Early Childhood Research Quarterly, 10, 317-337 (1995)

Age-Related Differences

in the Organization of Parent-Infant l~teractions During Picture-Book Reading

~~ni~ue S&zCchal

Carleton Unive~ity

Edward tf. Cornell

Lorri S. Broda

University of Alberta

These observations indicate how the organization of book reading events differs when middle- to upper-class suburban parents read picture books to preverbal and verbal infants. Twelve parent-infant dyads for each group of 9-, 17-, and 27-month-old infants were videotaped in their homes. On each of three visits, two different books were read. The books either contained sentences describing the illustrations or did not contain any sentences. The quality of parent verbalizations changed with the age of the infant; parents reading to younger infants used more attention-recruiting verbalizations and more elaborations, whereas parents reading to older infants used more questions and more feedback. Analyses of sequential dependencies be- tween categories of behaviors suggest that, across these age groups, parents monitor and attempt to maximize their infants attention to the book. Parents verbalizations expand from labeling comments, to sequences of labeling questions, to dialogues that exercise the growing linguistic com- petencies of the infant. Finally, interactions with books containing no sentences led to more verbal behaviors by the parent and more vocaliza- tions by the infant.

Many parents and their young children enjoy the interactions that occur when reading picture books. Children can also learn from book reading events. Correlational evidence suggests that young children learn language

Data collection and analysis were supported by a Social Sciences and Humanities Research Council of Canada grant to Edward H. Cornell. We thank Wanda Rowat and Jo&e Malan for their assistance with data transcription.

Correspondence and requests for reprints should be sent to Monique S&&hal, Department of Psychology, Carleton University, Ottawa, Canada KIS 586.

318 S&chal, Cornell, and Broda

from book reading events (for a review, see Scarborough & Dobrich, in press). For example, 2-year-old childrens responses to book reading events predict their vocabulary size 6 months later (Crain-Thoreson & Dale, 1992), and frequency of book reading is associated with vocabulary size at age 4 (Senechal, Monker, & Thomas, 1993). Experimental evidence confirms that 3- and 5-year-old children can learn vocabulary from listening to book reading events (Sdndchal, 1993; Se&ha1 & Cornell, 1993), and that 2-year- old childrens learning is particularly sensitive to the type of interactions that occur during book reading (e.g., Pemberton & Watkins, 1987; Valdez- Menchaca & Whitehurst, 1992; Whitehurst et al., 1988). Most of the research on the benefits of book reading, however, has been done with children who were at least 2 years of age. A recent study by DeBaryshe (1993) points to the need to examine book reading interactions with younger children. DeBaryshe examined book reading correlates of language development. She found that age of onset of picture-book reading (mean onset was 7 months) was a better predictor of language at age 2 than were variables such as frequency of reading sessions per week and number of stories read.

There have been a few observational studies to examine the role of book reading in very young childrens language development. Bruner (1983) noted that the earliest interactions with books appear to be highly regularized, enabling the infant to learn about picture-book conventions such as illustra- tion and naming. For example, Ninio and Bruner (1978) observed a single parent-child dyad over an S-month period starting when the child was 8 months old. They found that the mothers behaviors during book reading could be largely described as a four-step sequence: verbalizing to get atten- tion, asking a question, providing a label, and giving feedback. Moreover, this sequence seemed to be relatively invariant over the 8-month period. The lack of change in book reading practices during the childs second year may be specific to the mother dyad observed and may not generalize to other dyads,

Snow (1983) suggested that the repetition of certain verbal behaviors during book reading stimulates imitation and rote learning by the child, and ultimately independent use of the recurrent linguistic forms. The use of routines is clearly illustrated by Snow and Goldfield (1983), who reported multiple observations of one parent-child dyad during book reading. The dyad was observed over a period of 11 months starting when the child was 29 months old. The parent frequently used the same lexical items and the same linguistic constructions in response to illustrations in the book, and the child learned many of these. The results of this case study may be con- strained because the parent-child dyad were always reading the same alpha- bet book. Reading familiar books may lead to different kinds of interactions than reading novel books. Goodsitt, Raitan, and Perlmutter (1988) examined

Picture-Book Interactions 319

the issue of book familiarity on reading interactions and reported that mothers of 24-month-old toddlers talked more with unfamiliar books whereas the toddlers talked more with familiar books. Moreover, the mothers made more labeling comments (i.e., direct naming or requests for a name of an item depicted in book illustrations) with novel books than with familiar books.

Other observations of parents reading to their young children have indi- cated that parental book reading routines are not constant, but vary in accord with the age and competencies of the child. For example, Wheeler (1983) observed that mothers tended to describe pictures in books when their child was 17 months, whereas these mothers asked more questions from the same children a year later. Ninio (1980, 1983) found that mothers behaviors dif- fered depending on whether their 19-month-old child was verbally active or not. That is, passive nonverbal infants listened to mothers labeling pictures, more active nonverbal infants were asked where questions, and verbal in- fants were asked what questions. Taken together, these findings suggest that parents are very sensitive to their childs linguistic abilities, adapting their book reading behaviors accordingly. In fact, these findings are in contrast with Ninio and Bruners (1978) observation of an invariant book reading routine from preverbal to verbal infancy.

