Pragmatics and Pedagogy: Conversational Rules and Politeness Strategies May Inhibit Effective

Tutoring
Author(s): Natalie K. Person, Roger J. Kreuz, Rolf A. Zwaan and Arthur C. Graesser
Source: Cognition and Instruction, Vol. 13, No. 2 (1995), pp. 161-188
Published by: Taylor & Francis, Ltd.
Stable URL: http://www.jstor.org/stable/3233712
Accessed: 04-09-2015 06:44 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3233712?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to Cognition and Instruction.

http://www.jstor.org

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
COGNITION
AND INSTRUCTION,
1995,13(2), 161-188
? 1995,Lawrence
Copyright Erlbaum Inc.
Associates,

Pragmaticsand Pedagogy:
ConversationalRules and Politeness
StrategiesMay Inhibit
EffectiveTutoring
Natalie K. Person, Roger J. Kreuz,
Rolf A. Zwaan, and ArthurC. Graesser
Departmentof Psychology
Universityof Memphis

In this article,we identifyways thatGrice's(1975) conversational

rulesandP.
Brownand Levinson's(1987) politenessstrategiesare commonlyemployedin
one-to-onetutoringinteractions. We examinedtwo cross-agedtutoringcorpora
fromresearchmethodsandalgebratutoringsessionsto showhow theserulesand
strategiescanpotentiallyenhanceandinhibiteffectivetutoring.Examplesof these
costs and benefitsare presentedwithina five-stepdialogueframeproposedby
Graesserand Person(1994). Thereappearto be differencesin the use of these
politenessstrategieswhenalgebratutoringprotocolsarecomparedwith research
methodsprotocols.We suggestthatpolitenessstrategiesaremoreprevalentin less
constrained domains,even thoughtheiruse may inhibiteffectivetutoring.

Although tutoringhas been employed as a pedagogicaldevice for millennia,only

recently has the process of tutoringbeen investigated scientifically (Fox, 1993;
Graesser& Person, 1994; Leinhardt,1987; McArthur,Stasz, & Zmuidzinas,1990;
Putnam, 1987; VanLehn, 1990). An understandingof the tutoring process is
important,because tutoringtypically is more effective thanclassroom instruction.
Althoughmost researchersin this areahave referredto a positive cognitive change
as an essential part of effective tutoring(e.g., Palincsar& A. L. Brown, 1984;
Newman, Griffin, & Cole, 1989), a universally accepted definition of effective
tutoring has not emerged. Researchers examining the advantage of tutoring
comparedwith classroom instructionhave reportedeffect sizes rangingfrom .4 to
2.3 standarddeviationunits(Bloom, 1984;Cohen,J. A. Kulik,& C. C. Kulik, 1982;

Requests for reprintsshould be sent to Natalie K. Person, Departmentof Psychology, Rhodes

College, 2000 North Parkway, Memphis, TN 38112. e-mail: person@vax.rhodes.edu

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
162 PERSON, KREUZ, ZWAAN, GRAESSER

Mohan,1972).Thisadvantage existseven thoughtutorstypicallyarenothighly

trained(Cohenet al., 1982;Fitz-Gibbon, 1977;Graesser,1993a,1993b;Graesser
& Person,1994).In addition,thisadvantage is notattributable
to agedifferences,
becausethe effect sizes previouslycited includeexpert,cross-aged,and peer
tutoring.
Becausethese effects arenot due to sophisticated pedagogicalstrategiesor
age differences,the truecausemustlie elsewhere.One possiblereasonfor the
effectivenessof tutoringrevolvesaroundthe dialoguethatoccursbetweenthe
tutorandstudent.Tutoringdialogueis moresimilarto normalconversation than
is thelectureformatusedprimarily in theclassroom(Resnick,1977).Thepurpose
of this articleis to illustratehow propertiesof conversationcan potentiallyaid,
as well as hinder,effectivetutoring.

PERSPECTIVESON CONVERSATION
THEORETICAL
The implicitrules and strategiesthatfacilitatenormalconversationwere first
describedby Grice(1975, 1978),who proposedthatconversationis governed
by one overarching rule:the cooperativeprinciple.Accordingto this principle,
participants makea "goodfaith"effortto contributeto and collaborateon the
conversation as it proceeds(Clark& Schaefer,1987, 1989).Grice(1975, 1978)
furthersuggestedthatthis cooperationis augmentedby a numberof conversa-
tionalmaxims:quality(do not say thingsthatareuntrue),quantity(do not say
moreor less thanis required), relevance(do not say thingsthatareextraneous),
andmanner(be brief,be orderly,andavoidobscurityandambiguity).
OthertheoristshaveexpandedGrice's(1975, 1978)approach.Forexample,
Leech(1983)suggestedanoverarching politenessprinciplewithseveralmaxims
(e.g., tact, generosity,approbation, and modesty).Leech maintainedthat the
politenessprincipleis necessaryfor Grice's(1975, 1978)cooperativeprinciple
to be effectivein normalconversation.
Thisinterestin linguisticpolitenesshasbeenmostfullyexploredby P. Brown
andLevinson(1987).In an analysisof languagesusedin threewidelydiffering
cultures(Englishin the UnitedStatesandBritain,Tamilin India,andTzeltalin
MayanMexico),P. BrownandLevinsonfoundthatparticular politenessstrategies
areuseduniversally.Theydescribeddozensof conversational strategiesthatare
usedto facilitatesocialinteraction. Centralto P. BrownandLevinson'sanalysis
is thenotionofface, orone'sself-image(Goffman,1967).Individuals in a culture
attempt to maintain a and
positiveself-image try helpto others maintain their
ownself-images.Thisis notalwayspossible,however,becausefaceis frequently
putin dangerby face-threatening acts.Suchactsincluderequests,criticisms,and
demands.Eachculturehas deviseda numberof linguisticstrategiesto mitigate
the impactof theseface-threatening acts.
P.BrownandLevinson(1987)discussedthreesuperstrategies thatspeakersmay
employ:positivepoliteness, negativepoliteness, andoff-record.Positivepoliteness

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICSAND PEDAGOGY 163

refersto an appreciation of thehearer'swantsandneeds.It includesthe acknowl-

edgmentof commonground(Clark& Carlson,1981;Clark,Schreuder, &Buttrick,
1983),the acknowledgment thatthe speakerandhearerare cooperators,anda
readinessto fulfillthespeaker'swants(P.Brown& Levinson,1987,p. 102).For
example,bysaying,"Thatsculpture is reallybeautiful,ina way,"thespeakerseeks
to avoiddisagreement withthelistener.Perhapsthespeakerdoesnotreallybelieve
thatthesculpture is beautiful;andbyhedging(i.e.,saying,"ina way"),thespeaker
is ableto providetokenagreement withthelistener.
in
Negativepoliteness, comparison, attemptsto minimizeimpositionson the
hearer.This somewhatconfusingtermdoes not implya lack of politenessbut
rathera methodof ingratiation. Negativepolitenesscan be accomplishedin
variousways: The speaker can minimize the threatto thehearer,give the hearer
the option not to act, or communicate the speaker'sdesirenotto impingeon the
hearer(P. Brown & Levinson,1987,p. 131).By saying,"Ijust wantto ask you
if you can lend me a tiny bit of paper,"the speakerminimizesthe imposition
on the listenerby makingthe requestas understated as possible.
A final superstrategy discussedby P. BrownandLevinson(1987) involves
going off-record.A speakermakesa statementbut does so in a vague way,
leavingthe interpretation of the commentopen to the hearer(Craig,Tracy,&
Spisak,1986).Forexample,by saying,"It'shot in here,"the speakerindicates
her or his desireto have someonetakeaction,perhapsby openinga window.
This requestis madeobliquely,however,and gives the listenerthe optionnot
to act.
Manyof theseoff-recordstatements canbe construedas violationsof Grice's
(1975) conversational maxims.For example,a speakermay violatethe maxim
of relevanceby providingthe hearerwith a hint or may violatethe maximof
quantityby overstatingor understating. The maximof qualitymay be violated
by the use of irony,metaphors,or rhetoricalquestions.The maximof manner
may be violatedby the use of vague or ambiguousstatements(P. Brown&
Levinson,1987,p. 214).Inall theseways,thespeakercansavefaceby delivering
face-threatening acts in an indirectway.
P. BrownandLevinson's(1987)analysisof politenesshas becomethedomi-
nantperspectivein the areaof linguisticpoliteness(Fraser,1990;Kasper,1990).
Thereare,however,severaldifficultiesin applyingthis theoryin empiricalre-
search.Mostimportant, as Craiget al. (1986)pointedout,thedozensof individual
politenessstrategiesare not mutuallyexclusive;manyspeechacts can be justi-
fiably coded underdifferentstrategiesand even differentsuperstrategies. For
example,manyrequestsfor actionareconventionallyindirect(negativepolite-
ness) andunderstated (off-record).
These difficultieshave led some researchers to proposenew approachesto
linguisticpoliteness(e.g., Penman's[1990] analysisof courtroomdiscourse).
However,no alternative schemehasreceivedmuchsupport.As a result,it is not
possible to use P. Brown and Levinson's (1987) taxonomy to test empirically

