INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI filmsthetextdirectlyfromtheoriginalor copy submitted.Thus, some thesisanddissertation copies are in typewriterface.while others may be from anytype of computer printer.
TheqlU:lityofthisreproductionisdepend ent uponthe quality oftbe copy Jubmitted. Broken or indistinctprint.colored or poorquality illustrations andphotographs.,printbleedthrough. substandard margins.
andimproperalignment can adverselyaffectreproduction.
Inthewililcelyeventthatthewthordidnot send UMI a complete manusaiptand there are missingpages,thesewillbenoted. Also,if W1aUthorizedcopyrightmaterialhadtoberemoved, a notewillindicate thedeletion.
Oversizema1e:rials(e.g.,maps.,drawings. charts) arereproducedby sectioningthe original. beginningatthe upper left-hand comerand contim.ringfromlefttorightinequalsectionswithsmalloverlaps.Each originalisalso photographedinone exposureandisincludedinreduced fonnatthebackofthebook.
Photograpbsincludedin theoriginalmanuscripthave been reproduced xerographicallyinthiscopy. Higherquality6" x 9" black and white photographicprintsare available foranyphotographs or illustrations appearingin thiscopyforan additional charge.Contact UMI directlyto order.
UMI
A Bell .I;HowdII:afbrmalioDCompany 300 NorthZe:ebROllli,AoDAd1or" MI 48106-lJ46 USA
313n61-4700 8OOr.I21.()6O(l
NOTE TO USERS
The original manuscript received by UMI contains light print.
All efforts were made to acquire the highest quality manuscript from the author or school. Microfilmed as
received.
This reproduction is the best copy available
UMl
THE EFFE CTS OF PARENT AL STYL EONNARRATIVE PRODUCTIONINPRES CHOOLERS:ANINTE RVE NTIONSTUDY
By
BeulahA. Jesso
A t.hesissubmit.t.ed t.o t.he School of Graduat.e St.ud ies in part.ial fulfilment. of the
requir ementsfor th e degree of Mas t e r of Science
Departme ntof psychology Me mo rialUniversityofNew f o und l a n d
Spring 1997
St . John's Newf ound land
Abstract
The ove r a l l narrative product i v i ty of econom i ca l l y disadvantaged presch oolers was eva l ua t.ed pri or ec and followi n g in t.e rv ent.ion. Participan t.s were cwencvcree c hce t ch i ld r e n and th ei r mot.h e r s, who wer e randoml y assign ed eo ei t.her anint ervention orcon e r ol gro up . Th e studyconsist ed of a preliminary test, 12 months of in t e rve n t ion an d a post test. Fourt.een childr en (7 in each group) also particip a ted in a follow- up asses sme n t tha t occ u rreda year after theend of intervent ion. Al lchildren'snarrativesfrom bot h the prete st an d postt.est we r e analysedforthequan t. ity an d length of propos itions, unique units of in fo rm a tion, dec ontextua l ized info rma t ion, and simple andcomp l e xtempo ral All na r r a t ives produc ed by the parent s in both the pre t e st and post test were analy s ed for the number of utt era n c e s. open -en ded prompts , yes/noand wh- quest i o ns. and bac k - c hanne l lin g. It wa s predict.ed tha t fol lowing tr aining thein t e rve n tiongr ou p would surpassthecont ro l gro up onall aspect.s of na r r at iv e productivit. y . Children in the int e rvent i o n group showed no imp rove me n t relaeive to the cont rolgroup in the posttest, excepton avocabularymeasure.
howe ver a year la t e r at th e time of follow- u p as s e s s me n t in t e rve n ti o n children produced decontextual ized descrip tionsofwhereand esp eciallywhenthe describedevent s
ii
took place. Such decontextualized language has been emphasized as imp o r t a n t for literacy acquisition.
i ii
Acknowledgment s
The aut h o r gra t efu lly acknowl edg e s Dr. Carole Peterson for he r assistan c e over the past several years. Wi t ho u t her continue d su pport, guidance and patience t.he completion of thische s iswou l d not have been possible.
Since r e thanks are extended tocb e parents and children who partic i pated in this st u d y. The i r coop e r acion and wi l l ingness co parta ke in this pr o j e c c was an invaluable contribution andaremuch apprec i a ted .
Al s o , specia l thanksto my committee: Dr. MaryCo ura g e andDr. MikeRabinowitz , to Penny Vo u t i erforher ass i s t a n c e withrel iabili t y trai ni ng, toElizabe th Votta an d Dr. Malco lm Grant for ass i s t a n c e wi t h stat ist i c a l an a l y s i s , to Marleen Biggs, Tina Parsons , and Gi na Ri d e o u t for follow·up data col lectionand to Kim F'roude for follow-updata ana l y s i s.
rv
ABSTRACT Acknowledgments LIST OF TABLES INTRODUCTION METHOD RESULTS DISCUSSION REFERENCES TABLES
CONTENTS
Page i i Lv vi
12 21 27 34 40
Tables
Ta bl e 1. Quantit,yand len gth of na r r a t.ive s produ c e d by the children (a n dsta n da r d dev iat i onsI
Tab le 2_ Mean nu mbe r of unique units of info rma tion in the ch i l d r e n'snarratives (a n d standard deviati onsI Ta b l e 3 _ Amount of deconte xtual i zing (wh e r e and whenl
in fo rm a tio n in the chi l d re n' s na r r a t ives (a nd st a n d a r d deviat ionsl
Ta bl e 4a. Mea n numberof simple temp o r a l terms (a n d standard deviationsl
Tab le 4b. Mean numberof compl extemporal terms (a nd standa rd deviationsI
Ta b l e 5. Mean number of pa rent measures (a nd st andard de v ia tionsI
vi
The Effects of Parental Style on Narrative Production of Preschoolers: An Intervention Study
Researchers interested in the development of child language can study a number of language units, including words, sentences, and discourse. Over the past few decades, however, focus has shifted increasingly toward discourse analysis. Many researchers are now assessing the various processes involved in discourse through the examination of narrative texts, and specifically, the personal experience narrative (Feagans, 1982; Hudson & Shapiro, 1991;
Liles, 1987; Peterson, 1990; Peterson & Dodsworth, 1991;
Peterson&McCabe, 1991; Snow, 1983; Snow &Dickinson, 1990).
Precisely what a narrative is varies from definition to definition but will be formally defined herein as one way of recounting past experiences whereby a speaker will verbally provide a sequence of clauses which coincides with a sequence of events that has actually occurred (Labov, 1972).
There are a number of reasons for the considerable attention given to the personal experience narrative. First, i t is the only form of narrative that can be elicited from very young children. Children as young as two years of age can tell about personal experiences that have occurred in the past {Eisenberg, 1985; Fivush, Gray & Fromhoff, 1987; Miller
& Sperry, 1988; Peterson & Bell, 1996; Sachs, 1983; Todd &
Perlmutter, 1980). Second, they are relatively identifiable units (i.e., they have a marked beginning and end) (Labov, 1972; Mandler & Johnson, 1977; Peterson & McCabe, 1983) and third, they are common (Peterson &McCabe, 1991).
One of the defining features of narration is that i t is a form of decontextualized speech (Graesser, Golding &Long, 1991; Peterson&McCabe, 1994). In other words, it is speech about events that are removed from the immediate context; it does not describe the here-and-now, but rather the there-and- then. This implies that narrative discourse should be able to be understood without additional supporting context. A listener who was not present at a described event should be able to understand the story. An important component of achieving this is provision of orienting information (who, when, where, why, and what object). In order to provide a coherent account of the experience, narratives should also be informative, contain explicit temporal and causal relationships, and be chronologically organized. The inclusion of such information would be indicative of a well- structured narrative. According to Graesser et al (1991), when children produce narratives they are no longer depending on the immediate environment but rather they can use mental images. This decontextualization allows the narrator to speak about times other than the present, to focus on the specific
attributes of events and to contrive alternate possibilities for events (French, 1986; as cited in Graesser et , al. 1991). Narratives are quite common in daily activities of the classroom, such as story telling, show and tell and "sharing tLme ". These activities often involve having the child verbally describe some object or produce a narrative account of a past event, with the teacher acting as a discourse facilitator, providing questions and comments (caaden , 1988;
Michaels, 1981). This mediation on the part of the teacher assists the children's narrative composition. According to Michaels (1981), events such as sharing time provide a link between the oral discourse that the child has experienced at home and literate discourse that is necessary at school. While such activities provide exposure to the kind of instruction and practice needed to acquire narrativeskills, children are expected to possess some discourse skills when they enter school.