The purpose of the observational study presented here was to provide a microanalysis of the structure of picture-book interactions to capture potential differences in parental reading routines during the transition from preverbal to verbal infancy. Our focus was on describing verbal and non- verbal behaviors that occur during book reading. We included an age range that allowed us to assess whether the labeling routine described by Ninio and Bruner (1978) would generalize to other dyads, and whether the routine differs with growing language skills. For this study, we videotaped 12 parent- child dyads for each group of 9-, 17-, and 27-month-old infants. Multiple verbal and nonverbal behaviors from both the child and the parent were analyzed. Specific verbal behaviors were described in three ways. First, the function of utterances were described to assess potential differences in book reading routines. Specifically, we wanted to assess whether the organization of parental attempts to attract their childs attention to the book and paren- tal use of labels, questions, and feedback differed with the growing linguis- tic capacities of the children. We hypothesized that the labeling routine described by Ninio and Bruner (1978) would be characteristic of the young- est children only. In the Ninio and Bruner routine, the parent asked some- what rhetorical questions in that the parent provided the answers to the questions posed. Presumably, as childrens linguistic competencies increase, parents should ask questions that require children to provide the answers. In addition to these structural analyses, we examined the frequency of occurrence of each type of behavior. We hypothesized, based on Wheelers

320 Scinckhal, Cornell, and Broda

(1983) results, that parents descriptions of book elements would be replaced by parental requests for the child to label or describe the book elements. Marked differences were expected between each age group because of the age range selected. Third, we examined whether individual differences in childrens verbalizations were associated with differences in parental reading behaviors.

Children and parents exhibit a variety of nonverbal behaviors during book reading episodes. On the basis of pilot observations, the analysis was constrained to two nonverbal behaviors that were judged to be integral to interactions for children of the age groups studied. Childrens looking at the books was selected as a measure of attention and/or interest. Parents point- ing to the books was selected as a measure of parents controlling the childs attention. Previous research has shown that parents use pointing behaviors when labeling pictures in books for their 9- and lCmonth-olds, and also when making queries for labels with their 20- and 24-month-olds (Murphy, 1978). Sequential analyses of child looking and parental pointing behaviors allowed an assessment of how parents may use pointing to attract and sustain their childs looking behaviors. Murphys (1978) results suggested that these pointing behaviors are established early and maintained.

A final feature of this study is that we observed parent-child behaviors with a wider variety of materials than in previous studies. Each parent read six different picture books. The books selected were unfamiliar to the dyads because we were primarily interested in how parents adjust their book reading behaviors to their childs growing linguistic capacities (Goodsitt et al., 1988). The selection of unfamiliar books also allowed an examination of certain book characteristics that may promote interactions during book reading. Book characteristics may affect the frequency of parental verbal behaviors and may also affect learning from the child. Illustratively, Pellegrini, Perlmutter, Galda, and Brody (1990) found that expository texts (labeling and alphabet books) elicited more maternal teaching strategies than narratives did, and Cornell, Senechal, and Broda (1988) showed that 36-month-old childrens recall of events within two new picture books was related to differences in content. These results suggest that characteristics of the book may be important determinants of the parents practices and the childrens competence in book reading interactions.

Picture books were selected to vary according to one dimension that might be important in facilitating labeling behaviors from the parent, that is, the presence of text in picture books. Books either included short sentences or did not include sentences. We hypothesized that the presence of sentences might constrain readers to emphasize the printed information. The absence of sentences, however, requires that readers select the informa- tion to be discussed and the vocabulary and syntax of the remarks. Presum- ably, choices reflect the parents knowledge of their childs interests and

Picture-Book Interactions 321

language comprehension abilities. Thus, the presence of text might lead to a literal reading, whereas books without text might led to questions, responses, and feedback.

METHOD

Subjects We recorded observations of 12 parent-child dyads for each age group of 9-month-old (M=40 weeks, 5 days; SD= 3 weeks, 3 days), 17-month-old (M= 73 weeks, 5 days; SD = 2 weeks, 2 days), and 27-month-otd (M= 119 weeks, 1 day; SD=4 weeks, 4 days) chiIdren. Each age group had equal numbers of boys and girls. Families were White middle- to upper-class socioeconomic status with homes in the suburbs of Edmonton, Canada. Specifically, the Hollingshead (1975) Four Factor Index of Social Status was calculated to be a mean of 49.3 (SD= 10.0) for the families of the I)-month-olds, 43.7 (SD= 10.2) for the families of the 17-month-olds, and 46.4 (SD= 12.7) for the families of the 27-month-olds. The Hollingshead in- dex takes into account the education, occupation, sex, and marital status of the parents, and ranges from a low of 8 to a high of 66. Indices associated with medium status business employment, technical workers, and minor professional workers are in the range between 40 and 54.

Because we asked that the parent in each family who typically reads to the children participate in the study, fathers read the books in three dyads. Specifically, one father of 9-month-old twins, and one father of a 27-month- old girl read books. These three dyads were included because all their scored behaviors were within one standard deviation of the mean for each appro- priate age group. Other studies have shown that parent-child interactions vary more as a function of context (type of activity) than by gender of the parent (Golinkoff & Ames, 1979; Lewis & Gregory, 1987). Before scheduling observations, we read the titles of the six books to parents and asked whether each book was part of the childs personal library or had been previously used. Families that were familiar with any of the six books did not participate.

Material

Six cardboard picture books were selected because they were designed for rough handling and depicted events that could easily lead to conversations because the topics were known to infants. The books were not narratives, but were thematically organized. For example, one book depicted party clothes, cake, balloons, and presents. The first four double-spread pages of each book were read to equate the length of the six books. This modification was possible because the books were not narratives but included pictures of thematic~ly related objects or of families.

322 Sbkhal, Cornell, and Broda

The books differed in the presence or absence of sentences; three books included short sentences and three books did not. For example, the accom- panying text for a picture of a snowman was Snowman. A snowman to make, cold and white. A planned comparison was conducted to test whether the presence or absence of sentences affected book reading interactions. This contrast was conducted on five of the six books, that is, on the three books that contained sentences and the two books that did not. Although the data for the sixth book was included in the omnibus analyses of variance (ANOVAs), it was not included in the contrast because that book included four words. Book titles and their category membership are reported in Appendix, The planned comparison was conducted for the entire sample in the case of a significant book main effect but was conducted for each age group in the case of a significant Age x Book interaction. The Bonferroni inequality was used to adjust tests of significance to guard against Type I errors, and consequently, a conservative significance level of .001 was used for all book comparisons.

