Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Grade one marks a period when children acquire the skills necessary to successfully read novel words based on letter-sound correspondences and blending letter sounds or phonemes together (Ehri, 1991). As children enter this developmental phase and begin to take over the role of the reader themselves, parents’ goals for shared book reading shift, such that fostering reading via this activity becomes as important for parents as enjoying books and fostering the parent-child relationship (Audet, Evans, Williamson, & Reynolds, 2008). Thus, parents incorporate a substantial amount of coaching when reading with their primary grade children (Evans, Barraball, & Eberle, 1998; Evans, Moretti, Shaw, & Fox, 2003; Mansell, Evans, & Hamilton-Hulak, 2005). In so doing they play an important role in extending the instruction that teachers provide in the classroom.

Part of instruction includes feedback , or responses to children’s reading errors , or what Goodman (1969) and followers referred to as miscues . For consistency, the term errors will be used here throughout. Studies have consistently found children’s word reading ability, and in some cases reading comprehension , to improve significantly more when feedback is provided to errors than when errors are ignored (Heubusch & Lloyd, 1998; Spaai, Ellermann, & Reitsma, 1991). Parents themselves appear to be implicitly aware of this as they rarely ignore children’s errors. Rates as low as four percent have been observed during home observations of shared book reading in beginning readers (Evans et al., 1998). This is lower than that observed in studies of teachers (Allington, 1980; Chinn, Waggoner, & Anderson, 1993; Hoffman et al., 1984), possibly due to the more extensive attention parents can give one-on-one within the home.

A limited body of research has investigated the corrective approaches naturally used. Parents most often supply the correct word (referred to as terminal feedback) (Evans et al., 1998; Mansell et al., 2005; Stoltz & Fischel, 2003), as do teachers (Allington, 1980; Hoffman & Clements, 1984). However, Mansell and colleagues’ (2005) longitudinal study of shared book reading highlighted that there are two distinct groups of parents. “Word suppliers ” provided terminal feedback most frequently across kindergarten, grade one, and grade two. “Code coaxers ”, on the other hand, supplied the word less often and relied instead on various clues (called sustaining feedback) to guide their children to correctly read a word. These clues were most often graphophonemic in nature – that is, they focussed on helping the child decode the graphemes or written marks on the page, such as by segmenting the word into smaller parts for the child to attempt, providing sounds for individual letters or letter clusters to help the child over the difficulty, and encouraging the child to use their knowledge to sound out the word. However code-coaxers also used some context clues such as pointing to a corresponding picture.

What child characteristics determine the nature of parent feedback during shared book reading remains largely unexplored. As outlined in the next section, research conducted thus far shows that children’s reading skill likely has an effect. The focus of the present research was to expand upon previous work to determine the extent to which child shyness and reading ability predict both parent and child behaviours in response to words the child cannot read. Consideration was also given to potential differences between boys and girls.

Child Characteristics Influencing Reading Behaviour

Child Reading Skill in Relation to Parent and Child Behaviour

Studies have shown that the amount and type of feedback provided by parents and teachers varies according to the child’s facility with word reading. For example, in their observational study of shared book reading with grade one children, Evans and colleagues (1998) found that parents more frequently provided graphophonemic cues to stronger readers and were more likely to encourage the child to try a word again. In contrast, pictorial cues were most commonly provided to children with poorer word reading scores. Similarly, Stoltz and Fischel (2003), and Mansell et al. (2005) found that picture clues were more frequently given to children with lower scores on a graded word recognition test. Straightforward pictorial cues provide less able beginning readers with a swift way to identify a word, while slightly stronger readers have the phonological and letter-sound skills to benefit from graphophonemic clues and to subsequently use this information to sound out future words (Evans et al., 1998).When children are a little older, the errors of relatively poorer readers are more often followed by graphophonemic coaching from teachers (Allington, 1980), while the less frequent errors of the more skilled children, which often do change the meaning of the text, are ignored or corrected via terminal feedback (Hoffman & Clements, 1984). Moreover, Evans et al. (2003) found that when a particular type of feedback was not successful, parents switched to a different, more transparent, clue to bring about success.