Discourse skills have been identified as a critical link to successful school achievement (Bruner, 1986; Miller, 1990;
Olson, 1982; Woo d , 1992). In particular, the ability to produce decontextualized speech is reported as being connected with academic success, especially literacy (Dickinson, 1991;
Olson, 1977; Snow, 1983). Snow (1983) explains that children show a developmental change from contextualized literacy skills (reading thename on a sweatshirt when accompanied by
a picture or read ing the nameona favo riteboxof cerea l ) mor e dec o ntext ua l ize d l i te r acy skil l s (e. g ., readi ng wor d sand sent e n c e s wit ho u t ac c ompanyingpi ctures ). Accor dingto this re search, it is th is tr ansi t. ion fr o m conte xtuali ze d to de c ontex tualizedlangua g e t.ha t enab l e s in dividuals coacqui r e literacyski l ls . Na r r atives are a particularly goodformatfor developingdec ontextualizedlanguage sk i lls becausethey ar e about events that are removedintime andsp a c e.
Ac c ord i ng to Feagans (1 9 8 2) , narra t ive skil l is a prerequi site for sc hoo l adaptation, and unfortunately many chi l dre n enterscho ol with poorna rrative sk i l ls _ Th i s is es p e c i all y true for chi ldren wh o come from communi ties with langua g ede mand s that are differe nt from the langu a g edemands of the cla s sroom.
The narrative skills of children have al s o been fre quently as s oc iated wi t h soc ia l cla s s. Once i t wa s bel i e ved that lowerclass ch i ldre n ofte n perform poore r in schoo l than do midd le cl a s s ch i ldre n due to a linguistic de f i c i e n cy, especiall yin syntax. However, overthe pas t few decades this notion ha s been discredi t e d. Ac cordingto Bruck and Tuc ker (1974). the li n gui s t i c sophist icat ion of lowe r class child ren is equivalent to thatof their middle cla s s peers. The difference in schoolpez-tlozmance is nowbelie v ed cobe dueto lac kof preparation forschoo l prog rams thatare geared toward childr e n who have already ac qu i r e d specif i c
language skills (Bruck&Tucker, 1974). Consistent with this premise is the finding that middle class children have the narrative skills necessary to meet the demands of the classroom, whereas lower class children enter school without having already acquired such skills (Feagans, 1982; Heath, 1981; Peterson, 1994).
According to Cairns, Cairns and Neckerman (1989), a relationship exists between socioeconomic status, school performance, and subsequent school drop-out rates. In a longitudinal study that examined behavioural, cognitive, and demographic factors associated with early school drop-out, Cairns et al (1989) reported that seventh graders
likely to attain a low level of academic performance if they had low socioeconomic status. In turn individuals who performed poorly academically were more likely to drop out of school. Walker et al (1994) reported that children from economically disadvantaged families performed more poorly on tests of verbal ability, receptive and spoken language, and academic achievement as measured by standardized tests in kindergarten through grade three. We can now address these differences in terms of what happens prior to the onset of formal schooling.
There is little disagreement among investigators that a child's language environment plays a crucial role in shaping the development of his linguistic performance. One way to
explo r e th e developmental dif ferences in la ngua g e between chi ld r en from diff e r ent soc ioeconomic classes is to compa re the ver b a l envi ron men tsof lowe r and middlecla s s children. Recently, se vera l rese a rcher s have conf irmed a number of diff erences in th e lan gu a g e style s of mi ddl e cl a s s child r en (s ee reviewin Fenson , Da l e, Rezni c k, Bates, ThaI&.Pet-hick, 19941. Althoug h parentalbehaviour may not beth eon lyfa c tor underlying th e s e child style di f f e ren c e s , most re s e a r chers viewparents as maj o r cont r i b u t o r s. Becau se in the first 5 years ofli f e par e nts ar e noton l ythe primaryca r e t a k e rsbut alsoth e prima ry teac he r s, the ro le of pa r e n t s ina child ' s ve r b al en vironm ent wil lbe the focus inthe present paper.
Theverba l int era c t io n be t ween mothe r andch i ldha sbeen shown to havean ef f ectonth echi l d'sla nguage. Fo r example,
Ne lso n (198 11 dis tinguis hed betw een two different la ngua ge
styles produced by chi l dre n, referene ial and express ive. Ref erent ial children ar e cha r acter ized by frequent us e of common nouns, ea rly vo cabulary acquisition , and the use of la n gua g e asa devic e fo r gaininginfo rmation . On the other hand , express ive childrenprodu cespee ch tha t has a scarc i ty of common nounsan d an abundance of pro nouns, are slowerat acquiring vocabu l ary, and use language mainly for soc ial inter ac t ion. Researchers ex amining the maternal speech of exp r e s s ive an d referential child ren report a numbe r of di fferences. Mo thers ofexpressive chi1<ir e n, asco mp a re dto
mothers of referent-ial children, use more person an d fewe r object re f e r e n c e s {Fu r r ow &: Nelson, 19B4l, respond le s s fr eque n t l y their children's attempcs initiate conversation, and provide fewerextensions and expansions of their child ren's utterances (Li e v en, 19 7 B).
Such re s e a r ch involve s language with a hezeva n d-n ow contex t: events that are taking place in the present. As childre nde velophowe verthe r eis increasingly moretalk about topics that ar e not inth e presen t conte x t (Sachs , 1983). It has be e nsug gested(Eise nbe r g, 1985;Peterson&:Mc Cabe , 1994) that th e kinds of inform a tion that parents reque s t from children provide children wi t h cues as to the kinds of in f o rma t i o n they should prov ide when pr o d u c i ng their own narr atives.
Peterson and McCabe (19 9 2, 1994) reported chat pare nts wh o re gu l a r l y asked many WH-quescions and prompced for concexcuali zingla ngua g e (s u c h aswh en and ....here the described event took place) had children wh o regularlyproducedsimilar information in their stand-alone narra tives. McCabe and Pete rson (1991) also discinguished among seve ra l types of pa rent a l sc y l e s of narrat i ve elicitation including copi c- exte n din g and topic - swi t c h i n g. Topic-ex t e n d ing is charac t e r i z ed by staying on the same topic whereas topic- switching impl i e s introdu c i n g many dif fere n t topics. Their data re fl ectthat pa re nt s whoare topic-extendinghad children
who produced lengthier narratives over time whereas parents who are topic-switching had children who produced relatively shorter narratives.
Fivush and Fromhoff (1988) also explored different maternal styles for conversing about the past. They too, observed two different types of conversational styles, elaborative style and a repetitive style. The former is characterized by a rich description of the past event being discussed and providing additional information with each additional question asked, whereas the latter is characterized by little reference to the past event being discussed and few, simple, and redundant questions. Maternal style influenced the type of information recalled by the children. Children of elaborative moms recalled nearly twice as much information as children of repetitive mothers. This was true for all types of information including location, people, objects, activities, and descriptives. Also, Fivush (1991) reports that the way mothers structure their conversations about the past will have an effect on the way their children will produce personal narratives themselves. According to Fivush, most children can produce simple temporal links but children who have mothers that provide more complex temporal narratives (i.e., by using more causal/conditional terms) earlier in development will produce complex temporal links themselves. Similar findings were also reported for the
number of propositions per conversational turn: children produced more propositions per conversational turn in a later interview if their mothers provided more propositions per conversational turn in an earlier interview.
