntill (I went to)
universiy. TherearemanyoldmeninaconpanyJ

48
The turn shift in ample (43) would also have been abnwt without

mania. Thereforenanlaisusedtogetme'stumandatthesametime
to avoid abruptness in turn shift. In the male grow conversatim,
thereweremanycasesWnentwoorthreepemle'suttenancescverlapped.
This snggests that getting one'sturnisharderthaninthefenale or
mixedgrow conversatim. Most likely, nankawas used as onne of the
devicestogainone'stum.
'Ihereismemorecharacteristicoftheuseofmnleinthemale
grow; itoconrredfinally. Whennankaoconrredsentence- or clause-
finnally, it is oftenaresultofpostposiig, flat is, nnainlemovedfrom
initial or medial positim to the final positim. Since this postposing
ofnankafrequentlyocoirredwiththemalemakers (especiallyinthe
mdxedgmm),1willsrowtheamplesofpostposednamfleinthena¢

sectim.

4.5. “grow

Thespakersof themixedgrowweretmferalesandtvomales.
As table 3. shows, however, the total number of name with fenale
makersismly2,butwithmalemakers96. Moreover,the
cmversatim was mostly dominated by male meakers (total nurber of
clauses: fenale 379, male 892). Therefore, the results of this grow
regardingtheuseofnankamainlycmcernthemalespakersinthe
presence of the fenale speakers.

NanJa often oconrred sentence- or clause-finally with male
makers. In the fenale grow cmversatm, too, nanka was used finally,
bititindicatesthemaker'sreluctancetocmtimemakingandthe
sentenceswereallinconplete. antheotherl'and, finalnankawiththe

49
male makers is clarly the result of postposirg. It is interesting

that postposed nanla oconrred frequently with male spakers, since
postposing is generally ocoirs more in men's meech (Shibannoto, 1987:
Simon 1989). Ebmple (44) shows that thogh name wonld normally be in

utterance-initial positim, it was added at the and. The sentence is

conplete.

(44) G: Zenzen dame nannda yo nne, numb.
atall nogood FP

'It is not good at all.'

ample (44) incanmicalwordorderwonldbe "Wmdmenanda
yone." Inthiscase, renaisusedtoshowanindirectness. Ebenple
(45a) below also shows postposing of nanka. Notice that the utterance-
finalocoirrenceofnankahereisdifferentinaturefrmtratinthe
fenale maker's ample (38), repated belowas (45b), in that (45a) is
a complete sentence whereas (45b) is inconplete.

(45a)G:Kocchikarawayasuikedo, mnnkoo karawatakai
here frodenapthoghcthersidefromSacpensive

rashiindayo.

har FP
'It's chap from here, but from the other side it's acpensive,
I hear.‘

F: Assoc, hooo.
ohyah, Isee

'anyah, I see.'

G:Dakara kocchi karabakkari okutteru mania.
so here frommly sending

50
'SoIanmsendingfromheremly.'

(45b) D: Atashi we are c yatte ichineme sugoi futtodnattan

I S that 0 do first-yar very gainweiglnt
dakeon sa.
be FP

'I didthat firstyarandgainedsomnchweight.’

C: S00 mun (LAID-I)
yah

'Yah? (That's) .... '

Thesentence-finalnankain (45a) wonldoconratthebeginnirg inanim-
postposed, canonical sentence. This finnal nanle is used to soften G's
whole utterance which started with Galena. Postposing is said to oconr
to acpress an importannt element (informatim) first, arnd the element
which amars in final positim is considered secondary informatim.
(Simon, 1989)

With the male makers in mixed grow, nanka often co-occurred
with the filler iya 'well' (eight times), which did not oconr at all
with fenale makers (even in a fenale grow cmversatim). In ample
(46), makarFuses iyaandnanka insuccessimat thebeginningof his

utterannce.
(46) F: m mania ne, tamani aru jan, koo tenjoo
well FP occasionally be like this ceiling

terasu yomi natteru yatsu.
light way becore thing

'Well, occasimally there is (a light) which lights the ceiling
like this, '

5)
Thefilleriyaisusedwhenthemakergrmesformrds. The

conbiantim of iya and mania seens to deenphasize the whole utterance.
In ample (47), spakers F annd G disonss how to play tennis, and
makerGuses iyaandnanlainsuccessim.