Research has also shown that children can approach words in four ways – recognise the word from their sight vocabulary, decode the word, use a familiar word that is orthographically similar as an analogy, or guess at the word based on context cues (Ehri, 1991), and that children’s approaches change as they become more highly skilled. Biemiller (1970) found that very early readers relied almost exclusively on context to guess at unknown words, as most of these children’s errors were semantically appropriate substitutions but did not visually resemble the actual printed word. In a second developmental phase, children frequently made no attempts and, when they did provide a guess, their guesses tended to have some similarity with the printed word such as the same first letter but the word might not fit the sentence. Finally in the third phase, these graphophonemic substitution errors continued but the reading errors also preserved the meaning of the sentence, suggesting that children were able to consolidate their developing phonetic knowledge with their more developed contextual understanding. Biemiller’s developmental sequence is consistent with the trajectories later put forth by Ehri (1991) and Whitehurst and Lonigan (1998) and shown most recently in a study by McGee, Kim, Nelson, and Fired (2015).

Child Shyness in Relation to Parent and Child Behaviour

Shyness is defined as a “tendency to react with tension and discomfort to strangers and social-evaluative situations” (Asendorpf & Meier, 1993, p.1072). For many, a biological basis underlies this behavior in that shy individuals appear to have a lower threshold for arousal (Marshall & Stevenson-Hinde, 2001). A recent literature review by Evans (2010) showed a modest but consistent negative relation between children’s shyness and language development across studies conducted in English speaking countries, as well as in French speaking Canada, Hong Kong, Germany, Norway, Saudi Arabia, Sweden, and Switzerland, with shyer children having lower scores in both the expressive and receptive language domains (Crozier & Badawood, 2009; Engfer, 1993; Kristensen & Oerbeck, 2006; Normandeau & Guay, 1998; Steinhausen & Juzi, 1996; Ting, 2008). While the research is sparser, modest negative associations with academic achievement assessed via reading and mathematics tests in the primary grades were also documented in this review.

The mechanism by which such negative associations occur is not well understood. Spere, Schmidt, Theall-Honey, and Martin-Chang (2004) suggested that shyness does not result in delayed language development, but rather that less shy children get more practice at language, accelerating their language learning. For example, shy children volunteer less speech, speak fewer utterances, and display longer latencies before their first utterance in social situations (see Evans, 2010), and speak fewer utterances and volunteer less content while reading a wordless storybook with their mothers (Reynolds & Evans, 2009).

This decreased verbal participation may stem from shy individuals’ fear of negative evaluation, unwillingness to take risks, and tendency to cope with anxiety through avoidance (Barrett, Rapee, Dadds, & Ryan, 1996; Hope, Rapee, Heimberg, & Dombeck, 1990; Levin & Hart, 2003). These same characteristics may influence children’s willingness to guess at difficult or unfamiliar words during shared book reading.

Observational findings also suggest that shyer or more anxious children elicit more controlling and overprotective behaviours from parents (Greco & Morris, 2002; Hudson & Rapee, 2002; Siqueland, Kendall, & Steinberg, 1996; Wood, McLeod, Sigman, Hwang, & Chu, 2003) and teachers (Evans & Bienert, 1992). These differences may reflect a general desire to protect shy children from experiences of failure (Wood et al., 2003). Rubin and Burgess (2002) proposed that adults naturally respond to expressions of social fearfulness in a child with sympathy and are quick to solve the child’s dilemma in response to this discomfort. This, however, can be overly protective and thwart the child’s development of problem solving and emotional regulation.

No studies have been located that examine the effect of child shyness on interaction when children do part of the reading when parents and children share storybooks together. As a suggestion of what may occur in this context, one may turn to studies of high- and low-control utterances used by adults with shy and non-shy children (see Edison et al., 2011; Evans, 1987; Evans & Bienert, 1992; Reynolds & Evans, 2009; Wood & Wood, 1983). High-control utterances include chains of questions that seek a specific response (such as what, why, where and yes/no questions), while low-control strategies involve the adult making personal contributions and remarks that seek to engage the child in conversation rather than to direct it. Studies have shown that high-control approaches tend to be used more often when interacting with shy children (Evans & Bienert, 1992; Wood & Wood, 1983) and selectively mute children (Edison et al., 2011). Moreover, it is thought that these tendencies in adults may exacerbate the already avoidant nature of shy and socially anxious children (Barrett et al., 1996; Coplan, Arbeau, & Armer, 2008: Dadds, Barrett, Rapee, & Ryan, 1996; Hudson & Rapee, 2001; Rapee, 2001) by reinforcing it and entrenching lower self-efficacy or ‘can do’.