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
164 PERSON,KREUZ,ZWAAN,GRAESSER

some of the issues thattheirtheoryraises.As a generalframework, however,

the approachhas clearutility.
P. BrownandLevinson's(1987) analysisof linguisticpolitenessmay have
greatrelevanceto tutoringinteractions. Accordingto P. BrownandLevinson(p.
74), thedegreeto whichan actis facethreatening is determined by threefactors:
(a) socialdistancebetweenthe speakerandhearer,(b) the relativepowerof the
speakerandhearer,and(c) thedegreeof impositionof the act.If socialdistance
is high,relativepoweris asymmetrical, andimpositionis high,thenthe"weighti-
ness"of the face-threatening act is high.Forexample,a lawyerinterrogating a
widowabouther husband'smaritalinfidelitiesmightmakeheavyuse of these
strategies.In mosttutoringsituations,thefirsttwofactorswill be relativelyhigh,
becausethe tutorandstudentdo not knoweachotherandthe tutoris clearlyin
control.The thirdfactor(degreeof imposition)will vary, dependingon the
seriousnessof the imposition.
Otherresearchers haveexaminedP. BrownandLevinson's(1987)politeness
strategies in thecontext of requests(Craiget al., 1986),medicaldiscourse(Arons-
son & Rundstr6m,1989), and courtroomdiscourse(Penman,1990). This ap-
proach,however,hasnotbeenutilizedto clarifytheprocessof tutoringdiscourse.
Yet it is likely thatthe variouspolitenessstrategiesarefrequentlyemployedin
tutoringinteractions.
The purposeof this articleis to illustratehow the politenessstrategiesare
used, both positivelyand negatively,duringone-to-onetutoring.We do not
attemptto quantifythevariouspolitenessstrategies,however.At thisearlystage
of research,it is premature to analyzestrategiesquantitatively at a fine-grained
et
level (see Craig al., 1986). The strategieswill no doubt vary considerably
amongtutors,topics,andstudentpopulations.Ourprimarygoal at thispointis
to documentsomeof the interesting waysin whichpragmaticprinciplesinteract
withthe tutoringprocess. Future studies will quantifythe use of thesestrategies
in differenttutoring contexts.
Theconversational rulesandpolitenessstrategiesdiscussedin this articleare
summarized Table Thisarticledoesnotaddressallof P.BrownandLevinson's
in 1.
(1987) strategiesbutratherthe subsetthatseemsmost germaneto the tutoring
process.In otherwords,therulesandstrategiesthatwe discusswereinductively
selected,basedon a closereadingof thetranscripts. Someof thecostsandbenefits
associated with the rulesand strategies pointed in Table1.
are out

OF TUTORING
FRAMEWORK
THECONVERSATIONAL

Graesserandhis colleagues(Graesser,1993a,1993b;Graesser& Person,1994)

examinedthe framework An extensiveanalysisof tutor-
of tutoringinteractions.
corporarevealedthata five-stepdialogueframeis veryprevalent
ing transcript
duringone-to-one tutoring:

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 0 v 0

U ~ , , o
U' *au t)
u
t
r.

CL b3 0u 3o
0 0

t) t)E Q) v, c;~
0 0Q Q
So _ CL
0
9 1* o o
o
2 ~ 't
a)
vo
t)
O ,
t)
<Co
X
0. ?
-q.
0
75
0 u
cl 0.
0 0 c
t) cz cz
4o 0-
4-& t o& o
CLC~c
cz
0 r
a. oo1~*C'
>0 0 'C v,35;
4-& c
o v

0 x
a) >j t)
Ec,4-& 3" in oo H C,
or
W
C) :0
t) ~ o
0 a

t) ,
t5 t) oc ~ -
CL W Ee
-4--

o CL 4-&
c(J)?
o
o

0 E
00
r.c r ~ C',
75 cl 4C.
E;
o t)
CC
0d ~
t) L -~
0
as

0 g ar.
r E?f
= u

e:
v n
05 0
0 0 0 0
%o

C42
.-
&
l
MscY
E
?0 .E
E
E
E

0
EE 20
CL
u Pt%
->
U.

0
W Lv

165

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 0

"o t)at

> 4-4
0
ob
E Zie
~0 E

v, Q) 0
0,k
40.
02

00(S .o o
2
U4 U'

00
C, H
0-0
2
tU M l H

cn cnOr$., D 0
o 0
E E 4-&CL s CL-1
$- E
cz tto 1

O
C ~ to -10 5
0 U' ~ 0 c > U 0

0. u .G
>%
-1

U
72

co :C: 450 *5
e
E
00U

C ;; E 00 U'

0 0
30cn 4.
'3
0
2 (

co u n
F w c2
E
04C. 0
Ct)
c
o 60, oto~ o 03
0 Z0 0 l
C., UE E~~.~

Q) 0

WQ

~L cn u t,4 u"
cz cn
3 n u dv
q3 v

166

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICS
AND PEDAGOGY 167

1. Tutor asks a question.

2. Student answers the question.
3. Tutor gives feedback on the answer.
4. Tutorand studentcollaborativelyimprovethe quality of (or embellish) the
answer.
5. Tutor assesses student's understandingof the answer.

An example of a five-step dialogue frame from our corpus is shown in Example

1. The studentwas being tutoredon the concept of factorialdesigns in a research
methods tutoringsession:

Example 1:
Step 1
1.1 Tutor: So, how manyF scoreswouldbe computed?
Step 2
1.2 Student:Three.
Step 3
1.3 Tutor: Three[agreeingwith the student].
Step 4
1.4 Tutor: Andwhatnumbers[referring to a matrixof cell means]wouldyou
use?
1.5 Student:You woulddo one for humor[one of the independent variables].
1.6 Tutor: And whatdoes thattell you?
1.7 Student:I'm not sure[laughs].
1.8 Tutor: OK, why do you do an F score?Whatis an F score?
1.9 Student:To see the size, uh, significance?
1.10 Tutor: The size of the significance.
1.11 Student:The size of the significance.
1.12 Tutor: Right,how statisticallysignificanta variableis.
1.13 Student:Right.
1.14 Tutor: So, you areright,youwouldhavethree[F scores]:oneforcaffeine,
one for humor,andone for ... ?
1.15 Student: The scores ... from caffeine and humor?
1.16 Tutor: Interaction,
the interactionof the two, right.
1.17 Student:Um hmm.
1.18 Tutor: [Explainsindependence of maineffects]
Step 5
1.19 Tutor: Do you see whatI'm saying?
1.20 Student:Um hmm.

There is some flexibility in the implementationof this five-step frame. For

example, a step may consist of more than one turn. In Example 1, Step 4 starts
at 1.4 and continues through 1.18. It is also possible that steps may be omitted.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
168 PERSON,KREUZ,ZWAAN,GRAESSER

Forexample,Step3 maynot occur-that is, the tutormaychoosenotto supply

feedbackon the student'sanswer.
The heartof the tutoringprocesscan be foundin the last two steps of this
dialogueframe.DuringStep4, thetutorandstudentelaborateon theideasraised
in the earliersteps.In a classroom,however,this elaborationis less likely to
occuror to be individuallytailoredto a particular student.In addition,Step 5
allowsthe tutorto monitorcloselythe student'sunderstanding. Onceagain,this
monitoringis less likelyto occurin a classroomsetting.
Mehan(1979)reported thatonlythefirstthreestepsoccurin normalclassroom
instruction.Giventheadvantages of tutoringoverclassroominstruction, it seems
reasonableto assumethatSteps4 and5 are especiallyimportant in correcting
andrepairing knowledgedeficits.In particular, it is duringStep4 thatthetutoring
sessionbecomestrulycollaborative. Thatis, bothtutorand studentcontribute
crucialpieces of information to arriveat the correctsolution(Graesser,1993b;
Graesser& Person,1994;fora relatedpoint,see Resnick,Salmon,Zeitz,Wathen,
& Holowchak,1993). In Step 5, however,relativelylittle informationis ex-
changed;the tutorattemptsto assess the student'sgraspof the topic at hand.
This relianceon the student'sself-assessmentis problematic, becausestudents
arerarelyableto calibratetheirown comprehension (Chi,Bassok,Lewis,Rei-
& &
mann, Glaser,1989;Epstein,Glenberg, Bradley,1984;Glenberg, Wilkinson,
& Epstein,1982;Person,Graesser,Magliano,& Kreuz,1994;Weaver,1990).
In the remainder of this article,we showhow Grice's(1975, 1978) conver-
sationalrulesandP. BrownandLevinson's(1987)politenessstrategiespositively
andnegativelyaffectthepedagogicalprocessin thefive-steptutoringframe.We
drawexamplesfromtwodifferenttutoringdomains,whicharedescribedin detail
in the next section.

