JOINT BOOK READING ACROSS CULTURES: A COMPARISON OF ...

3 downloads 0 Views 1MB Size Report
primarily result from differences in frequency of joint book reading expe- riences. ... pictures and text, suggesting that these types of digressive tactics are more.
JOINT BOOK READING ACROSS CULTURES: A COMPARISON OF SURINAMESE-DUTCH, TURKISHDUTCH, AND DUTCH PARENTCHILD DYADS Adriana G. Bus LEIDEN UNIVERSITY

Paul P.M. Leseman UNIVERSITY OF AMSTERDAM

Petra Keultjes LEIDEN UNIVERSITY

This article reports about in-depth analyses of how parents from different cultural groups mediated a simple narrative text to their 4-year-old children. The sample included 19 Surinamese-Dutch, 19 Turkish-Dutch, and 19 Dutch low-SES dyads. The sessions videotaped in the families' homes were transcribed and coded with a detailed behavioral coding system that maps the function and content of parent and child behaviors. In addition, the sessions were rated on four 7-point scales for how parents interacted with their children in terms of supportive presence. Overall the study supports the hypothesis that when reading is less important for the parents personally, they are less inclined to deviatefrom the text in order to negotiate meaning. Their children initiated more interactions than other children did, but lowcognitive-demand behaviors such as naming details or identifying pictures of characters characterized these interactions. The ethnic groups also differed in how parents interacted with their children, but these characteristics of the reading session were not related to parental literacy.

J LR V. 32 NO. 1 2000

PP. 53-76

Downloaded from jlr.sagepub.com by guest on August 21, 2016

has proven to be a major stimulating contribution to young children's reading and language development (Bus, van IJzendoorn, & Pellegrini, 1995; Sulzby, 1985), we still have to determine whether this contemporary trajectory of becoming literate can be generalized across cultures or ethnic groups (Fracasso, Lamb, Scholmerich, & Leyendecker, 1997). The research presented here explores the potential role of book reading across culturally divergent groups (e.g., Baker, Scher, &Mackler, 1997; Leseman & de Jong,i998).We focus on what may be keys to the literacy-stimulating function of book reading: textual changes and conversations accompanying the reading of text. Not only descriptive research (e.g., Bus & van IJzendoorn, 1995; Jones, 1996; Martin, 1998), but also experimental studies (Whitehurst et al, 1988) support the hypothesis that children's learning does not only depend on general affective support such as smiles and hugs, but also on the parental cognitive attempts to connect the familiar to the novel. ALTHOUGH PARENT-PRESCHOOLER BOOK READING

JLR Bus, Leseman, & Keultjes

Insofar as book reading sessions differ between cultural groups, this may primarily result from differences in frequency of joint book reading experiences. The parents' personal literacy skills may inhibit parents to read to their children. For example, four out of five African American mothers in Edwards' (1989) study were motivated to read to their children, but they did not read fluently themselves. With inexperienced children, earlier studies demonstrated much more elaboration and communication about the pictures and text, suggesting that these types of digressive tactics are more characteristic of infrequently reading dyads (Bus & van IJzendoorn, 1995; Martin, 1998; Pellegrini, Brody, & Siegel, 1985). Alternatively, immigrant parents may have different experiences with reading themselves, and they may also perceive parent-child book reading differently from low-socioeconomic status (SES) Dutch parents (Bus & Sulzby, 1996). In interviews, IOW-SES Dutch parents indeed emphasized joint book reading as a source of pleasure more than IOW-SES immigrant parents (de Groot & Bus, 1995). The mothers from ethnic minorities acknowledged the importance of reading narratives to young children. However, most of them stated that book reading is not so much for pleasure as for learning words or reading conventions. When parents do not perceive book reading as a source of entertainment reading, routines may not have developed to the same extent. Parents who are not heavily involved in literacy themselves may not know how to make books understandable and enjoyable even when stories are at a simple level, as in the present study (Baker et al., 1997; Serpell, 1997). Heath's (1983) ethnographic work has indeed suggested that IOW-SES African American caregivers do not use questions as much as IOW-SES White mothers. Using a quantitative methodology, subsequent studies (e.g.,