(47) G: tekubi tsukawanai yone.
wrist use-NEE} FP

'We dm't use wrist, do we?‘

F: Tsukawanai yo.
use-NBS FP

'We don't use it.'

G:Iyanlnlatenisukookoo notokiniyatteta
well tennis high school when in do-PASI'

ndakedo,sontoki itsunoiwaretetano mahcnkano
but thattime always wastold thing S other

mono oshitaraikeaitte.
thingom nnot

'Well, IplayedtenniswhenIvasinhigh school, but at that
timeIwastoldnottodootherthings (sports).'

Tosunmarize,throghthreegrows,nanleisusedmostlytosoften
thestatenents, toshowuncertainty,andtoavoiddirectandabrtwt
utterances. Wheinanlaisrepeated, itnnotmlyincrasesuncertainty,
but also fuctims as a filler. A co-occurring disconrse marker or
filler with nanka differs depending on the grain: however, name
functims,sotospak,asabufferineadncase. Table6isasmmary
of characteristic featuresofnanlabyfenale, male, andmixedgrows.

Inthena¢sectim,Iwilldisoissthefmnctimsofmnkabased

mtheabcveresults.

52

 

 

Table 6. Sumary of characteristic features of nanka
by fenale, male, and mixed groups

Characteristic features of nanka

Fetale group 1. uncertainty, softener, indirectness
2. co-occurrence of dam and Hanna
3. renetiticn
4. incomplete sentences

Male group 1. uncertairmy, softener, indirectness
2. co-occurrence of dakara and nanka
3. turn entry device

Mixed group 1. uncertainty, softener, indirectness

2. postposirg
3. co-occurrenceof iyaandnanla

 

m5
DWANDW

As it has been denmstrated, there are disconrse markers in
Japanese. As faras nydata are cencerned, theyarenanla, dalara,
dam, datte, and sorede. 53de has its own literal use. In a
discourse centem, howeer, it often playsadifferent role.

JapaneseandEnglishdisooursenarkersdonotalvayshavethesane
functiens. Nanka functions all as softener, tum initiator, and filler
andsuchawordcamotbefcundinEnglish. Dalera,whichenunasizes
thefoflowirgstatennemdoesrntravetheserefmnctimasthefinglish
wordsso. Soredeindicatesaresultasadisoonrsenarkernuchnore
strengly than the English discanrsenarker and. 'Iheuse of datte, which
asserts ene's opinion, is also quite different from the English
discanrsenarkerbecauseammt. Chlythenarkerdemseenstohave
almost the same fmnctian as its English equivalent but; it indicates
centrastwiththepriorsubject. 'ItnsthehypothesisAthatJapanese
discanrsenanjcershavealnnstthesauefmnctiensasthcseofmnglish
discoursenaricersisnotcenfirmsdexcentforthenarke‘dem.

Thediscoursenarkersinnydatamightseanveryfevinmmber. I
anspectitisbecmsetheintevievswereinfomalarflthesubjectswere
generallyywng (nostofthenwereintheirtwenties). It is easily
enmnisedtlatiftheinterviedsmanversatienmmdbeenmefornaland