TWOTUTORINGSAMPLES

We examinedtwo differenttutoringcorpora:college studentslearningresearch

methodsandseventhgraderslearningalgebra.Wechosethesecorporaforvarious
reasons.First,thesecorporawereanalyzedpreviouslyin the contextof question
askingandanswering(Graesser,1993a,1993b;Graesser,Person,& Huber,1992,
1993).Second,thesecorporaareexamplesof cross-agetutoring,whichis theform
of tutoringcommonin most school systems(Cohenet al., 1982;Fitz-Gibbon,
1977).Third,thesecorporaaredrawnfromverydifferenttutoringdomains;the
algebratutoringcan be thoughtof as a relativelyclosed-worlddomain(Collins,
Warnock,Aeillo, & Miller,1975;Fox, 1993)comparedwithresearchmethods
tutoring,whichhasless specifiedboundaries.Itmaybe thecasethatthepoliteness
strategiespreviouslydescribedmayoperatedifferently,dependingon thetypeof
tutoringdomain.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 AND PEDAGOGY
PRAGMATICS 169

Sample1:CollegeStudentsLearningResearchMethods

Students and tutors. Tutoringprotocolswere collectedfrom 27 under-

graduatestudents enrolled in a psychology researchmethods class at the Uni-
versity of Memphis. All studentsparticipatedin the tutoringsessions in orderto
fulfill a course requirement.The tutoring protocols, therefore, were a repre-
sentativecollege sampleratherthana samplerestrictedto studentswho were
havingdifficulty.
The tutorsweresix psychologygraduatestudentswho hadeach receivedan
A in both an undergraduate and a graduateresearchmethodscourse.Eachof
these graduatestudentshad tutoredon a few occasionspriorto this studybut
not in the areaof researchmethods.

Learningmaterials. The courseinstructor selectedsix topicsthatarenor-

mally troublesome for in
students a researchmethods course.A list was prepared
for eachof the six topics;threeto five relevantsubtopicswerealso includedin
the list. The tutorswere instructedto cover these topics and subtopicsin the
tutoringsessions.The topicsandsubtopicswereas follows:

Variables: Operational definitions,typesof scales,andvaluesof variables.

Graphs: Frequency distributions, plottingmeans,andhistograms.
Statistics: Decisionmatrices,TypeI andTypeII errors,t tests,andprob-
abilities.
Hypothesistesting:Formulating controlgroups,
a hypothesis,practicalconstraints,
design,andstatisticalanalyses.
Factorialdesigns: Independent variables,dependentvariables,statistics,mainef-
fects, cells, andinteractions.
Interactions: Independent variables,maineffects,typesof interactions,and
statisticalsignificance.

The students were exposed to the material on two occasions prior to their
participationin the tutoringsessions. First, each topic was covered in a lecture
by the instructorbefore the topic was covered in the tutoring session. Second,
each student was required to read specific pages in a research methods text
(Methods in Behavioral Research, Cozby, 1989) prior to the tutoring session.
This ensured that the students would have some familiaritywith the topics and
provided more common ground for the tutoringsession.

Procedure. Each studentparticipatedin two types of tutoringsessions: un-

structured
(or "normal") tutoringsessions.This
tutoringsessionsandstructured
to
included
manipulation, addressother issues
pedagogical (see Graesser,1993a),
is not addressedin this article.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
170 PERSON,KREUZ,ZWAAN,GRAESSER

Threeof the tutorswere assignedto the unstructured tutoringsessions.In

these sessions,the tutorswerenot given a specificformatto follow. Whena
studententeredthe tutoringroom,the studentwas instructed to sit in view of a
cameraandto readaloudthe list of topics.Thetutoringsessionthenproceeded
in the directionthatthe tutorandstudentsaw fit.
The otherthreetutorswere assignedto the structured tutoringsessions.In
thesetutoringsessions,tutorsandstudentsworkedthrougha setof predetermined
problemsrelevantto the topicsandsubtopicsprovidedon theirlist. All six of
thetutorswerealsoinstructed to encouragestudentquestionsandto avoidsimply
lecturing to thestudent.Each tutoringsessionwasvideotapedandlastedapproxi-
mately 60 min.
Eachof the 27 studentsparticipated in fourtutoringsessions.A counterbal-
ancing scheme was designed so that(a) studentneverhadthesametutortwice,
a
(b) each tutorcovered all six topics,(c) eachtutorwas assignedto 18 tutoring
and
sessions, (d) a student participated in two unstructuredtutoringsessionsand
two structured tutoring sessions. Therefore, each tutorinstructedthreestudents
on each of the six topics,whichyielded 108 tutoringsessions.Twenty-fiveof
the tutoringsessionscouldnot be transcribed due to audioproblems.

Sample2: SeventhGradersLearning
Algebra

Studentsand tutors. Thissampleconsistedof 22 tutoringsessionsin which

high school studentstutoredseventhgradersenrolledat a middleschool in
Memphis.This corpusincludedall of the algebratutoringsessionsfrom the
schoolduringa 1-monthperiod.Algebrateachersidentified13 seventhgraders
who werehavingdifficultyin theircourses.Thetutorswere10 localhighschool
studentswho normallyprovidedtutoringservicesfor the middleschool. On
average,thetutorshad9 hrof prioralgebratutoringexperience.Liketheresearch
methodssample,the algebrasamplewas exampleof cross-agetutoring.This
an
sample,however,differedfromthe researchmethodssamplein thatthe seventh
in thetutoringsessionsin orderto remediate
gradersparticipated theirknowledge
deficitsratherthanfor coursecredit.

Tutoringtopics and sessions. Most of the tutoringsessionsfocusedon

threetopicsthatare frequentlytroublesometo seventh-gradealgebrastudents.
These included(a) positiveand negativenumbers,(b) fractions,and (c) con-
structingalgebraicequationsfromwordproblems.Thesetopicsweretypically
associatedwith homeworkproblems,examinationitems, or a chapterin the
algebratext.Tutorsandstudentsfrequentlyreferredto this materialduringthe
tutoringsessions.The tutoringsessionslastedapproximately45 min, whichis
roughly comparableto the researchmethods sessions. Each session was video-
taped by a researchassistantfrom the University of Memphis.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICSAND PEDAGOGY 171

andCodingof the Tutoring

Transcription Sessions
Transcribersreceived a 1-hrtrainingsession on how to transcribethe videotapes.
They were instructedto transcribethe entiretutoringsessions verbatim,including
all "ums,""ahs,"word fragments,brokensentences, and pauses. The transcribers
specified whether an utterancewas made by the student or tutor. In addition,
transcribersalso noted messages that appearedon the markerboard, hand ges-
tures,head nods, and simultaneousspeech acts thatoccurredbetween the student
and tutor. Each transcriptionwas verified for accuracy by a research assistant
who spot-checkedrandom segments of each of the videotapes.

FROMTHEPROTOCOLS
EXAMPLES

In this section, we provide examples of the positive and negative consequences

of the conversationalrules and politeness strategies in the tutoringcorpora de-
scribed earlier. The following paragraphsillustrate the use of these rules and
strategies within the five-step dialogue frame proposed by Graesserand Person
(1994).

Step 1:TutorAsksa Question

Typically, a tutoring exchange is initiated by a question posed by the tutor.
Depending on the tutoring domain, the question may be relatively open-ended
or relatively constrained.The topics addressed by the tutors in the research
methods sessions lacked the specificity of the topics addressed in the algebra
tutoring sessions. Example 2 shows a typical topic from the research methods
corpus:

Example 2:
2.1 Tutor: OK. All right.So, we've specifiedour hypothesis.OK.Whatelse
do we need to do beforewe performa t testor an F test?

This can be contrastedwith a Step 1 question from the algebracorpus:

Example 3:
3.1 Tutor: Let'stryanotherone. Ah,numbereight,thenumberof seatson the
new 525 airlineris a 36%increaseover the old model.The new
planeseats374 passengers.
Howmanypassengers didtheold model
seat?

In Example 2, the tutor gave the student a great deal of latitude in specifying
how to move from a hypothesis to a statisticalanalysis. For example, the student
had the option to declare a populationor a sample or to define operationallythe

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
172 PERSON,KREUZ,ZWAAN,GRAESSER

dependentmeasure.In Example3, however,the tutorwas requestinga numeric

responsethatthe studenthad to generateby applyingthe appropriate equation.
Appropriate answersweremuchmoreconstrained in Example3 thanin Example
2. Althoughit oversimplifiesmatters,hereafterwe assumethatalgebrais a more
constrained,closed-worlddomainthanis the researchmethodsdomain.
The conversational rules andpolitenessstrategiesprovidedin Table 1 both
facilitateandinhibitthe tutoringprocess.Forexample,at this earlystagein the
five-stepframe,the tutormay attemptto putthe studentat ease by minimizing
the impositionof his or herquestion.Thisis demonstratedin Example4, which
was drawnfroma researchmethodssessionon factorialdesigns:

Example4:
4.1 Tutor: Youcantell mea littlebitaboutthereasonsforusinganexperiment
withmorethantwo levelsof an independent variable.