54 Downloaded from jlr.sagepub.com by guest on August 21, 2016

Anderson-Yockel & Haynes, 1994; Neuman, 1996) have shown that lowproficiency parent readers and their children were more likely to engage in book-focused interactions, compared with the interactions of more proficient parent readers who display more meaning-based interaction strategies. De Temple and Snow (1996) coded to what extent maternal utterances moved away from what could be seen on the page, including spontaneous connections to the child's own world. Although they reported quite a range in reading styles of the mothers, those with the lowest score on a scale that assessed the family's literacy involvement used very little non-immediate talk. When parents are less inclined to initiate discussions, children produce more spontaneous verbalizations, probably because merely reading the words of a text aloud is not sufficient to make the story comprehensible for inexperienced readers (e.g., Anderson-Yockel & Haynes, 1994). Child-initiated discussions are highly valued (e.g., Kaderavek & Sulzby, 1998; Pellegrini, Perlmutter, Galda, & Brody, 1990). We wonder, however, to which extent discussions initiated by inexperienced children may reveal meaning-based interaction strategies necessary for bridging the relations between the text and children's own experiences. Early parent-child interactions may also vary as a function of cultural background (Bornstein & Lamb, 1993; Bornstein, Tamis-LeMonda, Pêcheux, & Rahn, 1991; Field, Sostek, Vietze, & Leiderman, 1981; Richman et al., 19 8 8). Culturally divergent groups may differ in educational beliefs that result in different ways of responding to and supporting their children during activities such as book reading (cf. DeBaryshe, Huntley, Daley, & Rodarmel, 1992; Sonnenschein et al., 1996). From a study among the Surinamese-Dutch parents, for example, it appeared that these parents are generally more restrictive and discipline oriented than their Dutch counterparts. Observing free play situations, van IJzendoorn (1990) found that Surinamese-Dutch mothers showed less responsiveness to infants' signals. These parents consider obedience and respect for adults as important characteristics for their children and instill these values at an early age (see also Riksen-Wakaven,Meij,Hubbard,&Zevalkink, 1996).Van IJzendoorn suggested that Surinamese-Dutch parents waver between complete adaptation to the Dutch norms and their own somewhat stricter criteria for controlling the child's behavior. This ambivalence of complete adaptation versus adherence to their own culture may result in a less optimal, inconsistent child-rearing pattern (i.e., less responsive behavior), especially in culturally less integrated families. By carefully describing how parents from different cultural backgrounds deviate from the printed text to negotiate meaning for children and how they sustain children's curiosity and exploratory behavior during a reading

55 Downloaded from jlr.sagepub.com by guest on August 21, 2016

7 LR Joint Book Reading Across Cultures

JLK Bus, Leseman, 6-Keultjes

session, we may add to a better understanding of the psychological resources that parents bring to joint book reading (Fracasso et al., 1997). Composing groups of parents from different cultural backgrounds, we controlled for confounding variables such as economic and social-class differences. Economic and social-class differences have been widely documented as a source of differences in parenting style and in book reading (Ninio, 1980; Payne, Whitehurst, & Angelí, 1994; Snow & Ninio, 1986; Sonnenschein et al., 1996). The first group consisted of indigenous Dutch children and their parents. The other two groups were first- and second-generation immigrants from Suriname (South America) and Turkey (Europe). The present in-depth analyses of how these parents and their 4-yearold children shared a narrative provided by the examiner test the hypothesis that parental literacy is associated with deviations from the literal text in order to make reading a meaningful event. Textual changes and conversations accompanying the reading of text maybe less frequent when parents are not heavily involved in literacy. To better understand how any differences in reading behavior emerge, we also studied the dynamics of the scaffold by investigating which participants initiated or continued particular themes of discussion. Individual differences are embedded in sequences of other utterances, and it is only through analyses of interactive sequences that we can begin to understand organized social units such as book reading (Pellegrini et al., 1990). Another aim of the present study is to test the hypothesis that culturally specific child-rearing patterns explain differences in book reading. In addition to previous studies, we made an attempt to test the overlap between culturally specific child-rearing patterns and crosscultural differences in how parents scaffold joint activities such as book reading.