Note that the books selected were commercially available because we wanted to use books that are readily available to parents reading to young children. In selecting these books we varied, across the text dimension, other book characteristics that may influence book reading routines. For example, three books had bold and three books had soft colors, and three books con- tained many illustrated items and three books have few items illustrated. We realized that this selection does not control all potentially important book characteristics. Our intention was to indicate the generality of the effects of the presence or absence of sentences in a variety of common picture books. We sacrificed some control over the stimuli to gain ecological validity. We acknowledge that the claims we make about the different aspects of picture books will be exploratory in nature. Nonetheless, the findings will help direct more experimental research on book characteristics that promote parent-child interactions.

Procedure Each parent-child dyad was videotaped in their home on three occasions. During each visit, the dyad looked at two different picture books. The order of presentation of the six books was counterbalanced across dyads with the constraint that one book with text and one book without text be read at each visit. The order of presentation of the book with text and book without text within each visit was counterbalanced across dyads. Parents were in- structed to look at and read each book as they would do with any new book for their child.

The book reading episode was modified in two ways to enable the scoring of behaviors. First, parents were cued to turn a page every 30 s. Parents wore small headphones connected to a portable tape player. The tape con-

Picture-Book Interactions 323

sisted of short tones heard to signal the page turn. This procedure was neces- sary to allow page-by-page comparisons of behaviors within time frames of equal duration. The duration of book reading events would likely have varied across books, dyads, and age groups if we had not imposed this regularity. The headphones, however, could have distracted both parent and child from the book reading. To check this possibility, we conducted preliminary anal- yses to assess whether parents interrupted speech during the signal or whether children handled the headphones. Parents showed no evidence of disruption during the brief signal and there was one occasion where one child acciden- tally pulled the headphone wires.

The second modification of the book reading episode was the presence of a small videotape recorder. The child was positioned on the parents lap on a couch, and the videotape recorder was positioned on the floor between and in front of the parents feet. A zoom lens was adjusted so that the view- finder image included the bottom of the book on the bottom of the frame, the parents arms on the sides, and the top of the childs head at the top. This ~rangement was necessary to allow a close record of pertinent behaviors and verbalizations. Participants were able to interact without distraction because the camera was below their gaze. We waited until parents and children became familiar with the equipment before starting each reading episode. The presence of videotape equipment, however, could have affected the parent and child behaviors. To verify this possibility, we hypothesized that if videotapes had a distracting effect, this effect would be most evident on the first visit. Presumably, parents and children would be more familiar with the equipment and procedure by the third visit. Preliminary analyses were conducted to assess variations in parental utterances and children vocalizations across visits. These analyses failed to reveal any significant variation across visits, all Fs< 1.68.

Verbal behaviors. Written protocols, transcribed from videotapes, were scored to index the function of parental utterances. For these analyses, an utterance was defined as a sequence separated from other speech by a pause, or as a grammatically complete string of words (Golinkoff & Ames, 1979). The functional analysis consisted of classifying each parental utter- ance into one of four categories: attention-recruiting, elaborations, question, or feedback. Utterances consisting of reading the text in books were excluded from the functional analysis to allow an index of parent-child interactions that went beyond the print. However, it can be noted here that additional analyses examined the amount of talk by including both the text read and the functional utterances considered here. An attention-recruiting @I) ver- ~~~iza~~o~ was defined as an utterance used to attract the childs attention to

324 S&&hal, Cornell, and Broda

Table 1. Examples of Categories of Verb~izations

Category

Attention-recruiting (A)

Elaboration (E)

Question (Q)

Child vocalization (V)

Feedback (F)

Example

Oh! Look at this! See the pull toys?

This is a ball, Hes inside the jacket,

or Thats her favorite toy.

Whats that? Can you show me a grandpa?

Do you have teeth like that?

Ball, Green, I eat cake, Thats mommy.

Yes, thats right. Red, good girl,

or No, its a bail.

the book or a particular aspect of the book. An elaboration (E) was a label, a description, or a comment. A question (QI was an utterance requesting some kind of response, either from the child or from oneself (that is, parents may ask questions and provide the answers). included in the feedback (F) category were utterances made in response to ones own question (e.g., when child fails to answer) or to a childs vocalization. Specifically, included in the feedback category were answers (to ones questions or to a childs questions), praises, imitations, corrections of the childs vocalizations, or comments about the childs vocalizations. Other parental utterances that followed a child vocalization but were not related to that child vocalization were coded in their appropriate category (i.e., either an attention-recruiting, elaboration, or a question). Any vocalization (V) by the child was coded as such; these included vocalizations made by the nonverbal infants, words used to label book illustrations, responses to questions, questions, and com- ments. Examples of each category are provided in Table 1,

The classification used in this study was based partly on Ninio and Bruners (1978) description, with the exception of elaborations. The elaboration cate- gory replaced the labeling category used by Ninio and Bruner because many parents made comments about book content that were more general that a label. Thus, the elaboration category included labeling comments (e.g., Its a ball) and more general book-related comments, for example, comments that try to make links between the book content and the childs experience (e-g., You have a ball just like that). All protocols of verbalizations were coded by one principal research assistant, with a second research assistant coding an overlapping 10% of the data. Kappa values for these behaviors ranged from .80 to .97, with a mean of .91. Disagreements were not resolved; the data recorded by the principal research assistant were used in the analyses.