In fact, Example4 illustratesat least threepolitenessstrategiesat worksimul-

taneously:The tutorwas beingoptimistic("Youcan tell me"),understating the
request("a little bit"),and minimizingthe impositionof the request("a little
bit").By doingall thesethings,the tutorwas facilitatingthe student'sresponse.
The tutorwas indicatingthatthe studentknew the answer,and the tutorwas
tellingthe studentthateven a minimalresponsewouldbe appropriate. On the
otherhand,the tutor'srequestmayhavebeenunclear(howmuchdoes the tutor
wantto know?),andthetutormayhavesetthestudentupforfailure(byassuming
thatthe studentwouldbe able to providethe requestedinformation).
Sometimesthetutorandstudentmustnegotiatein orderto producea question
thatthe studentcananswer.Graesser(1993a,1993b)documented that,whenthe
tutorasksmultiplequestions,they becomeprogressivelymoresimpleuntilthe
studentcan providea response.As a result,the studentultimatelyanswersa
questionthathasbeengreatlysimplified.This canbe seen in Example5, drawn
froma researchmethodssessionon variables:

Example 5:
5.1 Tutor: Whattypeof scalewouldthatbe?
5.2 Student:Oh, let me think,whichone. I don'tknow.
5.3 Tutor: Tryto think.Nominalor... ?
5.4 Student:
Ordinal,
yeah.

In Example5, the studenthadansweredcorrectlybutonly afterthe numberof

possibleanswershadbeenreducedby the tutor.Thisexamplealso demonstrates
the use of ellipsis and the giving of hints to simplifythe question.If these
strategiesoversimplifythe tutor'squestions,a studentmay neverbe challenged
to answerquestionson the frontierof his or herknowledge.
Tutorsoccasionallypose questionsthatareunclearor vague, so thatthe student
may encounterdifficulty in attemptingto provide an answer. In such cases, the

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICSAND PEDAGOGY 173

tutormay apologize, as illustratedin Example6, drawnfrom a researchmethods

session on formulatinga hypothesis:

Example6:
6.1 Tutor: Whatis the effect of no limitedrelationversuslimitedrelation?
6.2 Student:Whatdo you mean?
6.3 Tutor: I'm sorry.I knewI was askingthe wrongquestion.I've gottabe
careful... um,there'ssomethingthatmakesa correlationalmethod
not so wonderful[...] andsomethingthatmakestheexperimental
methodwonderful.Do you knowwhatthatwouldbe?

The tutor apologized to acknowledge explicitly the incorrect and potentially

confusing question at 6.1. By apologizing, the tutordemonstratedher sensitivity
to the student'sneed for clarity.There would seem to be little cost in apologizing
for a mistake on the partof the tutor. If the tutorcontinuedto do this, however,
the studentmay have begun to doubt the credibilityof the tutor.

Step2: StudentAnswersthe Question

During Step 2, the studentmakes his or her initial attemptto answerthe tutor's
question. By definition, the tutor plays a relatively passive role, although much
of the tutor's effort during Step 1 is directedat constructinga question that the
student can answer. Clearly, the student's answer will affect the feedback that
the tutorprovides (see Griffin & Humphrey,1978). Our focus here, however, is
how tutors use conversationalrules and politeness strategies.Because the tutor
does not participatein Step 2, we do not discuss this step in furtherdetail.

Step 3: TutorGivesFeedbackon the Answer

Clear,discriminating,and accuratefeedback by the tutor is presumablyessential
for effective tutoring.Previousresearch,however, has shown thattutorsprovide,
with roughly equal likelihood, both positive and negative feedback to students'
error-riddenanswers(Graesser,1993b). Vague answerson the partof the student
are normally met with positive feedback from the tutor rather than negative
feedback (Graesser, 1993b). Why do tutors provide positive feedback in these
cases? It seems likely that tutors avoid negative feedback as much as possible,
because it is very face threateningfor the student.Consequently,this convention
of normalconversationmay inhibit the effectiveness of the tutoringprocess.
Example 7 illustratesinappropriatefeedback by the tutor, drawn from a re-
search methods session on variables:

Example7:
7.1 Tutor: Whatis an inferentialstatistic?

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
174 PERSON,KREUZ,ZWAAN,GRAESSER

7.2 Student:A statisticthatgives you informationaboutsomethingthat'sgonna

happen. Or couldhappen,possibly.
7.3 Tutor: Also,somethingthat'skindof implicitwithinferentialstatisticsthat
peopletendnot to thinkaboutis [...] thatif peoplewentout and
surveyedand surveyedthatthey'dget the sameinformation as a
statistic....

At 7.2, the student's answer was incorrect. He seemed to have confused the
conceptsof predictabilityand generalizability.At 7.3, the tutorviolatedthe maxim
of quality (i.e., assuming that the tutor realized that the student's answer was
incorrect).We would argue that the tutor should violate the maxims of quantity
and mannerin order to provide effective feedback. That is, the tutor needs to
provide much more information,because the student's answer is error-ridden.
Perhapsthe tutorfailed to provideappropriatefeedbackbecause she was attempt-
ing to avoid disagreementwith the student. The student, however, may have
mistakenlybelieved thathis answerwas correct,because the tutorused the word
also, ratherthan no at 7.3. A more appropriateanswer at 7.3 would have been,
"No, inferentialstatistics refers to the ability to generalize to a populationfrom
a sample and not what may happen in the future."
We can contrast Example 7 with Example 8 (from an algebra session on
fractions), in which the tutor gave appropriatefeedback that addressedthe stu-
dent's misconception:

Example8.
8.1 Tutor: Let's trythis one:5/lsthsminusnegative3/6ths.
8.2 Student:Uh, thisone wouldjustgo like that[pointingto previouslyworked
problem]?
8.3 Tutor: Well,um, actually,no, no, you couldn'tdo that.Sorry.

The feedback in Example 8 was more appropriate,althoughpoliteness,strate-

gies were still being employed:The tutorapologized(perhapsallowing the student
to save face after an error).
When a student commits an error in a tutoring session, the tutor has the
responsibilityto acknowledge and correct the error.Because, by definition, this
is a face-threateningact, it seems likely that the politeness strategies will be
employed to make the feedback less aversive. In Example 9, drawn from a
research methods session on graphing, a student had constructeda frequency
polygon but had not labeled the axes. The tutor attemptedto make the student
realize this on her own. Because he chose to be conventionallyindirect,however,
this requireda numberof turns to accomplish:

Example9:
9.1 Tutor: OK, you've got the right numbers.
9.2 Student:Yeah,OK, I just needed...

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 ANDPEDAGOGY 175
PRAGMATICS

9.3 Tutor: OK,is thereanythingelse aboutthisgraphthatyou wouldwantto

do beforeyou finish?I mean,do you considerthis finished?
9.4 Student:I guess I could,I coulddrawa line ...
9.5 Tutor: Whatif you walkedup to this graphandyou hadneverseen [it]?
9.6 Student:Oh, I guess I shouldlabelthese [pointsto the axes].
9.7 Tutor: Yeah,yeah.

Although some might argue that studentsshould be encouragedto discover such

problems on their own, in this example the studentfocused on a less important
error in her graph (i.e., drawing a line to connect the points on the graph). By
being conventionally indirect, the tutor created ambiguity and took time away
from correctingthe more importantproblem.
The problem of being too indirecthas been noted in other domains. In their
study of Swedish allergists, Aronsson and Rundstr6m(1989) mentioned that
allergists must frequentlyask patients to remove their clothes. Clearly, this is a
face-threateningact; and the allergists, as expected, made their request in very
indirect ways. This led to confusion on the patients' part, because they were
typically left unsure whetherto remove their clothes at all or how much clothing
to remove.
As we have seen previously, thereare examples in this step in which the tutor
used multiplepoliteness strategiesto addressstudent'serrors.Example 10, drawn
from an algebrasession on variables,shows the simultaneoususe of two negative
politeness strategies:be conventionallyindirectand state the face-threateningact
as a general rule:

Example10:
10.1 Tutor: ... OK,now, whatit is, justFOIL.OK,FOIL.It standsfor "first,
outside,inside,last."OK, so whatyou do is you take [the]first
one, right?You multiplythesetwo, andyou takethe outside,the
inside,andthe last.Do you see how thatworks?
10.2 Student:Here'sthewayI'vedoneit [studentmumblessolutionto theproblem
fromthe bookto himself].
10.3 Tutor: Right.Well,see that'sone way to do it, buttheylikethis [theFOIL
method];thisis reallytheway mostpeoplelike to do it [elaborates
on reasons].