Evidence that parent-child interactions affect developing child is consistent with Vygotsky's (1 97 8 ) developmental theory. According to Vygotsky, development can be explained with reference to the zone of proximal development which is "t he distance between the actual developmental level as determined by independent problem solving and the level of potential development as determined through problem solving under adult guidance .... " (pp . 86). Diaz (1991) describes the zone of proximal development as having two facets. The first of these is joint collaboration, which refers to the active participation or sharingof task responsibility by both the child and the adult, and the other is transfer ofresponsibility, referring to the increasing role of the child as the role of the adult decreases. The increased role of the child is achieved through the construct
of "s c a f f o l d i n g" (Bruner, 1983). By gradually decreasing the
amount of support, the adult provides the child with more opportunities to complete the task themselves. Initially, tasks will require much adult support, but as the child makes repeated attempts to perform the task, the adult can gradually withdraw the support until the child has mastered the task.
10 Vygotsky'sth eo r e t ic a l perspective suggeststh a t la n gu a g e in t e rvent i o n with parents at home wi l l enhance a ch i ld's language ski lls . The efficacyof language in t e rve n t i o n is suggested by a numbe r of studies. Many of the cu r r e n t in t e rve n t i o n st.udies in v o l v e ch i l dre n who ha v e developmental delays (Ea rhea rt,19 8 2 :'ra nno c je, 1988; Ta nno c k , Gi rola mett a&
Siegel. 1992) . A consistent fi n d i n g among these researchers isthatmothers of delayed chi ldrentendtobe more direct ive and less responsive th a n mothers of chi ldrenwi t ho u t de la ys . andth a tthe s e childrenini tia te fewe rin t e r a ction s. Th u s, i t appears th a t the mother'sstyle mayin t e r f e r e withthe social interaction (i .e.• turn taking, initiati ng ) sk i lls of the child. Bo rn fromthis int e rp r e t.a t io n are a number ofpar ent - foc u s e d in t e rve n tion programsaimedat. ch a ngi n g t.h e mcc herrs style of int e r a c t i o n . In on e such study , Tan n ock et al.
(1992}foundth at int.ervent i on le d to ch a n g e s inth e speech of
mothers su ch th a t th ey became more responsive and les s dir e ct ive. More imp orta n t l y , thi s changein mat e rnalspeech was accompaniedby aninc r e a s e in the number ofcon v e r s atio n al turns producedbyth e i r children .
Whitehurstand Valdez-Menchaca (1 9 8 8) imple me nte d ahome inte rvent ion te c hniqu e to teach middleclass mot hers touse te c hnique s that altered the role of mo ther and child whi le reading. Ul t ima t ely the childwould swit ch from being th e listener to beingth e teller and the motherwo u l d switch fr om
11 being the te l l e r to bei ngan active listener. A compar iso n gro up received no in t e rve n ti on. Whit ehurst and Valde z - Men c haca foun d that thechi l dre ninvolve din thein t e rvention ha d an in cre a s e d me a n le ng t h of utte r a nc e as wel l as subs t antial gains on standardized tes t s of la nguag e de v e l o p me n t. Asi mila r procedure was imple me n t e d with lower cla s s childrenwhoattendeda Mexicandaycare(Vald e z - Me chac a
&Whitehurst , 1992 ). However, graduate st u d e n t s rather th a n
mothers carriedou tthe intervention. None-the-less,language gains (i n c r e a s e d number of verbal productions and increased sc o r e s on standardized language tesCs) were re p o r t e d for ch i l d r e nin the in terve n t i o n gro up .
Recently Wh i tehurst, Epste in, Angell, Pa yn e, Crone &
Fishell (1 994) have reported th a t Lcw-Lncome fa mili e s wi t h paren tswhoengaged inact i veboo k rea dingathome wi thth eir childre n had ch i ldren who preformed better on standard i z e d tests of langua ge, wri t i ng, linguistic awarene s s, and print conce p t s than did chil dre n who were not envolved in such reading . Aswasdisc u s s e d earlier,childre n fromecono mi cal l y disadvantaged backgrounds often experi ence dif f i cul ties in school and, in particular, in thei r producti o n of decontextualized sp e e c h. This decontextualized speech is characteristic of na r r at i v e discourse.
In th e present expe r i me n t, lower class parents were trained to use certain techniques when elic iting pers o nal
12 narratives from their child. These techniques included having the parents talk to their child frequently about past experiences. consistently spending time on a single topic, inserting many wh-questions and few yes/no questions. Also parents were instructed to listen carefully to what their child wa s saying and to aim at having their child say than one sentence at a time byusing responses such as "um- hum". "really?" or "tell me more" or simply by repeating what their child had just said. Parents were also instructed to follow their child' s leadbytalking with them about whatever i t is they wanted to talk about. It is hypothesized that the way parents prompt their child for personal narratives will have an influence on the types and complexity of narratives the child will produce.
Method Participant.s
Twenty children, 10 male and 10 female, and their mot.hers, participated in the study. All families were lower- class, living in subsidized housing and in receipt of social assistance. The children entered the study at a mean age of 3;7 (range .. 3;3 t.o 3;11) and were followed for 12 months.
Approximately a year later when the children were 5 1/2years old (mean age SiB), fourteen children (7 from each group)were located for a follow-up assessment.
13
~
The chi l d r e n we r e quasi-randomlyassigned to eitheran in t e rve n t i o n or control group.each group consisting af 5 boys and 5 girls. All childrenwere visie.ed in t.heir homebyth e re s e arc h e r. The reseacher establ i s he dra p p o r t byplayingwi t h t.he children during the first 2-3 visits wh i c h took place wi t h i n two we e k s . Once rapport was es t abli s he d the Peabod y Picture Vocabu laryTest (Pf'VT ) . the Clinical Eva l ua t i o n of Langua g e Fundament a l s (CELF) andanar rative eli ci ta t i o n task we readministe r e d to all chil d r e n. Followingthis asses sm en t , th e child ren we r e visited in thei rhomes approx i ma te lyev e ry ot.h er monthfor 1year.
PPVT. The PPV'l' is a standardi zed, individually administered te s t. ....hichmeasures receptive vocabulary. Each t.est contains 5 practice it e ms , followed by 175 test it e ms which are ordered fr o m most easy to most diff icult . When present.ed withan arrangement.of fourpi ct.ures , che subjec t is required to choose ebe pict.ure that. beec i l lust.rates t.he meaning of a word presented orally by the examiner.
CELF. The CELF is a st.andardiz ed test. of language funda me nt. als whi c h asses s e s both re c e p t i v e and expressive langua g e. Linguistic con c e p c e, ba s i c conce pts. and sentence st.ruct u r e meas ure s are use d to def i ne rec eptive lan guag e . Recalli ng eent.encee in co n t.e x t. , formu lat:.in g labels and word st:.ruc tur e ar e usedto define exp ressivelangua g e.
14 Narrative Elici tation Task. Al l subjects rece ived a pretest whereby the experimenter met. wi th ea c h chi l d in d i vi du all yfor appr oximat.e ly 30 mi nu t e s. During thi s t.i me th eexp e r i me n t. e r el ici t.edpersona l expe r ience narra c ives fr o m th e chi ld by present.ing standard ized l is t s of na r r a t i v e prompt s . These sho r t na r r ativ es were ins e rt e d wi th i n a context of play wit h the ch i ld, and each was fol lowe dby a general prompt. such as, -otaanyth ing l ik eth a tev e r ha p p e n co yo u?" One such ex a mple is as fol l ows . "I went. tri ck or t.r eat i ngonce and someof th ecos t u me s werere all ysc ary. Did youevergo t rickor tr e a t ing?" Ra the rtha n restructur ingthe na r r at i ve s of chi ldre n , t.he exper ime nter' s commen ts were res t r i c ted to genera l ind icatio n s of int e rest an d enc ouragemen t suchas nUh-huh " , "Yeah? ", "Really?" , "Andth e n what ha p p e n e d ?" or rep e t i tion s of what th e chi ld ha d jus t sai d. Acc ordi n g toPetersonandMc Cabe (1 9 8 3) , thes ecomments are suc c e s sful at en c o u r agin g narrati o n witho u t impo s i ng struc t ure .