Method Participants The participants were selected from a larger sample collected by Leseman and his colleagues. The sample included Dutch, Surinamese-Dutch, and Turkish-Dutch parent-child dyads (Leseman & de Jong, 1998). Children were considered to belong to one of these ethnic groups if both parents of the mother were born in the Netherlands, Suriname, or Turkey. The final sample (N= 125) was recruited through 29 schools in the four major cities in the Netherlands. All participating schools were classified as educational priority schools with a high percentage of educationally disadvantaged students from Dutch lower social classes and ethnic minorities. The selected

56 Downloaded from jlr.sagepub.com by guest on August 21, 2016

schools without exception received extra governmental support because of their deprived population. The families of 4-year-old Dutch, Surinamese-Dutch, and TurkishDutch children were contacted through their teachers. A small booklet explained the research purposes. It was emphasized that the focus of research was school-related development. The booklet also explained that participating families would receive a gift voucher equivalent to u.s. $15. The positive response rates for Dutch, Surinamese-Dutch, and TurkishDutch families were respectively 56%, 54%, and 63%, which was satisfactory as these deprived populations are difficult to engage in research. In comparison to national family statistics, the recruited families were representative of their cultural groups (Leseman, Sijsling, Jap-A-Joe, & Sahin, 1995). To understand more fully the effects of the cultural background on book reading experiences in the family, we matched the three groups on educational attainment level and job status. We took the smallest group of the sample (Leseman & de Jong, 1998), the Turkish-Dutch, as the point of departure. For each Turkish-Dutch pair in the original sample (« = 24), we tried to select 1 Dutch and 1 Surinamese-Dutch dyad. For 5 of the lowest educated Turkish-Dutch families, there were no matches in the sample. The more educated Dutch and Surinamese-Dutch middle-class families were not represented, because there were no matches for these families in the Turkish-Dutch group. As appears in Table 1, the socioeconomic status of the families in the three groups was as similar as could be expected from the sampling procedure, F(2,54) = .16, ns, but the reading habits of the parents, F(z, 54) = 4.4, p < .001, r| 2 = .14, significantly differed; particularly reading of books, ^(2,53) = 4-i,p < .02, T|2 = .14, or periodicals for entertainment, F(2,53) = 5.5, p < .001, r|2 = .17. Post hoc testing (Tukey) revealed that the Dutch parents read significantly more books and journals for pleasure than the Turkish-Dutch, with the Surinamese-Dutch falling between these two groups. The parental reading proficiency was similar across groups; the number of miscues and the number of times that they read by sounding out words did not distinguish between the three groups (see Table 1). The three groups were similar in the children's ages, F(i, 52) = 1.3, ns, and sex, %2 (df = 2) = .7, ns (see Table 1). The home visits were scheduled around the time that children in the Netherlands enter kindergarten. The children were on average 4.3 years old. Procedure

During the home visit of a female researcher of the same ethnic origin, the

57 Downloaded from jlr.sagepub.com by guest on August 21, 2016

J LR Joint Book Reading Across Cultures

Table 1. Characteristics of Dutch, Turkish-Dutch, and Surinamese-Dutch Parent-Child Pairs Dutch (n = 19)

Characteristics Age (in months)

Turkish-Dutch Surinamese-Dutch (n = 19) (n = 78;

Univariate F

50.66 (3.16)

51.95 (3.73)

52.48 (3.77)

F(1,52) = 1.3

10/9

8/11

10/8

X2(2) = .7

SES

-.01 (.86)

-.15 (.92)

-.00 (.84)

F(2,53) = .2

Parental literacy13

1.82 (.66)

1.39 (.30)

1.64 (.29)

F(2, 53) = 4.3*

Reads instructive books'3

1.55 (.64)

1.22 (.34)

1.31 (.41)

F(2,53) = 2.4

Reads novels'"