The Current Study

The current naturalistic study sought to extend the extant literature on children’s reading behaviour and adult feedback to their errors through investigating the extent to which reading ability and shyness, as well as the two in conjunction with each other, predict the way in which children react to unfamiliar words, and parents respond to their difficulty. Shared reading sessions in each child’s home were audiotaped and non-verbal behaviours were recorded by an observer. Audiotapes were transcribed and coded for parent and child behaviours. Parental utterances in anticipation of a child being unable to read a word correctly or immediately following a reading error were coded for type of response. Children’s utterances also were coded for the type of attempt made upon encountering a word they could not read.

In the study shyness and reading skill were treated as continuous variables, rather than dividing the sample into extreme groups of high and low shyness and reading skill. Given previous reports of parental sensitivity to children’s ability as they read, it was expected that parents increasingly would provide graphophonemic clues to foster sounding out words and non-specific encouragements to prompt another attempt at a word to children of increasingly higher decoding skill . Conversely, they decreasingly would provide clues that directed the child to make a guess on the basis of the illustrations or context as reading skill increased. With respect to child shyness, it was expected that as shyness increased, parents would increasingly adopt a more protective parenting style by increasingly supplying the word after an error and interrupting their child to give the next word in anticipation of the child having difficulty with it. Conversely, they would decreasingly provide non-specific encouragement to try a word again as child shyness increased.

For children’s behaviours, weaker reading skill was expected to be associated with more incomplete attempts at words and errors and more requests for assistance. In contrast, higher reading skill would predict more errors in the form of substituting the printed word with another. Finally, as suggested by previous research on risk-taking (Barrett et al., 1996; Hope et al., 1990; Levin & Hart, 2003), children with increasingly higher shyness were predicted to more frequently hesitate or request assistance and less often to attempt unfamiliar words. Hence they would make fewer word substitutions and incomplete attempts at decoding.

Methods

Participants

Participants were 6-year-old children enrolled in elementary schools in three small cities and surrounding rural areas in south-western Ontario, Canada. They were recruited in junior kindergarten as part of a larger longitudinal study on emergent literacy. All required data for the current study were available for 94 children consisting of 44 girls and 50 boys. The parent who took part in the home reading session was the one who viewed himself or herself as most often reading with the child. There were 87 mothers, 6 fathers, and one mother-father team. The average age of children at the time of the grade one home visit was 6 years and 6 months. All families reported reading predominantly English storybooks and seven families reported that a second language was spoken at home. Maternal education ranged from the completion of high school to graduate-level education, with 76 % of mothers having a college diploma or undergraduate university degree. Father’s education ranged from the completion of grade ten (three fathers) to the completion of a graduate-level degree. For fathers, 72 % reported having a college diploma or undergraduate university degree. Finally, average family income, measured using eight rating categories (less than $16,000, $16,000 – $26,000, $26,000 + − $40,000. $40,000 +− $55,000, $55,000 + − $70,000, $70,000 + − $85,000, $85,000 + − $100,000, $100,000+) was 6.37, indicating that, on average, the families earned between $70,000 and $85,000 annually. For comparison purposes, the median family income in Canada in 2011 was $76,000 (Statistics, 2013).

Materials

Storybooks

The following nine illustrated narrative storybooks were brought to the home and provided for parent and child to choose from. Listed in order of increasing difficulty, these were: Cat Traps (Coxe, 1996), I Can’t Sleep (Graves, 1994), Dear Zoo (Campbell, 1982), In My Backyard (De Vries & Zimmermann, 1992), Five Silly Fishermen (Edwards, 1989), A Kiss for Little Bear (Minarik, 1968), Moon Boy (Brenner, 1990), Tickling Tigers (Currey, 1996), and Grandma and the Pirates (Gilman, 1990). These books varied from simple repetitive texts appropriate for very early readers through to longer and more complex stories appropriate for children with higher reading skill. Only the occasional parent commented that they had any of these books in their home.