App rox i ma te lyon e year la t e r, al l chi ld r e n recei ved a pos t t e s t, admini s t eredby an independent researcherwho was blind to thegroupmember ship of th e chi ld . This wassimp ly a repeat of th e pre l imi narytesting using a di f fe r e ntli s t of standard izedprompts {e. g. , "I went to a birthd a y party at McDonald 'sonce. Haveyou or anyof your fri e n d s ev er had a birthday partyat McDonald's?"
lS One year after the posttest 14 (7 fromeach group) of the 20 children partic ipated in a follow-up asses sment which consisted of repeat of the posttest wi thyet a differe ntli s t of standa rdized prompts.
On vi s i t s between the pretest and post test. the expe rimente r eng aged inplay wi th the children, duringwhich t ime addi t i onal attempts to elicit personal experience narrati veswer e made. As inth e pr e t e stan dposttest s, short narrativeswe r e inser t e d wi th i nthe contex t of playandeach narrative was followe d by a ge ne ra l prompt such as "Di d anythin g like that eve r ha p pen to you?- A.gai n , since th e experi me nte r wa s in t ere s ted in what the child wo u ld say spontaneously,no specific questionswere asked. Instead, the experimenterusednon- s p e cif i c prompts such as-y eah? - , -and?- or simply re p e a t e d what the child had sa i d with quest i o n inton a t i o n. The experiences prompt ed for we re common expe ri encestomostchi ldren, such asha ving a birthdayparty, getting a needle or fallingoff a swing. All sessio ns were audio- r ecorded and later tr anscribed.
In tervention. On c e theexperimentercomp le t ed a sin g l e narr a t ive se ss ion, the parents of th e childre n in the inte rvention group we re informed of the type of nar r a t i ve interact ion that can fo s t e r the i r chil d re n's languag e development an d were co nt inuo u sly encou rag e d to act accor d inglyin the following manne r.
rs II The primarygoal was to establish rappor t with the mothers and to in t e r e s t them enough so th a t th e y would participate in the st u d y. At th e start of the researc h
project th e researcher aimed t.o establish th e import a nceof this work. Included he r e was in f o rma t io n concerning t.he aspectsof storytel l i n g (abo u t past event s ) th a t are link e d to school success, particularly readingand wri ting. It wa s exp l ainedthatchil dre nwh otel l goodna r r a t i v e s arelikelyto adj ustwellinsc h o o l. When ch i ldren canprodu c e moretha n on e se nte nc e spontaneously, theyaremore li k e l y to • fit in' . 2' Moth e r s were informed of th e typesof research that has been co n d u c t.e d and relevant fi n d i n gs, specificallythat mothers speak tothl!ir childrenin dif f e r ent ways, andthat so me ki n d s of.tal k i n g are betterthan othe r s.
3) Th e researcher explained to mothers some ways that the y could assi st thei r children in ~com ing be t t e r st ory tell ers . Th e fo ll owingpoin ts we r e in c lu d e d and reinfo rced us i n g bi -weeklyphone con v ersa t io n s .
al Ta l k to yourchild frequentlyandco n s i s t ent l yabou t past experiences . Set a time eachdaywhe n you cantalkwith your child.
b) Spend a lot of ti me on a single top ic.
cI Ask plentyof wh-quest i o n s and fewyes/no questions. d) Listencare ful lyandpay closeattention towh a t you r child is saying.
17 e) Encourage your child to say more than one sentence at a time. This can be achieved byusing responses such as
"urn-hum", "really?" or "tell me more" and simply by repeating what your child has just said.
f) Follow your child's lead. This means talk with them about whateveritis they want to talk about.
4) At this point the researcher showed the parents actual transcripts and had them listen to transcripts that contained the types of prompting we wanted them to employ.
5) The researcher then practiced these steps with the mother through role-play.
Control. The parents of children who served as controls were simply informed that this research was being conducted to learn more about how children develop narratives.
Prior to providing any information regarding the study to either group of parents and again at the end of the study, an audio recorded conversation between the parent and child was collected.
Measures of Analysis
Child Data. All narratives produced by the child in the pretest and posttest were analyzed. Any instance of talk about a specific event which is removed in time and consists of at least two related clauses was considered a narrative.
This definition is consistent with that used by other
18 researchers (Peterson, 1990; Umiker-Seobek, 1979). A clause was considered any ut terance containing both a subj ect and a predicate, as defined by Peterson and McCabe (1 9 94 ) .
Number and Length of Narratives. The number of narratives produced by each child, including both those narratives that were prompted for we l l as those produced
spontaneously by the child, and their average length was tabulated. The latter was determined by the average number of clauses in the longest three narratives. Each narrative was also scored for the average number of clauses per conversational turn (i.e., the number of clauses that produced without adult interruption). Back-channelling (i.e.,
"Tell me more", "Uh-huh?" Really?") was not considered an interruption. The number of prompts was also counted. This included all attempts by the interviewer to elicit information from the child. This would include back-channelling as well as any direct prompts (i. e . , "Have you ever been to a birthday party" .
Unique Units of Information. All instances in which novel bits of information were produced was also tabulated.
This is similar to the analysis of information by both Fivush (1991) and Peterson (1994). This included information pertaining to person (L e.,"Cor i nn ewas with me", nannylet me stay"), location (i. e., "I slept atSidney's house", ItI was at the mall"), activity (Le., "I playedwith the -tendo game.
19
"1 ha d to clean it up"). object:. (i. e.• "Th e teacher gave
some money", "When I goed tri c k - o r- t r e a ti ng I got some pumpkin sandsome friestoo"). and at t rib u t e (Lce.• "It. wasa big Easter Bunny" , "Thisguyfe l l down on a con c r e t e step"1. Attributeswe r e further divided into object attribute (i. 8. ,
"Andr got a newbike"). person attribute(i.e.• "The new baby wasli t t l e ") and stateattribute (i.e.,"It got reallydark"). Decontextualized Information. Each narrati vewa s scored forthe amountof decontextualizedinformation. Thisin cl u d e d allins t a n c e s ofte mp o ralcontext, ind i c a t e dbywhen (th e ti me
the event occurred) and spat i al context., indicatedby whe r e (t h e locat ionof the narratedevents ). Ex a mpl e s of tempora l co n t e xt in clude "Iwe n t ther e ye sterday" and "I ha d to get a needle when I was a baby and examples of spatial co nt e x t include "t wa s in my backyard" and "He bringed me to th e Janeway".
SimpleTemporal Terms. The numberoftempo r al termswas co un t ed fo r each narrat ive . Narratives co nt ai n events that
aret.e mp orall y l in ke d. Theselinksca n be expressedt.hr oug h
t.h e us eof t.emporal t.erms whi c hin clud e then,andth en, first , next, be fo reand after.
Complex Tempora l Terms. Na rra t. i v e s ca n also cont a i n eventst.hat are causally co nne c t ed . Ca u s a l connec t i o n sca nbe exp ress edthroug hsuch te rm sas because, so , wh e n , where, if, wh i l e andunti l .
20 Pa rentpata. Theeocber- child tra n s criptswereanalyse d for par ent al da t a. Eac h past experienc e about; which t.h e mothe r s quest i oned th ei r ch i ld was considered a narrative.