~o~verbul behaviors. Childrens looking at the books was selected as a measure of attention and/or interest. Looking was defined when the head and eyes of the child were judged to be aligned within the edges of the book cover. Parents pointing to the books was selected as a measure of parents

Picture-Book Interactions 325

controlling the childs attention. Pointing was defined when the index finger was extended and used to approach or held on portions of the page. Video- tapes were scored for each behavior independently. Scoring was done by one principal research assistant. Reliability was established by having a second assistant score 24 min of randomly selected tapes; mean percent agreement on the 1,440 l-s intervals scored was 89% for looking behavior and 84% for pointing behavior. Disagreements were not resolved; the data recorded by the principal research assistant were used in the analyses.

Sequential Dependencies Sequential dependencies analyses were conducted to test whether a specified behavior (criterion) was followed (or preceded) by another specified behavior (target) more frequently than would be probable in a model in which all behavioral categories occur randomly. Two types of analyses were con- ducted. First, verbal utterances were coded in a sequence of discrete events without a time base to assess the structure of book reading episodes. Second, parent pointing and child looking were coded within l-s time intervals to assess whether the onset of parental pointing is followed (or preceded) by the onset of child looking more frequently than would be expected by chance.

Sequential verbal behaviors were analyzed using a version of Bakemans (1983) ELAG program. Changes in the quality of verbalizations were the focus of these analyses. That is, the ELAG analyses are directed to adjacent categories of utterances that are not identical; a category may not follow itself in these behavior-sequence analyses. Each category or verbalization was used as a criterion and the remaining categories were used as targets. In all analyses, data were aggregated for each age group and across books because the contingency tables for each dyad and each book produced cells of insufficient size for the calculation of reliable z scores (for a discussion on the adequacy of aggregated data, see Allison 8z Liker, 1982, and Moran, Krupka, Tutton, & Symons, 1987). The advantage of using sequential analyses is that variations in base rate occurrences of behaviors are con- trolled. Sequential analyses between nonverbal behaviors were conducted with Symons, Wright, and Morans (1988) TLAG program. This program allows the analysis of parents pointing and childrens looking as two con- current streams of behaviors. Statistical reliability was assessed with the Allison-Liker z-score test (Allison & Liker, 1982), using Bonferronis cor- rection (for a discussion on the selection of appropriate z-score tests, see Gottman & Roy, 1990).

RESULTS

Three dyads accidentally skipped one page during observation of reading episodes. These missing data points were replaced by the average perfor- mance for both the appropriate book and age group.

:326 S&&hal, Cornell, and Broda

Verbal Behaviors

Reading Routines. We analyzed the sequential occurrence of verbal behaviors to assess possible dependencies among the categories listed in Table 1. Each type of verbalization was used as a criterion to examine whether there were any transitional dependencies between its occurrence and the occurrence of the remaining types of verbalizations. The analysis yielded mean z scores that were tested against a model of random event sequences (Symons & Moran, 1987). The significant patterns are represented as state transition diagrams in Figure 1.

The pattern for 9-month-old infants was quite simple; parents drew the childs attention to the book and labeled or described some aspects of the illustrations. Parents also gave feedback to the occasional vocalizations made by their infants. The pattern for 17-month-old infants was best char- acterized as the following progression: Parents began by alternating between attention-recruiting and elaboration of a topic, then posed a question; this was followed by the child making a response to the question, and the parent providing feedback to the childs response. A similar pattern appeared to constitute the initial portion of the sequence of interactions with 27- month-olds. However, the latter portion of the interactions was more vari- able because feedback from the parent could be followed by one of three behaviors: another response from the child, another question from the parent, or an elaboration. In other words, interactions with the 27-month- olds progressed to a dialogue.

Type of Verbalizations. Preliminary ANOVAs failed to reveal significant variations in performance associated with the gender of the child. Thus, gender was omitted from further analyses to increase the cell size for tests of age and book effects. Each dependent measure was analyzed with 3 (Age) x 6 (Book) x 4 (Page) mixed ANOVAs unless otherwise indicated. Analyses omitting the three fathers yielded results that were essentially the same as analyses including fathers, therefore, the analyses including the entire sample are reported here. Means for all verbal behaviors are reported in Table 2, and significant results revealed by the ANOVAS are reported in Table 3. Only significant effects, interactions, and contrasts are reported. Contrasts are significant at p< .OOl unless specified otherwise.

First, we analyzed each category of parental utterances and child vocali- zations. The number of attention-recruiting verbalizations and of elabora- tions made by the parent differed with the age of the child. Parents of 9-and 17-month-olds did not differ in number of attention-recruiting verbali- zations and elaborations (Fs< l), but parents of the 17-month-olds made more attention-recruiting comments and more elaborations than parents of the 27-month-olds, Fs(1, 33) = 12.95 and 9.54, ps

327

328 S&rhai, Cornell, and Broda

Table 2. Mean Number of Verbal Responses (and Standard Deviations) for each Age Group

Categories of Verhali7ations

Parent Child

Attcntion- Av Utterance Kerruiting ~labora:ion Question Fcrdhack \ocalization

9 months 37.4 (14.3) I 1.2 (4.4) 17.2 (4.9) 8.i (5.0) 1.1 (0.9) 1.2 (1.3) 17 months 47.1 (12.4) 1 I .9 (6.3) 18.8 (6.7) 13.8 (7.0) 5.6 (3.5) 6.6 (5.9) 27 months 41.9 (10.4) 4.1 (2.4) 9.3 (2.2) 19.3 (2.3) 9.5 (2.5) 17.6 (4.9)