Colorado Childhood Temperament Inventory

(CCTI, Bus & Plomin, 1984; Rowe & Plomin, 1977). Parents completed the CCTI during their child’s grade 1 year. The CCTI contains five items tapping shyness, all scored on a scale of 1 (not at all) to 5 (a lot). The items on the shyness subscale were: (1) My child takes a long time to warm up to strangers; (2) My child tends to be shy; (3) My child makes friends easily (reverse scored); (4) My child is very sociable (reverse scored); and (5) My child is very friendly with strangers (reverse scored). In the current sample, reliability as indexed via internal consistency was .84, consistent with the .83 reported by Bus and Plomin (1984) and Rowe and Plomin (1977).

Woodcock Reading Mastery Test – Revised, Word Attack Subtest

(WRMT; Woodcock, 1998). For our index of reading skill, children completed the Word Attack subtest of the WRMT in November or December of their grade 1 year. This task requires children to read isolated non-words (e.g., pruff), called pseudowords. This measure was chosen as it provides an index of children’s ability to read or decode words that they do not recognize on sight and as such reflects their ability to tackle and accurately read previously unseen or new words such as might be encountered during shared book reading. The internal consistency of the Word Attack subtest has been reported to be .87 (Woodcock, 1998).

Procedure

Schools were selected by the overseeing school boards to span a range a socioeconomic neighbourhoods. School visits to assess the children took place in November or December (the third and fourth month of the children’s 10-month grade 1 year). In January, February, or March of the same grade year, an observer visited the family’s home to observe parent and child reading together. Home observations took place when and where the family most frequently engaged in shared book reading, typically on a couch in the family room or at the kitchen table. Visits lasted approximately 1 h. The books the observers brought were spread out in order of increasing difficulty, and parent and child were asked to choose books that the child could read with some assistance, and to read together as they normally would. The observer allowed the dyad to proceed without any further guidance or interruptions. Sessions were audiotaped and observers recorded nonverbal behaviours as the dyads read, such as the parent pointing to illustrations or covering up parts of words. Audiotapes were later transcribed verbatim using the CLAN format of the Child Language Data Exchange System (MacWhinney, 2000), with the nonverbal notations added to facilitate coding parent and child utterances as outlined below.

Coding System

Coding of Parental Feedback

All feedback first provided by parents following a child’s error or request for assistance was coded. In addition parents sometimes anticipated that a child would have difficulty and intervened before the child attempted a word or requested the parent’s assistance by asking for help or signalling that they found the word difficult through a long pause, “um” or nonverbal cue. Both responses to miscues and anticipatory helps were coded according to the type of assistance the clue or prompt provided to the child, using the categories previously employed by Evans and colleagues (1998) and Mansell and colleagues (2005). The coding categories were as follows:

  1. 1.

    Ignore: The parent did not provide feedback. Note that this coding category does not apply to the coding of anticipatory help.

  2. 2.

    Encouragement: The parent encouraged the child to try the word without specific guidance (e.g., “Take your time”, “Look at it again”).

  3. 3.

    Graphophonemic clue: This included references to (a) letter details such as particular letters or letter combinations within the word (e.g., “That’s a ‘B’ not a ‘D’”, “(b) phonetic clues such as letter sounds, syllables, or word parts (e.g., “It starts with ‘chuh’), or (c) encouragement to the child to use letter-sound knowledge to decode the word (e.g.,. Don’t forget what ‘C’ and ‘H’ say when they’re together”; “SH” goes shhhh).

  4. 4.

    Context clue: This included (a) pointing to illustration wherein the word was depicted, (b) referring to the text’s meaning or to what word might fit best given the story’s plot, (c) calling up the child’s general knowledge wherein information from outside of the storybook directed the child to the unknown word (e.g., “It’s what you ride to school in”), or (d) retrievals of previous instances of the word (e.g. “It’s the same as this word that you just read”).

  5. 5.

    Terminal feedback: The parent supplied the unknown word for the child.

Parental responses to errors in a given category were tallied and added to the same category for anticipations of errors to arrive at a single score for each feedback category. The total number of anticipatory helps and total number of word supplies was also tallied.