1.66 (.71)

1.21 (.28)

1.37 (.35)

F(2,53) = 4.1*

Reads periodicals for information b

1.89 (.86)

1.53 (.44)

1.87 (.49)

F(2,53) = 2.0

Reads periodicals for entertainment 1 "

2.18 (.83)

1.58 (.40)

2.02 (.33)

F(2, 53) = 5.5**

Reading ability (miscues)

2.0 (3.0)

.9 (1.4)

1.1 (1.8)

F(2,53)=1.3

Reading ability (sounding out)

.5 (1.0)

2.8 (5.9)

.5 (1.2)

F(2,53) = 2.7

Sex (M/F) a

Post-hocc

D>T

D>T

D> T

a

z scores, bmaximum score: 3, Tukey. *pT/S

140 (57)

185 (107)

181 (116)

1.3

.82 (.13)

.59 (.11)

.71 (.22)

10.1***D/S>T

Naming0

.10 (.06)

Focus of conversations .07 (.04)

.15 (.10)

6.5** S>D/T

Parental part d

.80 (.27)

.75 (.26)

.77 (.20)

Parental initiative6

.96 (.11)

.50 (.15)

.23 (.05)

Parental partd

.10 (.07) .96 (.09)

.06 (.05) .95 (.08)

.07 (.05) .95 (.10)

Parental initiative6

.95 (.14)

Characteristics

Text-illustrationc

Surinamese-Dutch Univariate F(2,53)

Details of picture'

.26 (.09)

1.00 (.00) .26 (.20)

.92 (.29) .23 (.12)

Parental part d Parental initiative6 Own experiences 0

.74 (.19)

.55 (.30)

.74 (.33) .08 (.07)

.33 (38)

.69 (.21) .65 (.37)

.04 (.05)

.04 (.05)

.75 (.23)

.79 (.31)

.78 (.22)

.17 (.29) .17 (.09)

.00 .07 (.07)

.17 (.29) .08 (.07)

.86 (.19)

.83 (.28)

.82 (.28)

Parental initiative6 Paraphrasing0

.94 (.17) .09 (.05)

1.00 .22 (.15)

.75 (.43) .23 (.12)

Parental part d Parental initiative6

.91 (.25)

.88 (.22)

.88 (.19)

.92 (.29) .05 (.04)

.79 (.37) .21 (.15)

.88 (.25)

.90 (.23) .93 (.26)

.86 (.13)

.86 (.20) .84 (.27)

Parental part d Parental initiative6 Connections' Parental part d

Procedure 0 Parental part d Parental initiative6

.82 (.29)

.09 (.08)

2.3

.3

3.2* D>T/S

9.1*** D>T/S

9.1***T/S>D

12.6***T>D/S

Note. In case of significant F tests, a post hoc test was carried out (Tukey). a Total score on four 7-point scales; maximum score is 28. b Scores were corrected for differences in duration of the sessions; reported are numbers for 8 minutes. °Content variables represent the proportion of all utterances (by either child or parent) within one of the three samples that fall into the category. d The proportion of utterances that were spoken by the parent member of the dyad. e The proportion of exchanges that were initiated by the parent. *p < .05. **p < .01. ***p < .001.