At 10.3, the tutor erroneously told the student that his method works, when in
fact it does not. This is typical of how a tutorcorrectsa problem,albeit indirectly.
First, the tutoragrees with the studentbut then goes on to qualify this agreement
(in this case, by stating the face-threateningact as a general rule). This is prob-
lematic, because the student may, nonetheless, continue to cling to his or her
misconception.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
176 PERSON,
KREUZ, GRAESSER
ZWAAN,

Step 4: TutorandStudentCollaboratively the

Improve
Qualityof the Answer
Studentstypicallyprovideveryabbreviated answersto questions.Tutors,there-
fore, find it to
necessary encourage them to elaboratetheirresponses(Graesser,
1993a, 1993b; Graesser & Person,1994). As a result,the construction
of an
answercan be thoughtof as a collaborativeactivityin whichthe studentand
tutorworktogetherto constructan acceptableanswer(Graesser,1993a,1993b;
Kreuz& Roberts,1993;Resnicket al., 1993).
Thetutorcanuse severalstrategiesto facilitatethisprocess.Forexample,the
tutormay providea hint, ask an embeddedquestion,or employa scaffolding
techniquein orderto increasethe likelihoodof a student'sresponse.Whenused
to extreme,however,thesestrategiesencouragepassivelearning.Example11,
drawnfrom a researchmethodssession on interactions,shows an embedded
questionat 11.3.
Example11:
11.1 Tutor: ... All right,let's tryanotherone. Let'ssay we had,oh, wait.First
of all, let's translatethis in termsof ourexperiment.
11.2 Student:OK.
11.3 Tutor: OK.Whatdoes this meanas far as typeof drug?
11.4 Student:It hadno effect.
11.5 Tutor: In otherwords,using cornflakes... wereno differentthanusing
moodflakes. Therefore,moodflakesmustnothavebeena realdrug.
All right!Goodjob.

Example 11 shows some of the costs associated with encouragingthe student's

responses.At 11.5, the tutorexaggeratedby providingexcessive praiseeven
thoughthe student'sinputwas minimal.Onceagain,use of oversimplified em-
beddedquestions,combinedwithexcessivepraise,mayleadto a relativelypas-
sive role on the partof the student.
The tutor sometimes contributesinformationthat confuses the student, as in
Example 12. The various types of scales (nominal, ordinal, interval, and ratio)
were being discussed in a researchmethods session on variables:

Example12:
12.1 Tutor: OK, the way I rememberthat,um, whenI was tryingto learnit
was... I knewin Frenchthewordblackis noir,N-O-I-R,andyou
can remember it thatway.
12.2 Student:Well, yeah,yeah.
12.3 Tutor: So you can thinkof noir,N-O-I-R.
12.4 Student:So whatdoes that,whatdoes blackhaveto do with nominal?

Here is a mnemonic that has gone sadly awry. The tutor attemptedto provide
the student with a memory aid, but she did not make sufficiently clear how it

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICS
ANDPEDAGOGY 177

appliedto thenamesof thescales.Theconfusionarosebecausethetutorpre-

supposed commonground-thatis, thatthestudentknewthattherelationship
betweenthespellingof noirandthenamesof thescaleswasarbitrary.
Tutorsfrequently
dothestudents' workforthemin a tutoring
session.Tutors
often ask questionsthatprovidetoo muchinformation(a violationof Grice's
[1975]maximof quantity),
whichlessensthecognitiveburden
forthestudent.
Example13, drawnfroman algebratutoringsessionon fractions,illustrates
this:

Example13:
13.1 Tutor: And you wantto multiplythatby ... ?
13.2 Student:6, 42.
13.3 Tutor: Yeah.

In thisexample,thetutorprovidedthe operationinformation, andall the student

had to do was providethe numbers.One mightexpect this strategyto occur
duringthe earlystagesof a tutoringsession,whenthe studenthas notyet mas-
teredthe material.If the tutoradheredto this strategythroughoutthe tutoring
session,however,the misconceptions of the studentmightneverhavebeenad-
dressed,andthe studentwouldrely on the tutorto supplythe structureof the
dialogue.
A moreappropriate methodfor laterstagesin the tutoringsessionis shown
in Example14, in whichthe tutorandstudentwerediscussingthe next step in
computinga t test:

Example14:
14.1 Tutor: We aregoingto use the scores?
14.2 Student: Yeah.
14.3 Tutor: OK, What'sthe firstthingwe needto do?
14.4 Student:You haveto writethe scoresdown.It gets morecomplicated.
14.5 Tutor: OK, you would, we would need all of those scores. So, um, what
wouldwe do whenwe've got all the scores?
14.6 Student:Um, OK,you havea mean?
14.7 Tutor: OK....

In thisexample,the tutorneversaidmorethanthestudenthadalreadysaid.This
couldbe thoughtof as a violationof the maximof quantity,becausethe tutor's
arenotsupplyingnewinformation.
repetitions Thistechnique, however,promotes
activelearningby forcingthe studentto do mostof the work.
The student'sanswerin the nextexampleillustratesa violationof relevance,
becausethe studentfocusedon a relativelyminordetailin the tutor'squestion.
Example 15 is drawn from a researchmethods session on variables.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
178 PERSON,
KREUZ, GRAESSER
ZWAAN,

Example15:
15.1 Tutor: WhywouldNIMHnot give me five milliondollarsto do a corre-
lationalstudy?
15.2 Student:'Causeit's, well, that'sa lot of money.
15.3 Tutor: Say $1,500.It's asidefromthe moneyissue,if, OK...

At 15.3, the tutor dealt with this violation by explicitly redirectingthe student
away from the irrelevantaspect of the question.AlthoughExample 15 illustrates
a violation of relevance by the student, it is importantto note that the tutor
adheredto the maxim of relevance.
On the otherhand, tutorsmay need to violate the maxim of relevancein order
to introducematerialthat facilitates the student's understandingof the topic at
hand. The tutor may provide backgroundinformation,new examples, or alter-
native explanationsin orderto ensure student comprehension,even though they
may appearto be irrelevant.If, for example, a tutor proposes a confounding
variableto explain an experimentalresult, the new variablemay seem irrelevant
to the topic at hand, until the student realizes the underlyingrelationship.For
example, a tutormay ask the studentwhethertwo groupsof subjects were tested
at the same time of day. This question will appearirrelevantuntil the student
realizes that time of day may affect dependentmeasuressuch as reaction time.
Collins and his colleagues (1977, 1985; Stevens, Collins, & Goldin, 1982)
dissected the process of Socratictutoringas an importantpedagogicaltechnique.
In Socratictutoring,the tutordoes not correctan errorcommittedby the student;
rather,the tutor asks a sequence of carefully selected questions that expose the
student'smisconceptions.Interestingly,for the tutorto employ Socratictutoring,
the tutormust violate the maxim of quality, because the tutor,in effect, is acting
as if the erroris correct.
Example 16 illustratesSocratictutoringthat begins duringStep 2. It occurred
duringa discussionof how to design a studythatwould determinethe relationship
between divorced parents and depressed children. Ethically, this could only be
accomplishedby using a correlationalapproach.

Example16:
16.1 Tutor: Tell me first,ah, what kind of experimentwouldit be? I mean,
whatmethod?Wouldit be ... areyougoingto haveanexperimental
design?
16.2 Student:Yes.
16.3 Tutor: You are?
16.4 Student:Probablyso, um, becausenot all children,if they'redepressedare
going to be ... you'renot going to be ableto look at anydataon
depressedchildren.Well, I mean,let me startover.You can look
at dataon depressedchildrenandwhetheror not theirparentsare
divorced,butif you wantedto reallytestyourhypothesis,it would
be betterif you conductedan experimental researchdesign.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 ANDPEDAGOGY 179
PRAGMATICS

16.5 Tutor: OK, I'm curious,OK, now how wouldwe go aboutdoing that?
OK, this is all up to you.

During the next several turns,the studentstruggledto explain his answer, while
the tutor provided minimal input (e.g., "um hmm," "OK"). Socratic tutoring
requiresthatthe tutormaintainthe student'sfalse beliefs until it becomes obvious
to the student that these beliefs are false. After several such turns, the tutor
intervenes:

Example16 (continued):
16.11 Tutor. ... and,I don't,I couldbe wrong,butareyou manipulating
any-
thing?
16.12 Student:No, you'reabsolutelyright.No, I'm not.