Coding of Children’s Approaches to Unknown Words

The children’s initial approaches to tackling words they could not read were coded into one of the following categories:

  1. 1.

    No attempt: The child (a) paused, stopped reading, or remained silent, or (b) requested parental assistance.

  2. 2.

    Skips word: The child continued reading without attempting the word.

  3. 3.

    Incomplete attempt: The child provided an incomplete or non-word.

  4. 4.

    Substitutes a real but incorrect word: The child provided a real word that was not the word printed in the text.

Reliability of the Coding System

To assess the reliability of the coding scheme, a second investigator coded 20 % of the transcripts. Cohen’s Kappa was calculated for each parent and child code, as was the percentage of events coded by one coder that were also coded by the second. Cohen’s Kappa for coding of parent behaviour was .93, with 86 % agreement on events to be coded. Reliability coding of children’s approaches to difficult words yielded a Cohen’s Kappa of .85, with an 88 % agreement on incidents to be coded. Cohen’s Kappa was also calculated for each of the individual codes, and obtained values ranged from .79 to .99. All obtained Cohen’s Kappa values were considered strong according to the guidelines of Altman (1991) and Landis and Koch (1977).

Data Preparation

Table 7.1 presents means, standard deviations, and ranges for the data gathered in the study. Child shyness was indexed using the combined parent ratings for the five shyness items of the CCTI. Scores closer to the maximum score of 25 represented higher shyness in a child. Our index of being able to read words was performance on the Word Attack subtest according to norm-referenced scores which have a mean of 100 and a standard deviation of 15. Parental anticipations were a simple count of the frequency with which parents anticipated difficulty reading a given word and intervened. All other scores in Table 7.1 are percentages in order to control for the fact that some children made more errors than other children, due to such factors as reading more text in the books, or reading text that was more difficult for them. Specifically, parent scores in the first half of the table (except for anticipations) represent the percentage of words on which children erred to which parents provided feedback of a given type. Similarly, child scores in the bottom line of Table 7.1 represent the percentage of words on which children erred that fell in the no attempt, incomplete attempt, substitution, or skip the word categories. Skipping the word was infrequent (M = 1.53, SD = 2.09) and thus this variable was dropped from further analyses.

Table 7.1 Descriptive statistics for parent and child variables
Full size table

In order to determine whether and to what extent child shyness and child ability to read words predicts their behaviour and that of their parents, shyness and reading ability were entered as predictors in a series of multiple regressions. In addition an interaction term was included to determine whether the effect of one’s shyness or reading ability depended on the value of the other (e.g., whether shyness would have an effect only if reading skill was low).

Results

Preliminary Analyses

On average, children made 44.91 errors and/or requests for assistance per book reading session (SD = 25.45). As shown in Table 7.1, upon encountering a word they could not read, children substituted another word a little over half the time. Incomplete attempts to sound out the word comprised a quarter of their behaviour, followed by pausing/requesting parental assistance at 18 %.

Visible anticipations of child difficulty occurred roughly just once per book reading session (M = .98, SD = 1.71). Therefore, parent scores almost entirely reflect responses to children’s miscues. Parents ignored 13 % of children’s errors. About a third of their feedback consisted of supplying the word. Graphophonemic clues were only slightly less common. Encouragement to try the word again with no additional guidance constituted a quarter of parental feedback. Context clues were by far the least common strategy at just 5 %.

Sex differences were not anticipated but one-way analyses of variance (ANOVAs) were conducted to determine if scores for reading ability, shyness, or any of the behaviours varied by children’s sex, in case this needed to be considered in the regressions. Parent behaviours did not differ according to whether the child was a boy or a girl. Child reading skill on the Word Attack subtest also did not differ by sex. However, boys (M = 59.52 SD = 13.22) were more likely than girls (M = 49.57, SD = 14.03) to guess at a word, F(1, 92) = 12.49 p < .05. Girls (M = 23.10, SD = 19.37) were more likely than boys (M = 13.87, SD = 15.93) to pause or request assistance F(1, 92) = 6.47, p < .05 Thus, children’s sex was included as a predictor in the regression analyses for these two variables.