The mot h e r'sspeechwasanalyzedforthe followingcomponen ts:
Paren t Ut te rance s per Narr a t i ve. The number of utte r anc espe r Darra t i vewastabul atedin ord e r to provide a quan t ieat i ve measu re of ho .... much eac hparent ta l k e d about. each narrative top ic.
Open ended prompts. This co n s isted of al l que s t io n s and/ orcommands that promp tedfo r info rma tion butdid not ask for ori e n tati veco n text in form ationan d required more t.han a simpl e yes or no response. -Wha t ha p penedt.hen"? "What did
you do at schooltoday·? and ·What happened at the Janeway·?
are ex a mple s of open ende d pro mp t s.
Back channeling/Repeticion. This inc l u d e d all cases in wh i c h the parent repeats what the ch i l d ha s said IL e.•
Child: "Ali t t l e castle" . Pa rent : "A li t t le castle ? Wow-
Ch i l d: -And a big castle". Paren t: "And a big ca stle-? ) or
prcvtce s anindicationfor the childto go on IL e., "u rn- hu m"
"yeah?" "tel l me mo r e " ).
WH-contex t Queseion s. This inclu ded al lquest i ons th a t prompt for contextual in fo rma t i o n (i. e.,who ("Wh o visit ed you yesterday" ? , when (" Wh e n did nannygohome" ?) . where ("Where di d mommy takeyoutoday"?)and whatobjec t (Wh a t was in you r lun c h t.oday " ? ) . The numbe r of vh-c c n e exc que s tions per
21 narrative were calculated.
YES/NO Ouestions. This included all yes/no questions that provided context information about time (i.e., "Did we go to nanny's yesterday"?), location (L. e. • "Did we go to McDonald's), person (i.e., Does Dorothy drive your bus"?) and objects (Le .• Did nanny give you a new power jeep"?). as well as questions generally focused on actions (i . e . , Did you fall down"?) or evaluations (i.e., "Was it a good movie"?)
RESULTS
Child data will be presented first. It was predicted that increases will occur in the posttest of the intervention group on the number and length of narrativee , the number of unique units of information, the amount of decontextualized speech, and the number of complex temporal terms. The parent data will be presented second. Here, for the parents in the intervention group, it was expected that number of open-ended prompts, the amount of back-channelling and repetition, and the number of wh-context questions would increase.
Preliminary analyses that included gender as a separate factor were all nonsignificant for gender. Therefore the data are collapsed across gender.
CHILD DATA
.EfYI. The scores for the intervention and control groups at initial testing were 52.5 and 54.0, respectively, and at
22 th e posetest assessmentwere 59.0 and 55.5, respective l y. A.
repeated measuresANQVArevealeda groupX testin t e r a c t io n 1£
(1,18) "18.5 8,.Q< .011.wi th theint.e rve n t ion groupshowing improve me nt byhavin g hi gh e r scores than thecont r olgroupon t.h e post test but not on the pretest.
~ CELF scores showed an incre a s e in the pose tes t scores for the in t e rven t i on {x _ 88.4 VB. X '"'96.5 . 1 and control gr o u p tx_85.8 VB. x '"' 95.6) . Analysis didnat show a sign if icant interaction nor ma i n effect for gro u p . Ma i n ef fe c t for test, howe ver , wassignificant If:.{I.IS) _ 15.52, R< .01). withscores highe r onth e post t est.
NARRATIVE ANALYSES All narrativesproducedby thech i ld in the pretest and the postte s t we re analyzed for se vera l prop e r t i e s; the number of narrat i ve s, the me an number of clauses in the child's three lon g e s t na r r at i ve s, th e me an number of claus e s pe r conv e rs a tion a l turn , the nu mbe r of prompts requ i r ed to l!!licit the narrat i ve, uni que units of in f o rma tio n, decontextualized in fo rm a tion, and tem poral organiz a t io n.
Narrat.ives Clau sesand Prgmpts
The means fo r the numb erof narratives , the number of clau s e s per longest three narr a t i v e s , th e numbe r of clauses per conversational turn, and the number of experimenter prom pt s appe a r in Table 1. In order to determine if grou p membershipor time of testinghad asi gni f i c an t effect on th e
23 nu mbe r of na r r a tives, narrative length or cne numbe r of prompts, 3re peated measu r e s ANOVASwer e calcul a t e d wit h group (Int e rve n ti o n VB. COntrol) a between subjects va riable and Tes t (Pret est VB. Postt:.est) a with i n subject s va ri a b le. No si gn ificanteffe ctswereobtained.
Unique Unitsof Informatig n
Themean number ofuni queunitsof info rmat ion present in th e pretestand po st test narrat ives ofboth group sof children are reportedin Table 2. A re pe ate d measuresMANOVAwa s used t.o analyseth e frequency of th e va ri o us typesof informat i o n (obje c t,lo cat i o n,activ i ty, perso n and att ribute).as well as the tota lswi thGroup (Int e rventionV9.Co n t r o l) the be tw e e n- sub jecl:sva r i ab l eandTest (Pre test VB. Pasttest)andUnit s of In fo rma t io n bein g the with i n-s ubj e c t variables. Anal ys is produce d no significantGroup interacti on s. The r e washo we v e r a signi fican t int era c t ion between Test and Units of In f ormat i o n, £:C1.0e,6 ) . 3 . 0 9, P< .mt.
Ilecont e x t u a lizedInform a t i o n
Theamount of decont e xt u alizedinforma tionpr esen t in the pr e t e st and po a t t.eat; ofbothgroupsof childre n arepr e sen t e d in Table 3. (see Table 3.J Th ese data we r e also ana l y s ed usin g repea te d measures MANOVA, wi t h group (2 le v e l s) the between subjec ts vari abl e andte st (2 leve ls ) andcontext (2 levels : wh e n and where ) th e dep e nden t va r i a bles. N'o significant cont r a s t eff ect s wereob tained.
24 Temporalorga ni z a tio D
Temporalor g a ni z a tionis signal l e dbytheus eof tempora l terms whi c h can in clud e bothsimpl e and comple x forms. The childre n'sna r ratives weresearched forboth type sof temporal terms. The mean number ofsimpleand compl e x terms pr ese ntin thepretestand pos etes t na rra t i v e sof both grou p s ofchi ldren are presented in Table 4. {Se e Table 4. J
A repeated measures MANOVA wa s perf ormed for simp l e temporal terms wi th Group {I nt e rvention VB. Contr olI a between-subjectsvariableand Te s t (Pr e tes t vs . Pasetes c) and Simple Temporal Term (Slevels: then , and the n , first , next and before) the dependent. variables. The MANOVA for simple t.e mpo r a l te rm s produced no signi f icanteff ect s.
Anot.h e r repeated measures MANeVA was perfo rmed for comple x t.e rms, wi t.h Group {In t.e rve n t. i on V5. Con t rolI the bet ween subjec ts va riable and Test(Pre t e s t vs. Po s t t e s tland Comple x Te mpo r al Term (S level s: becau s e. unt i l . so, if and while) the depende n t va riable s. The re were no si gni fic a n t resu l ts.