Table 3. ANOVA Results for the Verbal Behaviors

Dependent Measure Effect/Interaction df F MS! P

Attention-Recruiting

Age Book

Elaboration

Age Book Age x Book Book x Page

Questioning

Age Book Age x Book Book x Page

Child Verbalization

Age Book Age x Book

Feedback

Age Book Age x Book Page

(2, 33) 9.79 37.17

(5, 165) 7.10 2.89

(2, 33) 5.64 85.07

(5, 165) 3.75 6.81 (10, 165) 3.21 6.81

(6, 99) 3.98 3.95

(2, 33) 14.74 41.68

(5, 165) 13.58 5.24

(10, 165) 3.56 5.24

(15, 495) 2.38 3.18

(2, 33) (5, 165)

(10, 165)

42.38 29.92 .OOl 8.36 3.29 .oOl 4.85 3.29 .ool

(2, 33) 32.74 9.71

(5, 165) 4.89 1.86 (10, 165) 2.28 1.86

(3, 99) 4.16 1.52

.Ol

.ool

.Ol

.Ol

.OOl ,001

.OOl

.OOl

.ool

.Ol

.ool ,001 .05 .Ol

describe a significant Age x Book interaction for the number of elaborations. Parents of 9-month-olds elaborated more when reading no-text books (M=20.5) than books with texts (M= 15.2), F(1, 165)=66.1O,p< .OOl. The number of elaborations changed as a book was read, as indicated by an Age x Page interaction.

Picture-Book Interactions 329

Frequency of questioning increased with age: Parents of 9-month-olds asked fewer questions than parents of 17-month-olds, and parents of l7-month-olds asked fewer questions than parents of 27-month-olds, &(I, 33) = 6.34 and 8.38, ps< .05, respectively. The planned comparison revealed that parents of 9-, and 27-month-olds asked more questions when using no-text books (MS = 11.2 and 20.8, respectively) than books with texts (Ms=5.0 and 15.8, respectively), Fs(1, 165)=105.75 and 67.15, respec- tively. There was also a significant Book x Page interaction.

Because questioning presses for thought and expressive language, we also recorded the type of questions parents asked. Parents usually sought labels, asking what- on 74% (MS= 81%, 64%, and 77% for 9-, 17-, and 27-month- olds, respectively) of the questions posed. Where- questions constituted 13% (MS= 6%, 24%, and 8% for 9-, 17-, and 27-month-olds, respectively) of the interrogatives, and who- questions comprised 12% (MS = 13%, I l%, and 13% for 9-, 17-, and 27-month-olds, respectively). The remainder were why-, how many-, or which- questions.

The number of vocalizations made by children increased profoundly across the age groups: Nine-month-old infants made fewer vocalizations than 17-month-olds, and 17-month-olds made fewer verbalizations than 27-month-olds, Es{ 1, 33) = 8.89 and 36.65, ps < -01, respectively. There was also a significant book main effect and an Age x Book interaction. Twenty- seven-month-olds talked more when examining books without text (M= 21.9) than books with texts (M= 14.1), F(1, 165)=264X%, p< .OOl.

The number of verbalizations providing feedback to childrens utterances or serving as a response to a parental question also increased with age, Parents of I)-month-olds gave less feedback than parents of the 17-month- olds, and parents of 17-month-olds gave less feedback than parents of 27- month-olds, Fs(1, 33) = 14.84 and 14.41, ps c .001, respectively. The amount of feedback also varied with book. Moreover, there was a significant Age x Book interaction. Parents of 27-month-olds gave more feedback when dis- cussing books without text (M= 10.9) than when discussing books with texts (M=8.3), F(1, 165) = 53.17, p< .OOl. Finally, there was a significant page main effect. Mean number of feedback verbalizations were 1.5, 1.4, 1.1, and 1.4 for the first to the last page, respectively.

Second, we conducted an additional analysis to verify that the observed advantage of no-text books was not an artifact of our procedure. In this analysis, we calculated total talk by adding the parental utterances, the sentences and words found in the books, and the childrens utterances. This analysis confirmed that some dyads showed more verbosity when reading some of the books than others. The planned comparison indicated that dyads including 9- and 27-month-olds talked and vocalized more when inter- acting with no-text books (MS = 44.2 and 69.9, respectively) than books

330 Skkhal, Cornell, and Broda

Table 4. Mean Number of Child Vocalization as a Function of Vocalization Level and Age Group

Vocalization Level

Age High LOU

9 months 2.0 0.2

17 months 10.8 2.3

27 months 21.0 14.1

with texts (MS =40.8 and 58.5, respectively), F(I, 165) = 11.88 and 134.07, p < .OOl , respectively.

Individual differences. Here we examined whether parental interactions differed as a function of childrens linguistic proficiency within each age group. This analysis was done by dividing the sample into two groups on the basis of childrens amount of vocalizations. Specifically, children in each age group were assigned to a high vocalization or a low vocalization group based on a median-split of the amount of vocalizations for each age group, The mean number of vocalizations for each vocalization group are reported in Table 4. Parental verbalizations were then analyzed using 3 (Age) x 2 (Vocalization Level) ANOVAs.

There were no main effects or interactions with vocalization level for the attention-recruiting verbalizations, nor for the elaborations. The amount of questioning and feedback, however, differed for each vocalization group. The ANOVA revealed a significant main effect of vocalization level for questioning, F(1, 30)= 1.15,MSe=25.06,p< .05. Parents of high vocaliza- tion children asked more questions (M= 15.1) than parents of low vocaliza- tion children (M= 11.6). Similarly, parents of high vocalization children provided more feedback (M= 6.4) than parents of low vocalization children (M=4.4), F(1, 30)= 7.38, M&=4,71,p< .Ol. There was also a significant Age x Vocalization Level interaction, F(l, 30) = 3.98, MSe = 4.71, p < .03. Tukey post hoc tests revealed significant differences for the 17-month-olds only: For this age group, parents of high vocalization children gave more feedback (M= 7.9) than parents of Iow vocalization children (M= 3.2). In sum, the distinction between high and low vocalization groups is the parental usage of questions and feedback.