Correlations were calculated to determine which parent and child behaviours were significantly related to child shyness and reading scores (see Tables 7.2 and 7.3). Shyness and Word Attack were uncorrelated but significant correlations were observed for each with child and parent behaviours. Specifically, shyer children were more likely to pause or ask for assistance and were less likely to make a guess at the word. Poorer readers also were more likely to pause, and they were less likely to make an incomplete attempt at the word. The different forms of sustaining feedback were not correlated with each other, but graphophonemic clues and encouragements to try again were negatively correlated with the parent giving the word. Shy children were more likely to be given the word or a context clue to it, and less likely to be encouraged to try the word again. Finally, parents more often interrupted poorer readers in anticipation of a difficult word, and more often gave the child clues to read the word and less often gave general encouragements to try the word again when the child made an error.

Table 7.2 Correlations of parent behaviours with child shyness and decoding ability
Full size table
Table 7.3 Correlations of child behaviour with child shyness and decoding ability
Full size table

Prediction of Behaviour in Shared Book Reading

To predict parent and child behaviours, child shyness, reading ability, and where relevant sex were entered in the first step, and the two-way (reading x shyness) and where relevant three-way (reading x shyness x sex) interactions of them entered in subsequent steps of the multiple regressions. None of the two- nor three-way interactions was significant. Thus, the simple main effects are reported below and displayed in Table 7.4.

Table 7.4 Statistics for hierarchical regressions of child decoding and shyness on behaviour
Full size table

Parent Behaviour

A significant amount of variance in how much parents gave graphophonemic clues was predicted by child reading ability, F(2,91) = 4.54, p < 05. Specifically, children’s Word Attack scores negatively predicted graphophonemic feedback, in that parents offered this type of feedback less as children displayed stronger reading skill. The squared part correlation (sr 2 column in Table 7.4) indicated that reading ability accounted for 9.00 % of variance in parent use of graphophonemic feedback. Child shyness had no predictive value.

A significant amount of variance in the percentage of context clues was also predicted, F(2,91) = 9.05, p <. 001. Context clues were negatively predicted by child reading ability, which accounted for 11.42 % of the variance, but were positively predicted by shyness which independently accounted for 4.49 % of variance. Thus, parents were less likely to rely on context cues as child reading ability increased, but were more likely to do so if their child was shyer than others in the sample.

Concomitantly, parents also less frequently encouraged shyer children to try the word again without offering guidance, but were more likely to do so with better readers. A total of 11.90 % of variance was accounted for by these factors, F(2,91) = 6.14, p < 01. Reading ability accounted for 4.16 % of the variance while shyness accounted for 6.97 %. The extent to which parents simply supplied the word was also predicted, F(2,91) = 3.61, p < .05. A greater percentage of responses to miscues entailed supplying the word if the child was shyer, with shyness accounting for 7.07 % of the variance. Decoding ability made no significant contribution.

For the percentage of errors ignored by parents, decoding ability but not shyness predicted 6.30 % of the variance, a small but statistically significant amount, F(2,91) = 3.21, p < .05. Parents ignored a higher percentage of their children’s errors when children were stronger readers. Finally, anticipations of difficulty reading words also were significantly predicted by reading skill, F(2,91) = 6.06, p < .01. This parent behaviour was more common when children were poorer readers, with decoding ability predicting 11.49 % of variance. Child shyness held no predictive value here.

Child Behaviour

The percentage of difficult words for which children paused or requested assistance from their parent was significantly predicted, F(3, 90) = 11.15, p < 001. Stronger readers made attempts more frequently, with their reading skill accounting for 10.76 % of the variance. Shyness also predicted the extent to which children attempted words, accounting for almost as much variance – 8.64 % – , with shyer children less frequently doing so. Finally, sex accounted for another 5.11 % of the total variance. As noted earlier, girls engaged in proportionally more of these behaviours than boys.

The percentage of miscues for which children substituted a word for what was on the page was significantly predicted, F(3,90) = 7.57, p < 001. Decoding skill accounted for 4.29 % of the variance, but sex accounted for double that at 10.50 %. As noted earlier, boys more frequently substituted a word than girls, as did children with stronger decoding skill. Shyness had no predictive value.

Finally, a modest but significant amount of variance (4.80 %) in incomplete attempts at reading the word was predicted by reading skill, F(2,91) = 3.69, p < .05. As reading skill increased across children, more incomplete attempts were made.