Parent Data
All narrat ivesproducedby the motherinth e initial and final ses sio n s wer e an a lysed for a number of measure s in c l udingthe numberof open-endedprompt s. back-channelling , wh and yes/no questi ons, and ot h e r utt e r anc e s (L e.. any
25 utterance that was not classi fied as one of the previous measures).Becausein t erven t i ontrainingfocusedon increasing the number of open-ended promp t.s, back-channelling and wh questions. a repeated mea s u res MANOVA ....as us e d to anal yse these thre e typesofut t e ran c e in the pretestand postte s t of bothgro up s ofchildren, wi th Group (i n terve n t i onVB. controll being the between subj ects va riable and Test (p r et e s t VB.
post-t e s t) and utterance type (3levels:openendedquestions, ba ck ch a nne l ling an dwh questions ) the dependent vari ab l e s . The MANOVA re v e a l e d a signific antgro u p Xtest inter a ct-ion, f:
(1 , 1 8 1 '" 5.56, 12< .OS, as wel l as a significant mainef f e ct fortest; .E{I ,I S I . 5.1 7 , .Q.< .05. Thus, cn e intervention mothers we re increasing the aggregate of the sorts of ut t era nc e s they wereenco u ra g ed toproduce more thandi d the control mothers. Interventionparents were alsoencouragedco decrease theiruse of yes/noques t i o n s;when the frequencyof thesequestion s we r e analysed wi t h group and testth e between- subjects and wi t hi n·subject variables, respectively, there we r e no significant ef fect s. Thus the in t e rve n t i o n didnot appear to affect the produc tionof yes/noquestions. saeene e were given no ad v i c e conce rni ng ot her ut t e ra n c e s . The freque n c yof th e s ewerealso analys edwi t h Grou p andTest the be t we e n - s ub jects andwi t hinsubj e c t variables, re s pectively, an d no signi f i c a n t effects were found.
26 Follo w- u p Dat a
Recently. 14 (7 in eachgroup) of the 20 subjects we r e locatedandre-interviewedin the sa mema nne r as theori gi n a l posttest. Th i s took pla ce app roxi mately 12 months after completionof the firststudy (s e e Tables 1-4) . Whi le groups didnotdiffer in the numbe randle n g t hof th e narratives , nor on the numberof the unique units of information, cn e v did di ffer on the amountof te mpo ral info rma t i o n produced. Mor e decontextualiz edinformation(e s p e c i a l l yte mpo r a l) aswel l as more complex te mpo r a l terms we r e producedby subjectsin cne intervention groupchanbysubjects in theco ntro l group.
Whe n the pretest and posttestscoresof th e 14 children whowe r e located for the follow-up assessmentwe r e compared wic h the pretest andpost test sco re s of the 6 children who coul d not belo c at e d for follow-upassessment nosighnif icant differenceswe r e found.
To analyseth e deconte xtualizedinformation aMANOVA was conducted with group membership(Co n trol vs. Int e rve n t i o n) as th e betwee n-subjectsvariableandtest (Pret est, sosu ceacand Follow-up} as th e within-subjects variable. When the ewe types of decontextualiz edinformation we r e analyzedseparately findi ngs for the amount of temporal (wh e n) informat.ion pr o duced revealed a signi f icant mai n effect for group CE {1,12}=3.64,P <.01, asi gnificanttest main ef f e c t 1.E( 2, 2 4)
= 4. 03, P-e.05) anda significant gro u p by test interaction
27
(.f.12,241 '" 4.56, P< .0 5 1 . This was also done for the ecee j, of all deconeextualized informat.ion and there was a signifi cant group Xtest interaction I.E (2,2 41 .. 3.69. Q <
.0 5 1. Overall , subjects in the in t e rve n t io n group produced significantlymoredecontextualizedinformation , a year afCer inte rve n tion ended, and especia l l y te mp o r a l decontex t:.ualizedinformation,tha ndid subjects in the control group.
Forthe anal y s es of co mple x tempor a l terms a MANQVAwa s also cond u c t ed , with group membe r s h i p (Co ntro l Interve ntion) as the be t we en-sub jects variable an d test (Pre tes t . Po s t test and Fo l l ow - u p) as the wi t. h i n - s u bje c t variable . Analysisshowe da si gnif ic a n t group maineffect (,E
11 , 1 2) "" 5.27, P< .05 ) wi t h subjects in the in t.e rvention groupproducing mor e co mpl ex temporalte rm s than subjects in the control group.
DISCUSSION
Previous researchhas indicatedthat childr enfrom lowe r class fami l i e s often do notposses s the languagepre r e qu i s i c e s ne cessary for school succe s s . Many studie s ha v e shown that language in te rve n t i on can be succe s s f u l at in c r e asi ng the school per f ormanc e of children; however, con clusions drawn from suc h rese archhave been re s t r i c te d by sa mple pa r a me t e r s (e. g., th e targetchildren were middleclass, deve lopmentally
28 delay ed or le arni n g di sabled l (Koniditsio t.is " Hun t er. 19 9 3; Tannock,Girolametta " Siege. 1992; Wi i g, 1990). The sample in the prese n t studyis excl usivelylowerclass.
The main focusof thisresear c h is to determine ifpa r e n t centered langua ge inte rven tionin lowe r cl a s s famil i e s could be ef fe c t ive. Mo s t re searc h e rs agr ee that. pa r ents aremajor cont. ribu to r s to the la n guag e sty l e of thei r child r e nand the kinds of in fo rma t i o n that parents reque s t are the kinds of in f o rmat i o n thatch i ldr e nwil llate rpro du c e on theirown. In fo cu sing on the importa nc e of comp e te nt ad ults ce a c h i ng new ski l ls to chil dren by providing a scaffold wh i c h is the n progress i vel y decreas e das child r en's masteryof thenewskill increases,we are taking a vygotski anapproach. a perspective wh i c h suggeststhat langua g e intervention wi th parentsat home will enhance a child's language skills.
It wa s proposed that training parents to el ic i t nar ratives from children using the previously described techniqueswould resul t inthe childr en later producing more complexnarratives. Parents wer e trai ned to use fe weryes/no questi onsbut mor e open- end e d pr ompt s,back channel responses an d wh quest i o n s. Comp l e xit y inthechildren's narrat i veswas measured,notonl y by the length ofnarrativ e sbut als o by the ove r a l l qualityof na rrati vestruc t ur e includ i n gth enumb erof uni qu e uni t s of in f ormationprovid ed, the amount of
29 decontextualized information and the number of complex temporal terms.
Analyses of the parent data indicate success in training the parents. At post testing, parents in the intervention group as compared to the control group increased their usage of the types of utterances targeted, namely, wh questions, back-channelling and open ended prompts. However, they did not decrease their use of yes/no questions which we had also aimed to do. Wh a t this tells us is that their style of talking with their children changed in important ways. By asking more wh and open ended questions and using more back- channelling, these parents are encouraging elaboration.
Because yes/no questions (i.e ., "Did you have fun"?, "Did you eat your peanutsvv) require only a one-word response they do not stimulate children to provide information to create longer narratives. Although intervention parents did not ask fewer yes/no questions over time, the proportion of all questions that were yes/no in form decreased since parents asked more wh and open-ended questions with time. An increase in the number
ofwh questions indicate that parents are now requesting more
contextual information from their children (who, when, where, why and what object). This is important because the type of information that is requested from children early in development is the type of information they will later produce spontaneously (Peterson & McCabe • 1994). An increase in
30 open-ended prompts is important because they are indications for a child to continue and they encourage children to produce spontaneous information.
Overall, parents increased the sort of utterances that were targeted in our intervention, as we anticipated. The primary question now is whether this change in parental language had an impact on the child's language. When we review our analyses of the child data, the findings from the original study are not consistent with what we anticipated.
The intervention group and the control group did not significantly differ on the length of their narratives, were there differences in the complexity of the narratives they produced. However, the PPVT scores of the children in the intervention group did increase from the pretest to the posttest, relative to the control group. This does suggest that the intervention children had gains in their receptive vocabulary.
In the follow up study i t is unclear whether the intervention was effective. Examination of the meansof the child measures in the follow-up assessment shows that the intervention may have had some effect since all the means are in the right direction. On average, the intervention children produced more narratives, their longest three narratives were longer and they produced more clauses during each conversational turn (Refer to Table 1). Also, they produced
31
more decantextualizinginformation(Re fe r to Table 3) andth e y provided more temporal terms (Re f e r to Table 4) .