Nonverbal Behaviors

Reading Routines. Sequential analyses were conducted to assess the relationship between the nonverbal behaviors. We first tested whether the onset of pointing to a page of a book by a parent was followed reliably by the child looking at the page. Next, we tested whether the onset of looking at a page by the child was followed reliably by a point to the page by the parent.

Picture-Book Interactions 331

Pointing may be used to recruit the attention of the child. Analyses revealed that typically when a parent extended the index finger to the page, a child who was not oriented to the book was judged to be looking at a 2-s lag. Mean z scores for the 9-, 17-, and 27-month-olds were 6.31,6.23, and 7.19, respectively. These z scores were all significantly different from zero (zcrit;cOl = 3.20); z scores for the l- and 3-s lags were not significant.

During the book reading interactions, pointing to illustrations that the parent considers to be topical may follow the infants attention to a page. In fact, parents of children in the three age groups were very responsive to their childrens looking behavior; typically when their child began to look at a page, the parents quickly (l-s lag) responded with a point. Mean z scores for the 9-, 17-, and 27-month-olds were 6.09, 6.04, and 7.12, respectively. These z scores differed significantly from zero (z~~;,;~~I= 3.20); z scores for the 2- and 3-s lags were not significant.

Childrens Looking. Children could look at each double-spread page for 30 s for a total of 120 s per book. Analyses revealed a significant age main effect in the duration of looking behaviors, F(2, 33) = 12.17, IV&= 600.41, p< .OOl. Planned contrasts revealed that the 9-month-old infants (M= 66.3 s) looked less at the books than the 17-month-old infants (M= 87.5 s) and, in turn, the 17-month-old children looked less at the books than the 27-month- old children (M= 106.5 s), Fs(1, 33) = 6.78 and 5.41, respectively, ps < .05.

Looking also decreased as the reading progressed, as indicated by a sig- nificant page main effect and a significant interaction of age with page, F(3, 99) = 9.5 1 and F(6, 99) = 6.07, respectively, MS = 21.12, ps < .OOl . An examination of simple main effects revealed a significant page effect for the youngest children only, F( 1, 99) = 20.79, p < .OOl . No other effect or inter- action was significant.

DISCUSSION

The findings reported here are especially important because this study included an age range not previously examined with a sample size of this magnitude. The findings replicate previous case studies by showing that parent-child interactions during book readings are progressive routines. The findings extend the case studies by showing that the routines occur when parents read new unfamiliar books to their children, and more impor- tant, the findings show a clear framework for expansion of reading from text. Interactions with 9-month-old infants were discontinuous, consisting of the parent attempting to draw the infants attention to the book and then proceeding to describe the illustrations. Although parents reinforced their 9-month-old infants vocalizations, parents did not reliably call for a label. Parents of older children, however, included questions in the reading routines.

332 S&k hal, Cornell, and Broda

When parents read books with their 17-month-olds, the routine included an initial cycle consisting of attention-recruiting, an elaboration, and often prior to posing a question, another attempt to recruit attention and elaborate features of the book. The initial cycle involving only the parent was followed by linear turn-taking: a question, followed by the childs response, followed by feedback. The questions were predominantly what-, and the infants responses and the subsequent confirmations or corrections by the parents were as described by Ninio and Bruner (1978). However, verbalizations given by 27-month-old infants, compared to those of 17-month-old infants, were associated with a qualitative shift in the pattern of verbal interactions: Only the older infants reactions to their parents questions lead parents to further the discussions. The parents feedback may consist of elaborations or other questions that again encourage responding from the infants.

The three routines described here seem to indicate that parent-child inter- actions are unidirectional in that parents decide what information to high- light and children simply respond to their parents questions. We contend that this interpretation of the directionality of interactions is only partly supported by the results for two reasons. First, consider the nature of sequen- tial analyses. The sequences described indicate the dependencies between behaviors. Thus, the results show that 17- and 27-month-old infants responded to their parents queries. This sequence, however, does not exclude occasions where children initiated cycles by making comments that were not prompted by queries. That is, children could have made comments that were not initi- ated by their parents questions. The sequences show, however, that when children spoke, their parents reinforced them. Second, an examination of the parents pointing behaviors also suggests that parents were sensitive to their childs responses to the book. The record of onset of pointing behaviors indicated that parents were monitoring infants interest in the book. When infants initiated looking to the book, parents followed with a point to a feature on the page within 1 s. These rapid sequences occurred reliably in observations of all age groups, suggesting that the functions of pointing are both established with preverbal infants and maintained with verbal infants. Parents are sensitive monitors of their infants attention, using points to follow and possibly sustain their infants attention. This finding is important in light of previous research showing that parents encouragement of attention during play with their Smonth-olds explained unique variance in the lan- guage comprehension of their children at 13 months (Tamis-LeMonda & Bornstein, 1989).

The pattern of interaction for the 9-month-old children was different than that noted by Ninio and Bruner (1978) who described reading cycles by a mother and son over longitudinal observations. When the boy was 8 l/2 months of age, they recorded that the mothers behaviors during book read- ing could be largely described as a four-step sequence: verbalizing to get

Picture-Book Interactions 333

attention, asking a question, providing a label, and giving feedback. In this study, parents reading to their 9-month-olds did label the illustrations in the books, but labeling queries were not likely. Two alternatives exist to explain this discrepancy in findings. First, it is possible that the labeling sequence described by Ninio and Bruner (1978) was specific to the dyad observed and does not generalize to other parents reading to very young infants. These parents may not favor labeling questions when interacting with nonverbal infants. In fact, in this sample, it was parents interacting with verbal 17-month-olds who showed the labeling sequence. Second, it is possible that our results were obtained because parents were reading unfamiliar books. Ninio and Bruner listed comments by the mother in their dyad which express that she knew the lexical items her son might produce. These comments in- dicate that the books had been used before and labeling questions were posed and answered by her in anticipation of her sons responsiveness. Pre- vious research has shown, however, that mothers of 24-month-olds ask more labeling questions when reading unfamiliar books compared to familiar books (Goodsitt et al., 1988). Book familiarity may have a different impact on book reading routines depending on the age of the child.