53

54
between people not so close or with older people, such as, in business

neetings, fomal parties, and classroan sitnnatiens, the result would
havebeendifferent. Social distance influences thewhole organizatien
of the cmversatim (Ventola, 1979). Thus, words such as shikashi
'hxt', kerehm'b.nt;hmever'whidnareusuallyusedinafornal
settingendcanfnmctienasdiscanrsenarkersvmldlikelymmenindne
settimsnenticnedabove,mttheydidnotintrychta.
finedisoanrsenarkerswhidnirxiicateindirectxessanflvagueness
werenotfoxmdexcentthenarkernanle. Itisworthnotingthatthe
occnrrenceofnankaoccupiesnorethansixtypercentofallthe
oconrrencesofdisccursenarkers. (SeeFigure l.) Interestingly,dnring
theinterviens,1didnotncticesonanyoconrrencesofmnka; however,
when I listened to the tapes after the interviews, nanka is actually
used very frequently and occurs at various points in a cmversatim.
nus,thehypotlmissthattherearesaneJapanesediscwrsenariosrs
which indicate indirectness and vagueness is partially supported by the
frequentocanrmceofnanka,hntotherdisomrsenarkersdonotstm
sudnfunctiens.
unethingtratlhadnotexpectedbeforetheintendewsuaslarge
individnaldifferencesinmnberofclauses. Inthefenalegrmp,ene
speaker(spekerD)spd<ealmsttwiceaslmgasotherspeakers(D=417,
speakerskmé, 3:197, @217). Intheualegram,speakerFuttered 516
clannsesvtfidnisalnnstsixtinesaslargeastlatofspeaherEmM,
vanoparticipatedleastinthesenegrwp.(8ee'1able4.) Isuspectthat
thereasanfortherelativelynanyutterancesWspeakersDandFis
that, since they were participating as the interviewer, they felt
respcnsihletokeentheoenversatienflowdnringtheinterviev.

SS
Omcerning thenale group, partianlarly speakers Fandeminated

the cmvermtian. The possible reasen for this is the age difference,
which is an extremely influential factor in Japanese discourse in
general. Nakane(l970)poim:soutthatineverydayaffairsenealvays
has tobeawareof theaddressee'sageandrankinorderto
differentiateone'sspeechstyles. Eventhoughallthespeakersinny
studyarefriendswmaluaysspeakininfomalstyle,speakersFandG
aregradnatesuflemsandspeakersnandfiareundergradnatesuflenzs.
Therefore, the difference in age and also rank (gradate or
undergradnate) might have influenced their cmversatim.

As in the male group, speakers F and G also dominated the
cewersatianinthenuxedgmnp,eventrnnghtherewerenctcanspiams
age and rankdifferences (accent for one fenale speaker who is an
undergradnate student). 'IhenwhydidspeakersFandeanimtethis
groupalso? Itiscenceivablethattheualespeakersvantedtoslm
theirsuperioritytotheferalespeakersbyspeakingabanttheir
knowledge and experience. Another possibility is the persenalities of
speakers F and G. They are generally talkative and always try to
entertain others (this certainly is true inthemalegroup
cmversatim). Asaresult,speakersFanthaJJcedalotccnparedto
theferalespeakes.

Interestingly in the nale gronp, the speakers with power over
others (resulting frcm higher rank, for exanple) spoke more, and as a
result,prod.ncednnrenanka. 'Ihenalespeakerswithlesspowertran
othersspdcelessandhenceprodncedfevernanka. 'Iherefore,inte:ns
of quantity of speech, I feel that power is an inportant factor
especiallyfornalespeakers. Regardirgthefetalegmnp,antheother

56
harriwhetherenehasponerornotdoesmtseentoinfluencethemrber

of utterances, at least in my data. However, it does influence the
ocanrrenceofmnle. Itneparticiparmwhohasnnrepcwerusednanlanore
frequently and the participant who has less power used it less
frequently.