66 Downloaded from jlr.sagepub.com by guest on August 21, 2016

group participated in a discussion about a particular theme or initiated such a discussion. One-way analyses of variance revealed that the number of discussion units accompanying the reading of text did not significantly differ across groups, F(2,53) = 1.9, ns, but there were differences in the number of textual changes, F(2,53) = 15.1, p < .001, rj 2 = .36. Post hoc comparisons (Tukey) revealed that the Dutch (D) parents were more inclined to use synonyms for words or to simplify phrases than the Turkish-Dutch (T) and the Surinamese-Dutch (S) (D > T/S). The focus of conversations also differed across groups. Pairs differed in naming of characters in illustrations, F(2,53) = 6.5,p < .003, T|2 = .20, discussing children's own experiences, F(2,53) = 3.2,p < .O5,T|2 = .n.making connections beyond the text, F(2,53) = 9.i,p < .ooi.T)2 = .26,paraphrasingtext,F(2,53) = 9.i,p < .001, T|2 = .26, and discussing the reading procedure F(2,53) = 12.6,p < .ooi,T|2 = .32. Post hoc comparisons (Tukey) revealed that the Dutch dyads paid more attention to connections going beyond the text (D > T/S) and focused more on children's own experiences (D > T/S), whereas the Turkish-Dutch and Surinamese-Dutch dyads often paraphrased text (T/S > D). The Turkish-Dutch dyads scored comparatively high on discussion of the procedure (T > D/S), and the Surinamese-Dutch on naming of characters in illustrations (S > D/T). The dynamics of the process may explain how these differences in cognitive demands come into being. The extent to which parents initiated conversation differed across groups, F(2,53) = 10.2,p< .ooi,T|2 = .28. As parents left initiatives for conversation to children, like the Turkish-Dutch parents were inclined to do (T < D/S), the overall demands made on the children were at a lower level than when the parents initiated the conversations about different themes. Table 2 shows that in each of the three groups, meaning-based interaction strategies that may help to understand the story (connecting the text to the illustration and making connections beyond the text) were primarily initiated by the parent. Parents also initiated paraphrasing the text and discussions about the procedure. Children, on the contrary, primarily initiated book-focused discussions about details of pictures and their own experiences related to their understanding of the book. Turkish-Dutch and Surinamese-Dutch children also initiated naming of main characters in the illustrations. In each group, parents had a main contribution to all discussions even when themes were initiated by the children. Results of the Discriminant Function Analysis

With the use of Wilks's criterion, the qualitative aspects of book reading

67 Downloaded from jlr.sagepub.com by guest on August 21, 2016

J LR Joint Book Reading Across Cultures

J LR Bus, Leseman, & Keultjes

were significantly affected by group membership, F(20,88) = 7.i8,p< .ooi, T)2 = .62.With the help of discriminant function analysis, differences among Dutch, Surinamese-Dutch, and Turkish-Dutch parent-child dyads were further explored. On the basis of the selection of 10 predictors that characterize the qualitative aspects of book reading (see Table 3), there was a reliable association between groups and predictors, %z (12) = 91.38, p < .0 01. After removal of the first discriminant function, reliable association remained, %2 (5) = 22.51, p < .001. The two discriminant functions accounted for 83.8% and 16.2%, respectively, of the between-group variability in discriminating among groups. As shown in Figure 1, thefirstdiscriminant function maximally separated Dutch and Turkish-Dutch dyads, with Surinamese-Dutch dyads falling between these two groups. The second discriminant function discriminated Surinamese-Dutch dyads from the two other groups. The loading matrix of correlations between the ten predictors and the two discriminant functions, as seen in Table 3, suggests that the best predictors (loadings of .33 and above) for the first discriminant function (distinguishing between Dutch dyads and Turkish-Dutch dyads) were textual changes, discussions about the procedure to be followed during book reading, discussions to make connections that go beyond the text, discussions of relations between the text and the illustration, and the part that moth-

Table 3. Results of Discriminant Function Analysis of Qualitative Aspects of the Book Reading Sessions Correlations of predictor variables with discriminant function Predictor variables

1

2

Univariate F(2,53)

p value

Supportive presence

-.14

.56

7.3

.004

Parental initiatives

-.36

-.02

10.1

.001

Textual changes

-.41

.39

15.1

.001

.39

.27

12.6

.001 .010

Procedure Naming

-.05

-.26

6.5

Text-illustration

-.33

-.02

2.3

Details

-.01

.14

.3

Own experiences

-.15

.17

3.2

Connections

-.37

.06

9.1

.001

9.1

.001

Paraphrasing

.28

-.45

Canonical R

.86

.60

Eigenvalue

2.91

.56

68 Downloaded from jlr.sagepub.com by guest on August 21, 2016

Canonical Discriminant Functions 3-

+ Group centroids

o 2-

• O

O

o

1 •

a Turkish

Q

o

o Dutch 3

%

a

O

A

O

^A S

oo -1 -

O

D

• Turkish

AA

0-

A Surinamese

D

o

.o

A

i

Surinamese A A -2 •

A

A A

-3 •

-4 • -6

1

1

A

1

-2

+2

+4

+6

Function 1 Figure 1. Plot of Three Group Centroids on Two Discriminant Functions Derived From Qualitative Aspects of the Book Reading Sessions