In addition to age-related differences in the book reading routines, there were changes in the frequency of occurrence of parental book reading behaviors, Parents reading to 9-month-olds used more attention recruiting verbalizations and more elaborations, whereas parents reading to 27-month- olds used more questions and more feedback. The interesting developmental implication here is that parents reading to 17-month-olds used all these behaviors fairly frequently, suggesting a transition period in which parents still use behaviors to attract their infants attention, but also introduce ques- tions into the book reading routine. This is an example of how the cross- sectional study of age-related differences can suggest phenomena that have not been previously described.

Children made more and more verbalizations with age. The important findings, however, were the individual differences in childrens amount of verbalizations across age groups. We found that children who were asked more questions and given more feedback made more verbalizations than children who were asked fewer questions and given less feedback. This find- ing has applied implications for pracitioners and researchers interested in pin~inting how picture-book reading events influence language develop- ment. For example, most of the intervention programs conducted to increase language development through picture-book reading are designed to enhance broad changes in parental reading behaviors (e.g., Arnold, Lonigan, White- hurst, & Epstein, 1994; Whitehurst et al., 1988). The results of this study suggest that intervention programs that focus on just a few behaviors, such as questioning and the use of feedback, may also be effective in producing increased verbal participation from children.

334 Scinirhal, Cornell, and Broda

With age, children also spent more time looking at the books. The young- est children, on average, looked at the books for 55% of the duration of the reading episode. The 17-month-olds looked at the books for 73% of the time, and the 27-month-olds looked at the books 88% of the time. In addi- tion, the youngest children looked at the book less and less as the reading progressed. This indicates that with age, children not only were able to maintain their attention for longer periods of time but also started to appre- ciate the routine of book reading.

Finally, we found consistent indications of the importance of certain book features as a stimulus for differential verbal behavior. Of particular interest were results that suggest advantages of book formats for encouraging verbal behaviors. The most novel effect was that interactions with books that contained no text led to more verbal behaviors by parents and more vocalizations by infants. Picture books that do not contain any text may promote verbal interactions because parents are free to select which infor- mation to emphasize. For example, a parent decided to concentrate on iden- tifying colors in the no-text book portraying family members. This decision produced numerous requests to identify colors and, consequently, numerous responses by her 27-month-old child. The other book portraying family members included sentences. The same parent read the text, made a few comments about the information in the text, but limited the number of questions asked.

All the parents in this sample read the text in the books. There are several possible reasons why parents may stick to topics and language that is speci- fied in print. Heath (1982) suggested that respect for the words of authors may lead parents to read veridically. It is also possible that the parents in this study felt that, because they were participating in a study, they should read more formally. The rapidity with which parents responded to their pre- verbal infants participation and the elaborations and conversations parents initiated with verbal infants suggests that this was not a serious problem in our home studies.

This finding has an applied implication. There is a strong assumption in the literature that parents stimulation and encouragements have a positive impact on young childrens language development. Many intervention pro- grams are constructed to teach parents specific ways to interact with their children. For example, Whitehurst and his colleagues showed that when parents were trained to encourage more verbal participation during book reading, 2-year-old children made greater expressive vocabulary gains than children whose parents were not trained (Whitehurst et al., 1988). Training parents, however, may not always be feasible. Whitehurst tried to circumvent this problem by modifying the text in books to encourage parents to involve their children during book reading (Whitehurst et al., 1989). The books were modified by incorporating questions in the narrative of books. This manipu-

Picture-Book Interactions 335

lation of the text in books failed to produce greater gains in vocabulary acquisition (but, for effects of videotape training programs, see Arnold et al., 1994; and for effects of daycare training programs, see Valdez-Menchaca & Whitehurst, 1992). The results of this study suggest that one way of en- couraging parents to involve their infants and toddlers during reading would be to use books without any text. This possibility can be tested empirically. Moreover, the use of picture-books without any text may be of value for practitioners who work with parents with culturalIy unique linguistic skills or who work with parents who have very low literacy levels (Dickinson, 1994).

REFERENCES

Allison, P.D., & Liker, J.K. (1982). Analyzing sequential categorical data on dyadic interac- tion. Psychological Bulletin, 91. 393-403.

Arnold, D.H., Lonigan, C.J., Whitehurst, G.J., & Epstein, J.N. (1994). Accelerating language development through picture-book reading: Replication and extension to a videotape training format. Journal of Educational Psychology, 86, 235-243.

Bakeman, R. (1983). Computing lag sequential statistics: The ELAG program. Behavior Research method & ~~trumentation, 15, 530-535.

Bruner, J.S. (1983). The acquisition of pragmatic commitments. In R. Golinkoff (Ed.), The transition from prelinguistjc to hnguistic communicatjon. Hiilsdale, NJ: Erlbaum.

Corbett, G. (1982). Guess who? New York: Dial Press. Cornell, E.H., Se&chat, M., & Broda, L.S. (1988). Recall of picture books by 3-year-old

children: Testing and repetition effects in joint reading activities. Journal of Educa- tional Psychology, SO, 531-542.