Socratictutoringis rarein most tutoringinteractions,becausethis methodrequires

a high level of domain knowledge, as well as a great deal of tutoringexperience
on the part of the tutor (Collins et al., 1975). Socratic tutoring may also be
uncommonbecause it forces tutorsto violate normalconversationalrules: in this
case, the maxim of quality.
In other cases, Socratic tutoring involves violating the maxim of relevance,
because the tutorintroducesa seemingly irrelevantidea thatwill, ideally, redirect
the student'sline of thinking. Collins (1977), for example, provided an example
of a violation of relevance:

17.1 Tutor: Wherein NorthAmericado you thinkrice mightbe grown?

17.2 Student:Louisiana.
17.3 Tutor: Why there?
17.4 Student:Placeswherethereis a lot of water.I thinkricerequiresthe ability
to selectivelyflood fields.
17.5 Tutor: OK. Do you thinkthere'sa lot of rice in, say, Washingtonand
Oregon?
17.6 Student:Aha,I don'tthinkso.
17.7 Tutor: Why?
17.8 Student:There'sa lot of waterup theretoo, butthere'stwo reasons.First,
the climateisn't conducive,and second,I don't thinkthe land is
flat enough.You'vegot to haveflat landso you canflood a lot of
it, unlessyou terraceit. (p. 351)

In 17.3, the tutor asked a question to ensure that the student'sreasoning was
focused on the relevant steps in the causal chain for growing rice. That is, rice
needs to be flooded. In 17.4, the student stated that rice grows in "places with
a lot of water."The tutor immediatelychose counterexamples(Washingtonand
Oregon)thatmight have seemed irrelevantto the student.However,this statement
forced the studentto think of other causal factors besides water.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
180 PERSON,KREUZ,ZWAAN,GRAESSER

In the classroom, teachers may attempt to explain difficult concepts by ap-

pealing to a conceptual metaphor(i.e., the atom is like the solar system). One
advantageto this approachis that a difficult concept may be simplified for the
students.The metaphor,however, may lead to even more severe misconceptions.
Feltovich, Spiro,andCoulson (1989) showed thatteachersoften use inappropriate
metaphorsto explain difficult concepts. We have found that a similar problem
occurs duringtutoring.In Example 18, the tutorand the studentwere discussing
main effects and interactions.This tutor frequentlychose to use a metaphorin
order to determinethe presence of a main effect in a graph. The tutor would
suggest that the student visually collapse together the lines in the graph;if the
resultingline had a slope, this impliedthe presenceof a main effect. This method
was referredto as the "squish"metaphor:

Example 18:
18.1 Tutor: Buttherearesomeneattricksto beingableto figureoutgraphically
[whethera maineffectof a variableis depicted].... Well,now we
can get fromhere [cell means]to a graph,right?
18.2 Student:Um hum.
18.3 Tutor: 'Causeonceyou'reat thegraph,it's realeasyto figureoutif there's
a maineffect for A, a maineffect for B, andan interaction.
18.4 Student:If they'reparallel,well ... If they'reparallel,there'sno interaction.
18.5 Tutor: ... UUm,we would do what's called collapsing the two lines. I call
it kindasquishing'em.... Whatyou woulddo, if thisline is hori-
zontal[pointsto new, squishedline] ... we wouldsay thatthereis
no maineffectforA. But sinceit's nothorizontal andis at a certain
angle where one end is different
from the otherendpoint,thenyou
can say thatthereis a maineffect for A. OK,let's see, how about
this [drawstwo lines on a graphthatdepicta maineffect for the
A variable]?Is therea maineffect for A?
18.6 Student:No.

The student's answer was incorrect(there is a main effect for A), even though
the tutorhad provideda supposedlyhelpful method for determiningthe answer.
Does this mean thattutorsshouldalways avoid the use of metaphors?Example
19, drawnfrom a session on Type I and Type II errorsin the researchmethods
tutoringcorpus, demonstratesthe utility of a conceptualmetaphor:

Example 19:
19.1 Tutor: You don'tsee it, butit's there[pointingto a decisionmatrixin the
text].Hereyou see it, andit's not there.... Theway I, I'll tell you
the way I remember it. A TypeI erroris like um ... you'rehallu-
cinating...
19.2 Student:Um hum.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 ANDPEDAGOGY 181
PRAGMATICS

19.3 Tutor: ... you see somethingthat'snot there.TypeII erroris like you're
blind.It's there,but you don'tsee it, andthat'sthe way I always
rememberthese[laughs].
19.4 Student:Oh, so let me writethesedown,hallucination andblindness.

The tutor and student spent the next several turns working throughan example
to determine whether a Type I or a Type II error was present. It is interesting
to point out that, later in this discussion, the following exchange occurred:

Example19 (continued):
19.19 Tutor: Um, does thatmakeit a littleclearer?
19.20 Student:Yes.
19.21 Tutor: OK....
19.22 Student:... becauseyou reallyneed somethinglike that.., analogies,
yeah.

Clearly,some metaphorswork betterthan others.Tutorsmustjudiciously choose

those metaphorsthat are clear, helpful, and accurateand avoid metaphorsthat
do not satisfy these criteria.
Tutors and students may elect to introducea new concept by stating it as a
general rule. This has the beneficial effect of minimizing the imposition of the
new informationon the hearer.In other words, insteadof saying, "Do this," the
speaker says, "Most people do this." Example 20, from a research methods
session on interactions,illustratesthe studentoffering a contributionindirectly.
The tutorhad forgottenan importantprinciple,in this case, the numberof intervals
on a Likert scale.

Example 20:
20.1 Tutor: Andourlevelof psychotic.Let'ssay we havea 10-pointscalefor
that.
20.2 Student:OK.
20.3 Tutor: So ...
20.4 Student:Whathappenedto the magicsix?! [laughs]
20.5 Tutor: Oh, OK.
20.6 Student:[Nameof instructor's] magicsix!
20.7 Tutor: The magicsix. OK,I forgotthe magicsix.
20.8 Student:Great.
20.9 Tutor: So, let's say we have a 6-pointscale.
20.10 Student:OK.

By invoking the name of the instructor,the student provided a face-saving ra-

tionale for the use of a 6-point scale. Instead of directly questioning the tutor's
selection of a 10-pointscale, the studentwas able to expressher opinion indirectly
in the form of a general rule.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
182 KREUZ,
PERSON, GRAESSER
ZWAAN,

It should be clear that Step 4 is a collaborativeprocess: The tutorand student

work togetherto constructan answer.In addition,violations of the conversational
rules and politeness strategiesoccur. Some violations are made by the student,
and some are made by the tutor.As we have shown, some tutorviolations may
be desirable for effective tutoring.

Step 5: TutorAssesses Student'sUnderstanding

of the Answer
In this step, the tutorcould encouragethe studentto identifyany specific problems
that still remainat this stage in the tutoringprocess. Instead,tutorstypically ask
very general, yes-no questions that do not tap the student's misconceptions.For
example, tutors typically ask:

Example21:
21.1 Tutor: And thenyou'reOK.Are you with me?

or

Example22:
22.1 Tutor: ... so you understand
that,right?

In other words, the tutorsare adheringto Grice's (1975) maxim of quantityeven

when they should not. That is, the tutors make global statements about the
student's comprehensioninstead of asking questions regardingspecific issues
that have alreadybeen addressed.This problem is very common in the tutoring
corpus and suggests an overreliance on students' self-assessment. In normal
conversation, it is assumed that individualsare accuratein assessing their own
knowledge (e.g., what they ate for lunch and to whom they are married),and
listeners do not challenge these reports. In tutoring, however, the student is
operatingon the frontierof his or her knowledge, and self-assessments may be
much less accurate.Therefore, the tutor should violate the maxim of quantity
frequently.
A much better approachappears in Example 23, drawn from a session on
constructingalgebraicequationsfrom word problems:

Example23:
23.1 Tutor: Do you haveanyproblemwiththesekindsof wordproblems[re-
ferringto a sectionin the book]?Wherethey say-
Ah, not really.
23.2 Student:[interrupts]
23.3 Tutor: You don't?You don't?You don'thave any troublewiththat?
23.4 Student:No.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 ANDPEDAGOGY 183
PRAGMATICS

23.5 Tutor: Let'sjust do one of them.Um, Dan earned$56, whichwas twice

morethanwhatJimearned.Now you'resupposedto writeanequa-
tion.
23.6 Student:Ah, I can'twritetheequations.

The tutordiscoveredthe student'sdeficits only by repeatedlyqueryingthe student.

Even this is insufficient;it was not until the student was challenged to perform
that the deficit was made manifest.Example 23 includes many violations of the
politeness strategies (avoid disagreement,be optimistic, and minimize imposi-
tion), but these violations were necessary to expose the student's deficits.
Sometimes a lack of common ground between the student and tutor may
adversely affect the tutoringinteraction.Specifically, the tutormay erroneously
assume that the student possesses informationthat the student does not. This
presuppositionof common ground can be seen clearly in Example 24, drawn
from an algebrasession on word problems:

Example 24:
24.1 Tutor: Now thatyou've workedthem,let's try number14. It's a little
differentonebut,ah,it's a lot liketheothertwo.A bottleof Produce
Timeapplejuicecontains64 ouncesandcosts99 cents.FarmFresh
juice,availableinbottlesthatcontainonegallon,foroneeighty-eight
[$1.88]each;ah, whichis the betterbuy?
24.2 Student:How manyounces,um, arein a gallon?
24.3 Tutor: Ouncespergallon,good question.You haven'thadthesein tables
before....