Table 7.5 Predictive strength of decoding ability, shyness, and sex for children’s approaches to difficult words
Full size table

Discussion

This study aimed to determine the extent to which children’s reading skill and shyness predicted the behaviours of parents and children when difficult words were encountered during shared book reading. Such difficulties are common as parents encourage their children to read to them instead of reading to the child. To accomplish this, counts were made of the types of assistance first offered by parents in response to an inability to correctly read words in the books read, as well as any parent comment in anticipation of a word being difficult. The frequency of different child behaviours (i.e., no attempt, skip word, incomplete attempt, substitute a word) was also calculated. As expected, both reading skill and shyness were significant predictors of child and adult behaviour. Generally the former was the stronger of the two predictors. However shyness equally predicted the extent to which children hesitated or declined to attempt difficult words, and it was an even stronger predictor of the extent to which parents simply gave the miscued word rather than encouraging their child to attempt it. In addition, although not anticipated, children’s sex predicted two of their behaviours; girls more frequently paused or requested assistance with difficult words, while boys more frequently substituted a complete word.

Child Approaches to Difficult Words

Shy children have been found to take fewer risks (Levin & Hart, 2003) and to report greater fear of negative appraisals resulting from failure (Hope et al., 1990; Keaten, Kelly, & Finch, 2000). Commensurate with this, shyer children in the present study more frequently paused or requested assistance when encountering words they could not read. As expected, poorer readers also more often paused or requested assistance, instead of substituting a wrong word or making a partial attempt. Unexpected was that a child’s sex proved to be a third predictor, in that girls more frequently paused or requested assistance, while boys more often substituted a word for the one printed. The different responses of boys and girls may also reflect differences in children’s tendencies towards risk taking . A meta-analysis by Byrnes, Miller, and Schafer (1999) found males to be riskier than females across almost all risk contexts and age levels, with sex differences in the willingness to take intellectual risks larger than for many other risk contexts. Guessing at an unknown word would fall into the category of intellectual risk taking, affecting the behaviour of both shy children and girls in this study. Such reluctance may be even greater in oral reading with teachers and in the classrooms, in that shy children are likely to feel less comfortable and secure there than in the home. Small reading groups, having shy children read to younger children, and ‘lightening up’ reading to the teacher by having stuffed toys or puppets as part of the reading dyad may make shy children more comfortable.

Parental Approaches to Difficult Words

Parents generally did not ignore their child’s reading errors. This is consistent with previous research showing that parents are vigilant in providing feedback (Evans et al., 1998; Mansell et al., 2005). In general, parents more frequently responded to a child’s anticipated difficulty, actual errors, or requests for assistance by supplying the word (roughly 31 %) than with other forms of assistance. Graphophonemic clues to help the child sound out the word were almost as common (roughly 27 %). In addition the two types of feedback were negatively correlated. This too is consistent with previous research (Evans et al., 1998; Mansell et al., 2005; Stoltz & Fischel, 2003) and lends support to the distinctions between “word supplier” and “code coaxer” categories of parent reading assistance put forth by Mansell and colleagues (2005). These categorizations are further supported by the finding that supplying the word was also negatively correlated with the other two forms of sustaining feedback – context feedback (largely directions to look at a corresponding illustration) and try again. Overall then, parents who frequently read the difficult word for the child less frequently provided clues to show how difficult those words might be tackled, and less often provided children with the opportunity to employ their own self-initiated strategies. Consistent with previous studies, general encouragements to try again and clues to assist a child before the attempted reading a word decreased as children’s reading skill increased across children.

The potential differential effectiveness of different types of feedback in shared book reading has yet to be established. Previous research (e.g., Lovett, Barron, Forbes, Cukst, & Steinbach, 1994; Meyer, 1982; Perkins, 1988; Spaai et al., 1991; van Daal & Reitsma, 1990) is sparse and inconsistent as to the value of supplying the word (as a whole or phoneme by phoneme) rather than encouraging another attempt. Similar inconsistency is seen in studies of meaning versus graphophonemic feedback (see Crowe, 2003). Given the accumulated evidence on the importance in learning to read of understanding that letters represent the phonemes of spoken language (the alphabetic principle) (e.g., Byrne, 1998; Reiben & Perfetti, 2010), the inconsistency may be a function of the different populations and levels of skill sampled in the studies cited. It is the case, however, that any form of constructive feedback appears to result in better reading gains than none at all. As such, the coaching parents provide during shared book reading to help children over difficult words, rather than ignoring their errors, likely provides a valuable supplement to the instruction and experiences received at school.