Crain-Thoreson, C., & Dale, P.S. (1992). Do early talkers become early readers? Linguistic precocity, preschool language, and emergent literacy. Developmental Psychology, 28, 421-429.

Curry, P. (1983). Play. Surrey, England: Worlds Work. DeBaryshe, D.B. (1993). Joint picture-book reading correlates of early oral language skill.

Journal of Child Language, 20, 455-461. Dickinson, D.K. (1994). Bridges to literacy: Children, famihes, and schools. Cambridge, MA:

Blackweft. Golinkoff, R.M., & Ames, G.J. (1979). A comparison of fathers and mothers speech to

young children. Child ~ve~opment, SO, 3842. Goodsitt, J., Raitan, J.G., & Perlmutter, M. (1988). Interaction between mothers and pre-

school children when reading a novel and familiar book. International Journal of Behavioral Development, II, 489-505.

Gottman, J.M., & Roy, A.K. (1990). Sequential analysis: A guide for behavioral researchers. New York: Cambridge University Press.

Heath, S.B. (1982). What no bedtime story means: Narrative skill at home and school. Language in Society, I, 49-76.

Hillman, P. (1983). The merry-mouse book of toys. New York: Doubleday. Hollingshead, A.B. (1975). Four factor index of social sfutus. Unpublished manuscript, Yale

University, Department of Sociology, New Haven, CT. Khalsa, D.K. (1984). Happy birthday, Baabee. Plattsburgh, NY: Tundra Books. Ladouceur, F. (1982). My first words: My family. Tokyo: Les Editions Heritage.

336 Shnichal, Cornell, and Broda

Lewis, C., & Gregory, S. (1987). Parents talk to their infants: The importance of context. First Language, 7, 201-216.

Moran, G., Krupka, A., Tutton, A., & Symons, D. (1987). Patterns of maternal and infant imitation during play. Infant Behavior and Development, IO, 477-491.

Murphy, C.M. (1978). Pointing in the context of a shared activity. Child Development, 49, 371-380.

Ninio, A. (1980). Picture-book reading in mother-infant dyads belonging to two subgroups in Israel. Child Development, 51, 587-590.

Ninio, A. (1983). Joint book reading as a multiple vocabulary acquisition device. Child Devel- opment, 79, 445-45 1.

Ninio, A., & Bruner, J.S. (1978). The achievement and antecedents of labefing. Journal of Child Language, 5, l-l 5.

Oxenbury, H. (1981). Family. New York: Wanderer Books. Pellegrini, A.D., Perlmutter, J.C., Galda, L., & Brody, G.H. (1990). Joint reading between

Black Head Start children and their mothers. Child Development, 61, 443-453. Pemberton, E.F., & Watkins, R.V. (1987). Language facilitation through stories: Recasting

and modeling. First Language, 7, 79-89. Scarborough, H.S., & Dobrich, W. (1994). On the efficacy of reading to preschoolers. Devel-

opmental Review, 14, 235-302. Sinechal, M. (1993, March). Preschoolers expressive and receptive vocabulary development.

Poster presented at the biennial meeting of the Society for Research in Child Develop- ment, New Orleans.

Senechal, M., & Cornell, E.H. (1993). Vocabulary acquisition through shared reading experi- ences. Reading Research Quarterly, 28, 261-272.

Sdnechal, M., Monker, J.A., &Thomas, E. (1993, May). Individua~differences in vocabulary acquisition. Poster presented at the annual meeting of the Canadian Psychological Association, Montreal

Snow, C.E. (1983). Literacy and language: Relationships during the preschool years. Harvard Educational Review, 53, 165-189.

Snow, C.E., & Goldfield, B.A. (1983). Turn the page please: Situation-specific language acquisition. Journal of Child Language, IO, 551-569.

Symons, D., & Moran, G. (1987). The behavioral dynamics of mutual responsiveness in face- to-face mother-infant interaction. Child Development, 58, 1488-1495.

Symons, D., Wright, R.D., &Moran, G. (1988). Computing lag sequential statistics on dyadic time interval data: The TLAG program. Behavior Research, Methods, Instruments, and Computers, 20, 343-346.

Tamis-I,eMonda, C.S., & Bornstein, M.H. (1989). Habituation and maternal encouragement of attention in infancy as predictors of toddler language, play, and representationa competence. Child Devefopment, 60, 738-75 1.

Valdez-Menchaca, M.C., & Whitehurst, G.J. (1992). Accelerating language development through picture book reading: A systematic extension to Mexican day care. Developmental Psychology, 28, 1106-l 114.

Wheeler, M.P. (1983). Context-related age changes in mothers speech: Joint book reading. Journal of Child Language, IO, 259-263.

Whitehurst, G.J., Falco, F.L., Lonigan, C.J., Fischel, J.E., DeBaryshe, B.D., Valdez- Menchaca, M.C., & Caufield, M. (1988). Accelerating language development through picture book reading. Developmental Psychology, 24, 552-559.

Whitehurst, G.J., Fischel, J.E., Arnold, D., Lonigan, C.J., Valdez-Menchaca, M.C., & Caufield, M. (1989, April). Intervening in early shared book reading: Scripts versus direct Iraining. Paper presented at the biennial meeting of the Society for Research in Child Development, Kansas City.

Picture-Book interactions

APPENDIX

Book Membership for the Dimension of Text (Present vs. Absent)

337

Book Title (and Author) Text

Fami/y (Oxenbury, 1981) Not included

Guess who? (Corbett, 1982) Absent

h%ppy E;ff~~~~ (Khalsa, 1984) Absent

My Fa~i~ (Ladouceur, 1982) Present

The Merry-Mouse (Hillman, 1983) Present

Play (Curry, 1983) Present

NOW. The book Family did not include sentences but did include four words.