In this example, the tutorpresupposedthat the studentknew how many ounces

are in a gallon. The studentdid not know, however, and asked the tutor for the
information.It is more typical,however,for a studentto hide his or herknowledge
deficits from the tutor, leading to a breakdown in effective tutoring. Tutors,
therefore, should exercise caution when they make presuppositionsabout what
the tutor and studentboth know.
Even when a concept or idea has been explicitly mentioned in a tutoring
session, the tutorcannotbe certainthat the studentboth understandsand remem-
bers the information.This can be contrastedwith normalconversationin which
contributionsby both participantsare assumed to be in the common groundand
completely understood (Clark & Schaefer, 1989). Tutors, therefore, must be
careful not to carry over this conversationalassumptioninto tutoringsessions.
Tutors occasionally preparestudents for a difficult problem by being pessi-
mistic aboutwhetherthe studentcan solve the problem.Example25, drawnfrom
a session on variables,illustratesthis:

Example 25:
25.1 Tutor: OK, thisone is probablya littleharderthanthe firstone.
25.2 Student:Yeah[laughs].

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
184 PERSON, KREUZ, ZWAAN, GRAESSER

This methodmay have unwantedconsequences.For example,it may lead the

studentto believethatfailureis expectedor even acceptable.This may lead to
a diminutionof efforton the partof the student.

CONCLUSIONS

We have suggestedthattutoringcan be examinedby employingthe strategies

andmaximsthatcharacterize normalconversation. Wearenotclaiming,however,
that tutorialand conversational dialogueare the same.Discourseexists on a
continuum, withinteractive,normalconversation at one end andless interactive
discourse(e.g., classroomlecturesandspeeches)at theother.Tutoringdiscourse
clearlyfalls somewherebetweenthese two extremesand probablyresembles
conversation moreclosely thanclassroomdiscourse.If this claim is true,it is
notsurprising thattutorsrelyon the implicitprinciplesof ordinaryconversation.
As we haveshown,Grice's(1975, 1978)conversational rulesandP. Brownand
Levinson's(1987)politenessstrategiesaffectthetutoringprocessin positiveand
negativeways.Tutorsshould,therefore, be cognizantof thesecostsandbenefits,
becausesuchawarenessmay enhancethe overalleffectivenessof tutoring.
Some steps of the tutoringprocessare more vulnerableto conversational
misstepsthanothers.DuringStep 1, for example,the tutorand studentmust
negotiatea mutuallycomprehensible questionthatwill be expandedduringlater
steps.DuringStep4, thetutormustelaborateon thestudent'sanswerandaddress
the student'sknowledgedeficits.Thesetwo stagesare crucialfor the tutoring
process,butthereis a highprobability thatface-threatening acts mayoccur.As
a result,we foundmanyexamplesof Grice'sconversational rulesandP. Brown
andLevinson's(1987)politenessstrategiesin thesesteps.
Theremaybe functionaldifferencesin how theserulesandstrategiesoperate
in the tutoringdomain.Grice's(1975, 1978)analysisfocuseson the contentof
utterances (e.g., quantityandquality),whereasP. BrownandLevinson's(1987)
approach addresses the socialandinterpersonal dimensionsof discourse.Future
researchmayprofitablyexplorehow theserulesandstrategiesinteract,andsuch
a fine-grainedanalysismayofferadditional insightsintothispedagogicalprocess.

DomainDifferences
We foundthatthe researchmethodsandalgebratutoringsessionsdifferedfrom
each otherin severalways. In particular,the tutorsin the researchmethods
sessionsseemedto rely on the politenessstrategiesmorethandid the algebra
to differencesin the statusof the tutors
tutors.Thisfindingcannotbe attributed
acrossthe tutoringdomains,becausebothsampleswereexamplesof cross-age
1977).Thisdifferencealso cannotbe attributed
tutoring(Fitz-Gibbon, to differ-
ences in the expertise of the tutors, because most of the algebra tutors and all

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICSAND PEDAGOGY 185

of the researchmethodstutorshad neverpreviouslyservedas tutorsin their

respectivesubjects.This is, in fact, typicalof the tutoringthatoccursin most
schoolsettings(Cohenet al., 1982;Fitz-Gibbon,1977).
Otherdifferencesbetweentheresearchmethodsandalgebrasessionsmaybe
responsiblefor the observeddifferences.For example,we cannotrule out the
effectof agedifferencesbetweenthetwogroupsof tutors.We suggest,however,
thatthe differencesin politenessstrategyuse maybe attributable to the domains
of the tutoringsessions.The researchmethodsdomaincan be thoughtof as an
open-worlddomain:Thequestionsandanswersdo notexist withinwell-defined
parameters. In contrast,the algebradomaincan be considereda closed-world
domain:The questionsand answersare typicallywell-defined(Collinset al.,
1975), and one can readilydistinguishbetweengood and bad answers.For
example,it is far easierto elicit the answerto an algebrawordproblemthanit
is to elicitthe drawbacksof a correlationaldesign.Theanswersaredifferent,as
well, because there is no ambiguity associatedwith a numericresponse:The
studentwhosays,"Theansweris five,"canreceiveclear-cutfeedback.Incontrast,
the studentwho says, "It'sless powerful,"abouta correlational designmaynot
completelyunderstand the underlyingprinciples,andthe tutormustfollow up
on the student'svagueanswerto ensurehis or herunderstanding.
For thesereasons,the tutors(andstudents)in open-worlddomainsmayrely
heavilyon theconversational rulesandpolitenessstrategiesthatfacilitatenormal
conversation. As we have shown,however,theserulesandstrategiescan create
pedagogicalproblems,even whenthey areemployedto expeditenormalsocial
interaction.

GroundRulesinTutoring
Establishing
As manyresearchers have demonstrated,studentsmusthaveprerequisite infor-
mationin orderto profitfromaneducational experience(Gagn6,1977;VanLehn,
1987).Forexample,studentsrequirerelevantbackground knowledgein orderto
comprehendtextbookinformation(McKeown,Beck, Sinatra,& Loxterman,
1992). In a similarway, studentsmusthave an understanding of the tutoring
process beforea tutoring sessionbegins.Specifically,we believethattutorsand
studentsshouldestablishconversational ground rulespriorto the tutoringinter-
action.The studentshouldbe madeawarethatthe tutorwill use negativefeed-
back,thatthe "normal"rules of conversationmay be violated(e.g., the tutor
may say, "No,youransweris wrong"),andthatthe studentis expectedto take
a very activerole in the tutoringprocess.In this way, knowledgedeficitsmay
be moreeasily exposedandmoreeasilycorrected.The tutoringprocessshould
be moreefficient.
As mentionedearlier,most of the effort in Step 4 of the tutorialdialogue
frame (student and tutor collaboratively improve the quality of the answer) is
contributedby the tutor.We suggest thattutorsimplementstrategiesthat encour-

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
186 PERSON,KREUZ,ZWAAN, GRAESSER

age the active participationof the student ratherthan the tutor supplying most
of the information.A trulycollaborativeexchange duringStep 4 allows for more
active involvementon the part of the student, as well as more opportunitiesfor
the tutorto identify the student'sknowledge deficits. The reason these activities
do not frequentlyoccur duringStep 4 may be the overreliance,by both tutorand
student,on the conversationalrules and politeness stfategiesof normaldiscourse.
This overreliance on the rules and strategies of normal conversation also
creates a problem in Step 5, in which the tutor assesses the student's under-
standing.Because this assessmentcan be very face threateningfor students,tutors
often assume that, if the materialhas been covered during the tutoringsession,
it has been understoodby the student.We suggest that tutorsactively probe the
studentsin orderto expose knowledge deficits. If the tutorsexplicitly inform the
studentsthatthis will occur, the studentswill regardthis assessmentas less face
threatening.

ACKNOWLEDGMENTS

This researchwas fundedby grantsawardedto ArthurC. Graesserby the Office

of Naval Research(N00014-88-K-0110, N00014-90-J-1492, and N00014-92-
J-1826) and by a Center for Excellence grant awarded to the Departmentof
Psychology at the University of Memphis by the state of Tennessee.
We are indebted to John Cady for providing access to the seventh-grade
algebratutoringsessions. The commentsof two anonymousreviewers were also
very helpful.