With both fenale and male speakers in the mixed grunp, the
oconrrenceofnanledecreasedomparedtothesenegendergranps,eocent
inthecaseofanenalespeakerklunghthepercemagestayedthesele).
(See Table 4.) This suggests that in this case the presence of omesite
sexnayhaveanaffectmtheuseofnanla. Adler (1978: 46)pointsout
that subjects tendtodecreaseintensitytothesane—sec ecperimenter
andincreaseintensitytotheqposite-sex experinenter. Inmystudy,
it seared that both fenale and nale speakers were smedhat censcious
abont thepresence of the omosite sex, even though they are in
relatively similar status in terns of age and rank. This result might
suggesttlattheocenrenceofnankaissensitivetohmnudnaspeaker
isreJaxed,thatis,whenthespmkerisrelaxedhe/sheusesmnlannre
freJuentlythanwhenhe/she isnctrelaxed.

Another point that I had not anticipated was that the male
speakersprodncedmnlenudnnoreoftenthanlecpected. Shibamoto
(1.985) states that wmen'sspeednismrepoliteandsofterthanmen's
speech. Asnankaiseneofthesofteningdevices,1hadexpectedthat
thedifferenceinfrequencyofnanlebetweenfenaleandnalespeakers
wmldbelarge;m»ever,itvasrntthecase. anthecentrary,nctanly
didthenalespeakersusenanle nudnnoreofteninthemixedgmnp,in
general, theyusedotherdisccursenarkersfrequentlyaswell (see
Figure 1.) Fenalearxinalespeakersusedrmflcain different ways.

5 7
Fenale speakers tend to use it in repetition, in succession with the

disconrse marker dam, and in incorplete sentences. Renetition and
n'ncorplete sentences are general characteristics of wonen's speech
(Shibancto, 1985). antheotherhand,ualespeakerstendtousenanla
asaturn-entrydeviceandasaresultofpostposing,anditoftenco-
oconrredwith dahara. 'Iherefore, thehypotlesisCthat there are
differencesintheuseofdisconrsenarkersbetweenmenandwonenis
supported; however, itisonlywithnanka. 'Iheuseofotherdisccurse
naricerswerenotanalyzedinthisstudy.

antheotherhand,irxiivionaldifferenceswerenbreprordnent.
Speaker A (fenale) and speaker G (nale) used name very often.
Furthermore, thesespeakersusednanyotherdisconrsemarkersand
fillers conpared to other speakers. Speaker A generally talked slowly,
andmenshetalkedsheseenedtobetryingtochoosethebestwordin
theconzextasnudnassheconld. Asaresult,sheusedmaryfiller
words. mreover,herpersonalitymiglmhavecontrihntedtoprodncing
frequentnenla. Altl'nnghshespdceoften,shealvaystriedtobenndest
and respected other people's opinions. It is possible that this
tendencyfrequemlytriggeredhertousenameasasoftener.

SpeakerGistheoldestparticipartinthisstudy. Hehadfive
yearsofmrkirgexperiencebeforehecauetotheU.S.andheis
generally considered a very mlebeable person. Since he likes to
sharehisexperience,themmberofhisutteranceswasverylarge. Also
heofteninternnptedotherspeakers'utteames,andrmnkavasusedasa
turnentrydevice. Isuspect thatthisisoneofthereasonswl’ynanka
occurredveryofteninhis utterances.

58
Ontheotlnerham,SpeakerEusednanIeleastanongtheeight

subjects of this study. He is theyoungestperson. Ihadnoticedprior
tothis studythatheisageconscionsandthathesonetinesusesfornal
speechstyleeventoclosefriendsaslongastheyareolderthanhe.
'Iherefore, forhim, thisageandrankdifferencenayhavekeat himfronn
participating mare actively in the conversation.

In conclusion, it is possible that the occurrence of raffle is
sensitive to age, rank (i.e. power) differences, arnd the degree of
relaxation, bit not gender. Moreover, prediction of mania is closely
related to indivional differences.