Deco ntextua l ized in f o rm a t i o n analysis showed th at th e intervention group used orientat ion to both when and wh e r e significantlymore th andid th eco ntr ol groupinth e foll ow -up assessment. Prov i dingcontextualinformat i o n isparamoun t in good st.ory t.elling and is one of the defin i ng fe ature s of na rration . Narrati ve discourse must be unders t ood by a listener who was not present at the t ime of the described This becomes increasingl y imp o rt a nt when ch i ldre n enter school since sc h ool-a g e d childrenare expectedto ta l k about ti me s other th a n th e he r e and now to peop le who we r e unl i k ely to be present at th e s e descri bed eve n t s . By provi d ing inf o rmati o n aboutwh en (L.e . •yesterday, onSunda y, la s t night )andwhere (L e. , at theplayground , at da yc a r e, at my nanny'shouse )inth e i r narratives th e s echild r en are usin g a form of dec ontextua l ized speech th at can be und e rsto o d without supportingco nte xt.
In additi on to being inform a t ive and a form of decontextualized speech, na r r a t i v e s should also co nt ai n explic i t simple and complex te mpo r al relat ionships . The inc lusion of such in f o rm a t i o n in d i cates a chrono logically organized andwe llpatterned narrative. Analysesof temp o r a l orga n i z a t ionin the follow -updata showedthatt.h e means fo r the total number of complexterms were greater in th e follow
32 up test for the in t e rve n t i o n group but not for th e con t r o l group. No differences ....ere found for th e tot a l numbe r of simple te rm s. Thi s is consistent ....i th Fi vus h {l991l who reported that mos t. chi l dre n can use simp l e temporal term s (i.e., chen. and then, first, ne x t and before l and th a t th ese do not di f f e r e n tiate children of elaborative parent.s who foste r complex language ski lls fr o m children of repetit ive paren t s who do not foster suc h language skills _ However.
Fivush (l991 ) found that child r e n wh o have mo t h ers that are elaborative and encourag e complex narrat.ives early in developmen t will produce mor e compl ex temporal links themselves (Le ., be c a u s e,until, so , ifandwm. J er.
We did not. find changes in the factors that. do not measurecomplexi ty, na me l y lengthofna r r a tiv e s and numb e r of sim ple complexterms. We werein t e re s ted inseeingacha ng e in t.he qual i t.y. not the quanti t y of th e narrative . Whi l e th ese findings were not made in the orig ina l study, the follow-up study suggests partial support. It appears th a t slee pereffectsha v e oc curred. According toSeitz 1198 11 i t is possibl e th a t a behavior altreatment can have longterm effects wit h o u t having earlie r ones (but see Clarke'"Clarke , 1982). It. is possiblet.hat if pare nt s continu e d to use the interve ntio ntechniques . children exposed to th e s e t.echniques ata la t erage maybemore ca p ab leof learningthe skillsthat
33 produce changes in the temporal organization of their narrati ve a .
In summary, it seems that language intervention with economically disadvantaged children and their mothers can be successful. However, because we have a small number of subjects and large variability in scores we have only suggestive pilot data but it is encouraging. The implications of the current findings are substantial though since the children in the intervention group were more successful at providing decontextualized texts which is strongly linked to literacy acquisition.
34 References
Bruck. M. & Tucker, G. R. (1974). Social class differences in the acquisition of school language. Merrill Palmer Ouarterly, 20, 205-220.
Bruner, J .S. (1986). Actual minds, possible worlds.
Cambridge, MA: Harvard University Press.
Bruner, J.S. (19B3). Children's talk: Learning to use language. New York: Norton.
Caines, R.B., & Caines B.D, & Neckerman, H.J. (1989).
Early school dropout: Configurations and determinants. Child
Development, 60, 1437-1452.
Clarke. A.D.I & Clarke, A.M. (1982). "Sleeper Effects"
in development: Fact or artifact? Annual Progress in child psychiatry and child development, 94-112.
Dickinson, D.K. (1 9 91) . Teacher agenda and setting:
Constraints on conversation in preschools. In A McCabe &C.
Peterson (Eds.), Developing narrative structure. Hillsdale NJ: Erlbaum, 255-301.
Dickinson, O.K. &Tabors, P. O. (1991). Early literacy:
Linkages between home, school and literacy achievement at age five. Journal of Research in Childhood Education. 6, 30-46.
35 Earhart, E.M. (1 9 8 2 ) . Building prerequisite learning skills for reading and mathematics. Advances in Early Education andziaycaz-e , 2, 159-184.
Eisenberg, A. R. (1985). Learning to describe past experiences in conversation. DiscourseProcesses,8. 177-204.
Feagans, L. (1 9 8 2 ) . The development and importance of narratives for schoal adaptation. In L. Feagans &; D. C.
Farrans (Ed a.) . The language of children rearedin poverty (p p . 95-1161_ NewYork:Academic Press.
Peneon , L., Dale, P.S., Reznick, J. S. , Bates, E.• ThaI, D. Pethick, S. (1 9 9 4). Va r i a b i l i t y in early communicative development. Monographsof thesociety for Research in Child
Development, 59(5).
Fivush, R. (1 9 91). The social construction of personal narrat ives . Merrill-Palmer Quarterly, 37, 59 - 81.
Fivush, R. &Fromoff, F. A. (1 9 8 81 . Style and structure in mother-child conversations about th e past. Discourse Processes, 11, 337-355.
Fivush, R., Gray, J.T. , & Fromoff, F.A. (1 9 8 7). Two year-aIds talk about the past. Cognitive Developmen t,2, 393- 409.
Furrow, D., & Nelson, K. (19 84) . Environmental correlatesof individual differences in language acquisition.
Journal of Child Language, 11(3) , 523-534 .
3 .
Gra es s e r. A., Gold ing, J.M.• & Long . D.L. 119 9 1}.
Na r rat i ve eep e e e e n e .aedo nand co mprehension. In R. Ba r r, M.L. Kami l .P.B .Mos e n tha l .ioP.O . Pe a r son, fEds .} . Thehandbookof readingrese a r c h (Vo l. 21 {p p. 17 1-2 0 5J. NY: Lon gma n.
He ath, S.B. (1 9 8 2). What no bedtime story means: Narr a t ive ski lls at. home andat.sc hoo L Languagein socie ty.
11, 49 - 76.
Labov , w. (1972) . Language in th e inner ci t.y.
Phi la d elphia:Universit.yof Pennsylvania Press.
Lie v e n . E.V . (19 8 4). Interactional styl e andchild ren' s la n gu a g e learni ng . Topi cs in languages dis orders. 4(4'. 15- 23.
Mc Cabe. A. , " seee re c n, C. (1 9 9 1). Geetingcne st.ory:
A longit.ud i nal study of paren t.a l styles in el i c i t i n g na r r atvies and developing narrat i v e ski l l . In A. McCa b e (, C.
Pet e rson (Ed s . l . Developing narrat ive structure, pp.217 ·254. Hillsd al e, N.J : LawrenceErlbaum Assoc i a t.es.
Michael s, S. (198 1) . -S haring Time:- Children's narrat i ve st.yles and di ffere n t. ial access to lit.e racy.
LanguageandSocie t y, la , 42 3-442.
Miller , L. (1990). The rolesof langua g e andlearningin the de velopment of l i tera cy. Topi c s in la n gu a g e dis orde r s . 10(2) 1-24.
37 Miller. P. J. & Sperry, L. L. (1988). Early talk about the past: the origins of conversational stories of personal experience. Journal of child language, 15, 293-315.
Olson, D.R. (1 9 7 7). From text to utterance: The bias of language in speech and writing. Harvard Educational Review, 47, 257-281.
Peterson, C. (1994). Narrative skills and social class.
Canadian Journal of Education, 19, 251-269.
Peterson, C .•&Dodsworth, P. (1 9 91 ) . A longitudinal analysis of young children cohesion and noun specification in narratives.
Journal of children language, 18, 397-415.
Peterson, C.• McCabe A. (1983). Developmental psycholinguistics: Three ways of looking at children's narrative.