REFERENCES

Aronsson, K., & Rundstrim,B. (1989). Cats, dogs, and sweets in the clinical negotiationof reality:
On politeness and coherence in pediatricdiscourse. Language and Society, 18, 483-504.
Bloom, B. S. (1984). The 2 sigma problem:The search for methodsof groupinstructionas effective
as one-to-one tutoring.EducationalResearcher, 13, 4-16.
Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language use. Cambridge,
England:CambridgeUniversityPress.
Chi, M., Bassok, M., Lewis, M., Reimann,P., & Glaser, P. (1989). Self-explanations:How students
study and use examples in learningto solve problems.Cognitive Science, 13, 145-182.
Clark, H. H., & Carlson, T. B. (1981). Context for comprehension.In J. Long & A. D. Baddeley
(Eds.), Attention and performance (Vol. 9, pp. 313-330). Hillsdale, NJ: Lawrence Erlbaum
Associates, Inc.
Clark, H. H., & Schaefer, E. F. (1987). Collaboratingon contributionsto conversations.Language
and CognitiveProcesses, 2, 19-41.
Clark, H. H., & Schaefer, E. F. (1989). Contributingto discourse. CognitiveScience, 13, 259-294.
Clark, H. H., Schreuder, R., & Buttrick, S. (1983). Common ground and the understandingof
demonstrativereference.Journal of Verbal Learningand VerbalBehavior, 22, 245-258.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATICSAND PEDAGOGY 187

Cohen, P. A., Kulik,J. A., & Kulik, C. C. (1982). Educationaloutcomes of tutoring:A meta-analysis
of findings. AmericanEducationalResearch Journal, 19, 237-248.
Collins, A. (1977). Processes in acquiring knowledge. In R. C. Anderson, R. J. Spiro, & W. E.
Montague (Eds.), Schooling and the acquisition of knowledge (pp. 339-363). Hillsdale, NJ:
Lawrence ErlbaumAssociates, Inc.
Collins, A. (1985). Teaching reasoning skills. In S. F. Chipman,J. W. Segal, & R. Glaser (Eds.),
Thinkingand learning skills (Vol. 2, pp. 579-586). Hillsdale, NJ: LawrenceErlbaumAssociates,
Inc.
Collins, A., Warnock, E. H., Aeillo, N., & Miller, M. L. (1975). Reasoning from incomplete
knowledge.In D. G. Bobrow & A. Collins (Eds.),Representationand understanding(pp. 453-494).
New York: Academic.
Cozby, P. C. (1989). Methods in behavioral research (4th ed.). MountainView, CA: Mayfield.
Craig, R. T., Trasy, K., & Spisak, F. (1986). The discourse of requests:Assessment of a politeness
approach.Human CommunicationResearch, 12, 437-468.
Epstein, W., Glenberg, A. M., & Bradley, M. M. (1984). Coactivation and comprehension:
Contributionof text variablesto the illusion of knowing. Memory & Cognition, 12, 355-360.
Feltovich, P. J., Spiro, R. J., & Coulson, R. L. (1989). The natureof conceptual understandingin
biomedicine: The deep structureof complex ideas and the developments of misconceptions. In
D. A. Evans & V. L. Patel (Eds.), Cognitive science in medicine: Biomedical modeling (pp.
113-172). Cambridge,MA: MIT Press.
Fitz-Gibbon, C. T. (1977). An analysis of the literature of cross-age tutoring. Washington, DC:
National Instituteof Education.(ERIC Document ReproductionService No. ED 148 807)
Fox, B. (1993). The human tutorial dialogue project. Hillsdale, NJ: Lawrence ErlbaumAssociates,
Inc.
Fraser, B. (1990). Perspectives on politeness. Journal of Pragmatics, 14, 219-236.
Gagnd, R. M. (1977). The conditions of learning (3rd ed.). New York: Holt, Rinehart& Winston.
Glenberg, A. M., Wilkinson, A. C., & Epstein, W. (1982). The illusion of knowing: Failure in the
assessment of comprehension.Memory & Cognition, 10, 597-602.
Goffman, E. (1967). Interactionritual: Essays on face-to-face behavior. GardenCity, NY: Anchor.
Graesser,A. C. (1993a). Questioningmechanismsduringcomplexlearning.MemphisStateUniversity,
Memphis, TN. (ERIC Document ReproductionService No. ED 350 306)
Graesser,A. C. (1993b). Dialogue patternsand feedback mechanismsduringnaturalistictutoring.In
W. Kintsch (Chair),Proceedings of the 15th Annual Conferenceof the Cognitive Science Society
(pp. 126-130). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc.
Graesser, A. C., & Person, N. K. (1994). Question asking during tutoring.American Educational
Research Journal, 31, 104-137.
Graesser, A. C., Person, N. K., & Huber, J. D. (1992). Mechanisms that generate questions. In T.
Lauer, E. Peacock, & A. C. Graesser (Eds.), Questions and informationsystems (pp. 167-187).
Hillsdale, NJ: LawrenceErlbaumAssociates, Inc.
Graesser, A. C., Person, N. K., & Huber, J. D. (1993). Question asking during tutoringand in the
design of educationalsoftware.In M. Rabinowitz(Ed.), Cognitivesciencefoundationsof instruction
(pp. 149-172). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc.
Grice, H. P. (1975). Logic and conversation.In P. Cole & J. Morgan (Eds.), Syntaxand semantics:
Vol. 3. Speech acts (pp. 41-58). New York: Academic.
Grice, H. P. (1978). Furthernotes on logic and conversation.In P. Cole (Ed.), Syntaxand semantics:
Vol. 9. Pragmatics (pp. 113-127). New York: Academic.
Griffin, P., & Humphrey,F. (1978). Task and talk at lesson time: Children'sfunctional language
and education in the early years (Final Report,CarnegieCorporation).New York: CarnegieCorp.
Kasper,G. (1990). Linguisticpoliteness:Currentresearchissues. Journal ofPragmatics, 14, 193-218.
Kreuz, R. J., & Roberts,R. M. (1993). When collaborationfails: Consequencesof pragmaticerrors
in conversation.Journal of Pragmatics, 19, 239-252.
Leech, G. (1983). Principles of pragmatics. London:Longman.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions
188 PERSON, KREUZ, ZWAAN, GRAESSER

Leinhardt,G. (1987). Developmentof an expertexplanation:An analysisof a sequenceof subtraction

lessons. Cognitionand Instruction,4, 225-283.
McArthur,D., Stasz, C., & Zmuidzinas,M. (1990). Tutoring techniquesin algebra. Cognition and
Instruction,7, 197-244.
McKeown, M. G., Beck, I. L., Sinatra,G. M., & Loxterman,J. (1992). The contributionof prior
knowledge and coherenttext to comprehension.Reading Research Quarterly,27, 78-93.
Mehan, H. (1979). Learninglessons: Social organizationin the classroom. Cambridge,MA: Harvard
University Press.
Mohan, M. (1972). Peer tutoringas a techniquefor teaching the unmotivated.Fredonia,NY: State
University of New York, Teacher Education Research Center. (ERIC Document Reproduction
Service No. ED 061 154)
Newman, D., Griffin,P., & Cole, M. (1989). The constructionzone: Workingfor cognitive change
in school. Cambridge,England:CambridgeUniversityPress.
Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fosteringand
comprehension-monitoring activities. Cognitionand Instruction,1, 117-175.
Penman, R. (1990). Facework and politeness: Multiple goals in courtroom discourse. Journal of
Language and Social Psychology, 9, 15-38.
Person, N. K., Graesser,A. C., Magliano, J. P., & Kreuz, R. J. (1994). Inferringwhat the student
knows in one-to-one tutoring:The role of studentquestions and answers.Learningand Individual
Differences, 6, 205-229.
Putnam,R. T. (1987). Structuringand adjustingfor students:A live and simulatedtutoringof addition.
AmericanEducationalResearch Journal, 24, 13-48.
Resnick, L. B. (1977). Holding an instructionalconversation:Comments on chapter 10 by Collins.
In R. C. Anderson, R. J. Spiro, & W. E. Montague (Eds.), Schooling and the acquisition of
knowledge (pp. 365-372). Hillsdale, NJ: Lawrence ErlbaumAssociates, Inc.
Resnick, L. B., Salmon, M., Zeitz, C. M., Wathen, S. H., & Holowchak, M. (1993). Reasoning in
conversation.Cognitionand Instruction,11, 347-364.
Stevens, R., Collins, A., & Goldin, S. E. (1982). Misconceptionsin students' understanding.In D.
Sleeman & J. S. Brown (Eds.), Intelligent tutoringsystems (pp. 13-24). New York: Academic.
VanLehn, K. (1987). Learningone subprocedureper lesson. Artificial Intelligence, 31, 1-40.
VanLehn, K. (1990). Mind bugs: The origins of procedural misconceptions.Cambridge,MA: MIT
Press.
Weaver, C. A., III. (1990). Calibration and assessment of comprehension.Unpublished doctoral
dissertation,Universityof Colorado,Boulder.

This content downloaded from 130.56.64.29 on Fri, 04 Sep 2015 06:44:37 UTC
All use subject to JSTOR Terms and Conditions