Aswehaveseen, thedisconrsenarkernankaseenstohaveseveral
functions. First, name contributes greatly to softening the utterance.
Mnynanleisusedsofrequentlytosoftentheutterance? This canbe
explained fron one of the characteristics of Japanese conversation. In
nychta, thereweresonecasesthatthespeakersexd’nangeddifferent
cpinions. In Japanese, peqnle generally try not to disagree, bit when
they have to do it Japanese peqnle usually hesitate rather than overtly
disagreeing (Hinds, I976: 141). This hesitation device could be the use
of nanka, which results as softening the utterance (opinion). For this
reason, nanka was frequently used.

Also,nankaasasoftenerrenresen:sarntherinrportantcharacter
of theJapanese language. Japanesepeqnledonot liketoexpressthings
directly. Kindaichi (1975) states that in Japanese one prefers to say
everything moderately (gently) as much as possible, and also that
Japanese people generally agnreciate indirect expressions. Moreover,
Itasaka (1978) observesthattheJapamselanguagehasatendencyto
avoid tellirng right fron wrong. His stateent might be too extrnene, but

59
toacertainextent it explainsthereasonwhynanleisusedso

frequentlytonaketheutterancevague,asiftowrapitinawafer.
NenJausnnallycbesnotoconrwhenaspeakertallcswithdefinite
certaintyorwhenaspeakerassertsaverystrongopinion. Therefore,
mice is oftenused in reported narratives, since in renorted narratives
itisthecasethatthestoryismttheirovneqnerience,bntsonething
they have heard fron sonnebody else, so the reliability of the story
tendstobelowandthespeakertellswhathe/shehasheardwith
uncertainty. 'Ihereforenankaoccurredfrequently.
Second,nan}ecanbeginanewturn. Asshcwnin'IableS.nanla
occurring initially shows the second largest percentage. Especially
with the nale speakers' conversation in which the utterances often
overlappedandtakingone'sturnvasnorediffionlt,nan1eseenedtobe
usedtogetone'sturn. Isuspectthisisduetothenatureofthis
word. Asthespeakeravoidsanabruptturnshift,thewordnnam<ais
uttered first to get an attention of other people.
Third,nankaalsohasafunctionasafiller. Sinceitisusedto
fillapausevmenaspeakerseeksawordjustlilceano,andsono.
WehaveenaminedthedisconrsenarkersinJapanese,partionlarly
ranka. Aldunglnmnbisusedeothmenarfiwonnenitissensitiveto
age, rank (power), andthefonnality level of talk. 'Ihefunctions of
nankaarel) softener, 2) turninitiator, and3) filler. Thisnrarkeris
usedineverydayconversationtonakeit coherent--whidncamnotbe
explained onlybythe literal sense of the word, "sane, any, sorething,
andanything."

MS
mammonsmmmm

In this section, I will discuss what could have been done for
obtaining acre legitinate result of this study and sore suggestions that
thisstudycculdleadinthefuture.

First,InadethehypothesistIattherearedifferencesintheuse
ofdisconrsenarkersbetweenmenandmnen,anditwassunportedwith
mania. On the other hand, there is not an inportant difference
accordingtothegenderinprodncirgthemnberofnanka,emceptinthe
mixedgronp. Ibelieve that the reason why there is nota large
differencebetweennaleanriferalespeakersisbecausetwonaleanfitwo
fenale speakers participated in two groups, n'ale/fenale andmixed. 'Ihat
is,speakersCandD(ferale)andspeakersFandG(nale)corposedthe
mixgroup. Myintentuasthattheconversationswonldbeininfornal
style, whichmsant that the participants had tobe friends whokncw each
otherequallywell. 'Ihespeakerswhoparticipatedinthemixedgronp
seared to be sonewhat influenced by participating in exclusively one
geriergronp,sincethemixedgronpintervievwasheldafterthesane
gendergroupintervieds.

Second the role of the interviewer slnculd have been considered
norecarefully. Inthefflearxinalegronps,thespeakerswtnwere
assignedtobeintervieverstendedtospeaknnrethanotherspeakers.