New York: Plenum.
Peterson. C. & McCabe, A. (1 9 9 2 ) . Parental styles of narrative elicitation: Effect on children's narrative structure and content. First language, 12, 299-321.
Peterson, C. &McCabe, A. (1 9 9 4 ) . A social interactionist account of developing decontextualized narrative skill.
Developmental Psychology, 30, 937-948.
Peterson, C. &McCabe, A. (1996). Parental scaffolding of context in children's narratives. In J.H.V. Gilbert&C.E. Johnson (Ed s .) , Children's Language, 9, 183-196. Mahwah, NJ: Erlbaum.
38 Sachs,J.(1 9 8 3). Talking aboutthereandthen:The emergence of displacedreferencein parent-childdiscourse. In K.E.. N"elson
(Ed .I.Children'slangu age(Vol . 4,pp.1-2 8) . New 'fork:Gardner
Press.
Seitz, V. (1 98 1). Intervention and sleepereffects:A reply toCl a r k e and Cla rke. Developmentalreview. 1(4) ,344-360 .
Snow, C. (1 9 8 3). Literacy and Language:Relationships during th epreschoolyears. Harvard Educational Review, 53, 16 5 -18 9.
Tannock, R. (1 9 8 8) . Mo t h e r's directiveness in their in t era c t i o n swith thei r chi ld r e n with andwith o ut down syndrome. Alnerican JounJal on MentalRetardation, 93, 15 4-168.
Tannock,R.,Girolametto,L .• "Siegel,L.S.(199 2 ). Language intervent ionwithch i l dre nwh o have developmenta l delays:Eff e c t s of an interactive approach. Special Issue: The relation of communication and langua g e development to mental retardation.
American Journal on Retardarion, 97 (2 ), 145-160.
umiker-Sebeok. (19 79) . Preschool chi ldren's interconve rsationalnarratives. Journ alot Chi ldLanguage, 6,91 -
ValdeZ- Menchaca, c. s, Whit e h u r s t, G. J. (199 2) . Accelerat inglanguagede v e l o pment throughpicture book reading :A systematicextension toMe x i c an daycare. Developmen tal Psychology, 28, 1106- 1 114 .
Vygotsky, L. s. (1 9 7 8) . Mi n d ineeeaeey.The development:ot higher psyc.!101ogical pr oc e s se s. In M. Cole, V. John-Steiner, scscricne r , & B. Souberman (Bd s .) . Cambridg e, MA: Harvard university Press.
39 Wa l ker , D., Greenwood, C.• Har t. B.• & Carta, J. (1 9 94). Predictionof school outcomes based on ea r l y language productionand sacioeconomi c alfactors. Child De v elopment . 65. 606-621.
Whi te h u r s t . G. J.• Epstein,J. N.• Angel l , A.L.• Payne, A.
C .•Crone, D.A. &:Fisch el, J.E . (1 9 94)_ Outcomesof an emergen t literacy in t e rve n t i o n in he a d start. Journal of Educational Psychology, 86, 542-555.
Wh i te h urs t ,G. J." veicea- a e n cba ca,M. C. (198 8 ). What. is the role of reinforcement in earlylanguage acquisition? Child Development. 59, 430-440.
Wi i g . E.H. (1991). Languag e-l e a rn i n g disabilities:Paradigms for the nineties. 41 s t. AnnualCo n f e r enc e of the Orton Dyslexi a Society (1 990 , washington.DC) . Annals-of Dyslexia.41. 3-2 2.
Wo o d, D. J.. (~ 9 9 2). Language, le a rn i ng and education. Educationaland Child Psychology, 9(2) 17-27.
40
Table 1: Qu a n t i t y and length of ne r rettvea producedbythe children (a n d standard deviationa)"
Timo!ofTest
Follow ·up
'*Nar r a t ives
Control (1. 9) 0.01 11.6 ( •• 0)
(1.3) (1. 5) 14. 0 (5. 8)
Clauses/longest3 narr.
Control (l.9) (l. ll 10.5(4.2)
(2. 0) 0.5) 19. 11
Cla u s e s/ t u r n at talk
Control (1.8) 10.91 (2.01
(2 . 0) (2 . 1) (3 . 3)
the pretestand po e t testmeans are from the entire sampleof 20children (lO/group) whereas the follow-up means come from only 14 children(7/g r o u p).
41
Tab le 2, Me a n number ofun i qu e units of informationin thechildren'snarratives (an d standarddevLatLona Le
Ti1lleofTest
Posttest Follow-up Object
ControlGroup 10.30.5) (2.7) 25.7 (17.3)
InterventionGroup 12 . 8) 13.2) 32.4 (1 7.1)
Location
Control Group
S.,
(1. 6) (2 . 51 7. ' (7. 6) Intervention GroupS.,
(2 . 6) 6.' (2.51 13.1 (6 . 9 1 Act i v i t yControl Group 2.6(1,7) 2.' 11.6) (2 5 . 11
Intervention Group (1. 8) 11. 9) 0).2)
ControlGroup (2 . 5) 12. 4) 21.9 (11 .0 )
Interve nt ionGroup e. a(3.6) '. 6 12.6) 32.7 06.0) Attributes
of Object
Control Group 1.2 (0.9) (1.0) 114.21
Int e rve n t i o n Group (0.3 ) (1 . 0 ) 21.) 114.5) ofPerson
ControlGroup 1.2 11.71 (1.3) 2.7 (2 . 8)
Intervention Group 11. 81 10.91 (6. 2)
ofState
ControlGroup (0.9)
...
10 . 9 ),.,
(6.1)InterventionGroup 1 . ' (0.7) (0.8) (17.61
Total Unique Units
Control Group )).6().O) 12 . 9) 117.9 180 . 51 Intervention Group ().Ol 27.9 14.0) 170.4 1102 . 81
the prete s t andpos t testme a n sare from the entire sampleof 2. children nO/group)wheras
'h.
follow-upmeans come from only 14 children l7/group ).42 Tabl e3: Amountof de con t ext ua l iz lng(wh e re and vhel'l linformati o nin "h e ch i l d ren's na r ra t i ve s (andst a n da rd dev iat i on s )-
Fo l l ow-u p Sp a ci alinfo.
Con trol Group In tervention Gr oup Tempora l info.Iwhen l
Co n t r o lGroup Int e rve ntion Group
Co ntro lGr o u p rne erve eerc e Group
theprete standposttest nlea nsere fr o m theentiresa mp leof20chi.ldren nO/g r o u p) whe re aslOh l! fo l low -upmeans cOIne frOCll only14chil d ren(7/gro u p l.
43 M.Olan numberofsimplee...mpor..l cerrns (a n d standard deviat ion s l-
Follow-up
Con trolGr ou p Interve n tio nGroup
Control Group Inre rve nt.ion Gr o up
ControlGrou p Inte rve ntionGroup
Con t rolGrou p InterventionGroup
Cont:rolGroup InterventionGroup Tot alsimpleterms
controlGroup Int;..rven t i on Grou p
-No t e: thepreeest and po " ..t .."eme an S arefr om t;heent i reS"llf!ll eof 2.0 <;hil d r e n (I O/ gro up) ·..he r e as th efol low- upmeans come froaoonly 14cmieeen(7/ g r o u pl.
44
Con~ rolGro..p In t .. rv.... tionGroup
ControlGroup Int e rve ntionGro".p
ControlGrou p In te rve n tionGrou p
Co n t r olGroup Int erventi.on Group
Co n t rolGroup In t e rve ntion~roup
Control Group [.u:e nre' l.l;ionGroup
°!fote:thepret es tan d poetr ee.. _ansar ef~t.ta.nt1re••"'Pl e of 20ch i l d r e n 110f'jrou pl where•• thefollov- up ....an..lIcome frQOlonly14<:hild-r e nl7/g ro u pl.