ers took in initiating new themes of discussion. Dutch parents (M = 4.4) made more textual changes than Turkish-Dutch (M = .0) and SurinameseDutch parents (M = .5). Dutch dyads were less inclined to discuss the procedure (AÍ = 5.6% of the discussion) than Turkish-Dutch dyads (M = 24.4%), with Surinamese-Dutch dyads scoring more like the Dutch than the Turkish (M = 9.9%). In the Dutch group (M = 17%), more discussions occurred to make connections that go beyond the text than in the Turkish-Dutch (M = 7%) and Surinamese-Dutch groups (M = 8%). The relation between text and illustration was more often discussed by Dutch dyads (M = 10%) than by Surinamese-Dutch (M = 7%) and Turkish-Dutch (M = 6%). In Dutch dyads, parents were more inclined to initiate the discussions (M = 82.2%) than in Turkish-Dutch dyads (M = 59.4%), with Surinamese-Dutch dyads scoring more like the Dutch (M = 71.1%). The primary predictors of the second discriminant function (separation of Surinamese-Dutch dyads from the other two groups) were supportive presence of the parent, paraphrasing, and textual changes. The Surinamese-Dutch dyads had a lower mean score on supportive presence

69 Downloaded from jlr.sagepub.com by guest on August 21, 2016

J LR Bus, Leseman, & Keultjes

(mean score on a scale ranging from 4 to 28 = 19.1) than the Turkish-Dutch (M = 21.4) and the Dutch group (M = 23.6). Similarly, the SurinameseDutch dyads were more inclined to paraphrase the text (M = 25.0% of all comments and questions) than the Dutch dyads (M=9.2%) with the Turkish-Dutchdyads(Ai=23.8%)verysimilartotheSurinamese-Dutchdyads. A comparison of actual group membership with group membership predicted with the help of the two discriminant functions reveals that from these pairs,50 (89.3%) were classified correctly, compared to 18.7 (33.3%) who would be correctly classified by chance alone. The Surinamese-Dutch were less likely to be correctly classified (77.8% correct classifications) than either the Dutch or the Turkish-Dutch dyads (both 94.7% correct classifications). The 77.8% classification rate for the Surinamese-Dutch dyads was achieved by classifying 2 dyads as Dutch and 2 as Turkish-Dutch, indicating that the interaction patterns in this group were more variable than in both other groups. Correlations Among Characteristics of the Reading Process Table 4 presents pooled within-group correlations among the predictors. Of the 45 correlations, 8 would show statistical significance at alpha equal to .01 when tested individually. (This significance level was chosen to protect against capitalization on chance.) Supportive presence is positively correlated with own experiences and connections beyond the text, and negatively with procedure, indicating that an overall higher score on supportive presence implies more questions and comments of the parent that may help to understand the story events.

Table 4. Pooled Within-Group Correlations Among Ten Predictors ; 1. Supportive presence 2. Parental initiatives 3. Naming characters 4. Text-illustration 5. Details of picture 6. Own experiences

-

2

3

4

5

.11

-.17

.10

-

-.25

.38* -.06

-

-.04

6 .41*

7

8

.38* -.23

9 -.40*

10 .05

-.11

-.15

.26

-.15

-.03

-.02

-.11

-.12

-.36* -.09

.05

-.17

-

-.22

-.08

-.13

.00

.01

-.06

-

-.08

-.06

-

7. Connections

-.57* -.38* -.05

.15

-.16

-.19

.04

-

-.20

-.35»

.13

-

8. Paraphrasing

-.01 -

9. Procedure

.15 -.09 -

10. Textual changes *p