60

6|
Wylnyselfvastheintevieverofflnefenaleanflmixedgmnps

(speakerD). I felt trat especiallyinthefenalegroup, Ihadtospeak
acre as interviewer because every time a tonic ended, other participants
asked ne (interviewer) for another tmic. The shift in roles
(interviewer and participant) might have influenced the participants'
speech style. this problem also nakes us think again what a linguistic
interview should be. It is inportant, or even crucial, to create a
natural settingsothatparticipantscbnotfeeltrattheyare
participating in a study.

Third, the social status (power) and the personalities of the
partich shonld have been taken into account. Though the
personality traits are less controllable, speakers' speech style, such
as fast and slow, talkative and less talkative, might very well
influence the result. Rank difference is a very delicate issue annog
Japanese people. I had assured that the difference of gradate and
mndergradnatedoesnotnakeanydifference; however, it actuallya sort
of superiority in male gradate speakers. Perhaps either just graduate
or undergradnate students should have been chosen. Especially in the
nalegrcup, thediffernences inageandrankwerelarge.

This study suggests nany possible future research directions.
First, hcwthedsgreeoffornalityof speechaffectstheoconrrenceof
nanka. For exannple, in fornal and informal parties and meetings, how
large a difference nay be dnserved. Second, as Sawyer (1992) analyzed
second larguage learners' acquisition of sentence-final particle ne, it
wouldbeinteresting toobservehmnanleorotherdisconrsenarkersare
acquired by non-native speakers of Japanese. Similarly, as Hays (1992)
onservedtheuseoffinnglishdisconrsenarkersbyrativespeakersof

62
Japanese, it will be worth finding out how proficiency level in Japanese

correlates withuse of disconrsemrkers; for instance, itnaybe that
if learners are proficient in speaking Japanese, do they use more
disconrsemarkersornot? 'Ihird, doesthelengthof residenceinthe
U.S. influence Japanese people's use of discourse narkers?
Althoughthepresent studyonly touchedononeaspect of Japanese
discoursenarkers, Ihopethat itwill helpus tounderstandnoreabont

Japanese infomal conversation.

nAPPiflHIEX

IJUHDRIHUPICRIJSEIEEP

Please answer the following questions.

1. Name:

 

2. Age:

 

3. Sex: Male Female

4. Classification (circle one): Fr. So.

5. Major field of study:

Jr.

Sr.

 

6. Hometown in Japan:

 

7. How long have you been in the U.S.?

Grad.

 

Thank you very much for your cooperation.

63

BIBLIOGRAPHY

Adler, M. K. (1978). Sex Differences in Human Speech: A
sociolinguistic study. Helmnt Buske Verlag Hamburg.

Beaman, K. (1984). Coordination and subordination revisited:
Syntactic complexity in spoken and written narrative
discourse. In D. Tannen. ed. Cbherence in Spoken and
written Discourse. Norwood, New Jersey: Ablex.

Chafe, W. (1980). The deployment of consciousness in the
production of a narrative. In W. Chafe (ed.), The pear
stories: Cognitive, cultural and linguistic aspects of
narrative production, 9-50. Norwood, N.J.: Ablex.

Cook, H. M. (1990). The sentence-final particle ne as a tool
for cooperation in Japanese conversation. In Hoji, H
(ed.). Japanese/Kerean Linguistics. Center for the

Study of Language and information. Stanford,California.

Goffman, E. (1981). Fbrms of Talk, 124-57. Philadelphia:
University of Pennsylvania Press. (Originally published
in Semiotica, 25: 1-29. 1979.)

Halliday, M. A. K. and R. Hasan. (1976). Cohesion in
English. London: Longman.

Hays, P. (1992). Discourse markers and L2 acquisition. .A
paper presented at the twelfth Second Language Research
Forum, Michigan State University, April.

Hinds, J. (1976). Aspects of Japanese Discourse Structure.
Tokyo: Kaitakusha.

Ishikawa, M. (1988). The cognitive and Interactional
functions of the Japanese sentence particle ne in
Discourse. Southern Review, 3: 24-35.

Ishikawa, M. (1992). A.synthesis of three approaches to
discourse: An analysis of the Japanese sentence particle
ne. .A paper presented at Linguistic Society of America
Annual Meeting. Philadelphia, PA.

65

Itasaka, G. (1978). .Nihongo no hyoojoo [Expressions of
Japanese language]. Tokyo: Kodansha.

Kindaichi, H. (1957). Nihongo [The Japanese language].
Tokyo: Iwanami Shoten.

Kindaichi, H. (1975). Nihonjin no gengohyoogen [Language
expressions of Japanese people]. Tokyo: Koudansha.

Labov, W. (1972). Sociolinguistic Patterns . Philadelphia:
University of Pennsylvania Press.

Labov, W. and D. Fanshel. (1977). Therapeutic Discourse:
Psychotherapy as Cbnversation. New York: Academic
Press.

Maynard, S. K. (1989). Japanese Cbnversation: Self-

Contextualization Through Structure and Interactional
Management. Advances in Discourse Processes Vol. 1.
Norwood. New Jersey: Ablex.

Mizutani, N. (1985). .Nichiei hikaku: Hanashikotoba no
bunpoo [The Comparison of Japanese and English: A
grammar of spoken language]. Tokyo: Kuroshio shuppan.

Nakane, C. (1970). .Nihon no Shakai [Japanese Society].
Tokyo: Kodansha.

Ohta, Ar S. (1992). Evidentiality and politeness in
Japanese. Issues in Applied Linguistics, 2, 2, 211-38.

Owen, M. L. (1983). Apologies and Remedical Interchanges.
The Hague: Mouton.

Polanyi, L. (1979). So what's the point? Semiotica, 25:
207-41.

Quirk, R. (1955). Colloquial English and communication.
Studies in Cbmmunication, 169-80. London: Secker and
Warburg.

Sacks, H., E. Schegloff, and G. Jefferson. (1974). .A
simplest systematics for the organization of turn-taking
in conversation. Language, 50: 696-735.

Sawyer, M. (1992). Toward developing pragmatic competence
in Japanese as a second language: The affective particle
ne. A paper presented at the twelfth Second Language
Research Forum, Michigan State University, April.

Schegloff, E..A. and H. Sacks. (1973). Opening up closings.
Semiotica, 7: 289-327.

66

Schiffrin, D. (1987). Discourse Markers. Cambridge:
Cambridge University Press.

Shibamoto, J. S. (1985). Japanese Nemen's Language.
Orlando:.Academic Press.

Shibamoto, J. S. (1987). The womanly woman: manipulation
of stereotypical and nonstereotypical features of
Japanese female speech. In S. U. Philips, S. Steel, and
C. Tanz, eds. Language, Gender, and Sex in Cbmparative
Perspective. New York: Cambridge University Press.

Shinmura, I. (1955). Kbojien. Iwanami Shoten.

Simon, M. E. (1989). An analysis of the postposing
construction in Japanese. Unpublished dissertation.
University of Michigan.

Stubbs, M. (1983). Discourse Analysis. Chicago: The
University of Chicago Press.

Stubbs, M. (1984). Applied discourse analysis and
educational linguistics. In P. Trudgill. ed. Applied
Sociolinguistics. London: Academic Press.

Uyeno, T. (1971). A study of Japanese modality: A
performative analysis of sentence particles. unpublished
dissertation. University of Michigan.

Ventola, Eija. (1979). The structure of casual conversation
in English. JOurnal of Pragmatics, 3: 267-98.

Von Glaserfeld, E. (1974). Because and the concepts of
causation. Semiotica, 12: 129-44.

Wootton, A. J. (1981). The management of grantings and
rejections by parents in request sequences. Semiotica,
37: 59-89.