While shyness did not predict parental anticipations, parents were less likely to encourage shyer children to try a word again, and more likely to tell them the word or point to a picture of it. This result echoes previous findings on the conversational interactions of adults with shy, anxious, and selectively mute children (Edison et al., 2011; Evans & Bienert, 1992; Greco & Morris, 2002; Hudson & Rapee, 2002; Moore, Whaley, & Sigman, 2004; Wood et al., 2003; Wood & Wood, 1983). Reading the word and alerting child to a picture of the word to name quickly solves the reading dilemma. However doing so may also reduce children’s sense of responsibility and agency for reading. Gene-environment interaction theory holds that shy children elicit protective and controlling behaviours from others that can then act to maintain or exacerbate their behavioural tendencies (Rapee & Spence, 2004; Rubin & Burgess, 2002; Wood et al., 2003). The observed tendency of parents to provide more direct feedback by giving the word, as opposed to giving strategies and clues to encourage beginning readers to think through the word, may constrain children’s feelings of confidence in attempting to read on their own. It may also decrease children’s accumulated experience in coming to a close- enough approximation of a word to recognize it and benefit from the self-teaching mechanism proposed by Share (1995). This could potentially hinder the development of their decoding skill over time, and account for the modest negative relation of shyness to reading skill that has emerged in the majority of studies of shyness and academic ability in children (see review by Evans, 2010).

Limitations and Directions for Future Research

Replication of these results would alleviate concerns that the findings presented here are specific to this sample of children. In addition, it is important to note that the present study was not an attempt to predict as much variance as possible. Indeed, no single regression analysis accounted for more than 27 % of the variance in any parental behaviour. Other variables that likely also influence the feedback strategies employed include mothers’ education and proficiency in reading (Neuman, 1996; Tracey & Young, 2002), the nature of the mother-child attachment relationship (Bus, Belsky, van IJzendoorn, & Crnic, 1997), parents’ own memories about how they themselves were taught to read as children (Evans et al., 1998), the goals that they have for reading with their children (Audet et al., 2008; Evans & Audet, D. (July, 2014), and the nature of the book read – alphabet or storybook (Davis, Evans, & Reynolds, 2010; Smolkin, Yaden, Brown, & Hoffman, 1992). Including all these variables would entail doubling the sample size and require even more intensive labour in transcription and coding.

The present study consisted of observations of shared book reading at a single point in time when the children were in grade one. Longitudinal observations beginning in preschool and extending through grade two would help to determine when child characteristics of shyness and sex begin to exert an effect on behaviour during shared book reading and how long it persists. Unlike Evans et al. (1998) who found that parents more frequently provided graphophonemic feedback as reading skill increased, the present study found the opposite. This is likely because the children in the present study were slightly older and had more advanced decoding skill than those in Evans et al. study. Given the substantial changes in letter-sound knowledge , phonemic awareness , and word reading skill from kindergarten through grade two (Speece, Ritchey, Cooper, Roth, & Schatschneider, 2004; Wagner, Torgesen, & Rashotte, 1994, different findings might be expected at earlier versus later time points. Longitudinal research with multiple time points for data collection would help to clarify this issue as well.

Future research should also further investigate the effects of shyness and reading skilling by observing shared book reading in teacher-child and child-child dyads. Top and Osguthorpe (1987) showed benefits for tutors, when the tutors have relatively poor reading skills for their age are paired with younger children having still weaker skill. Part of the gains these tutors make may be from encouraging and allowing them to try out their decoding skills with relatively little evaluative risk.

Conclusion

In the early stages of independent reading , children who were shyer and those who had lesser reading skill were more likely to pause or request assistance when reading to their parents. Shyness and decoding skill exerted differential effects on the type of feedback parents provided to their child’s reading errors. As children’s reading skill increased, parents provided feedback of a less supportive nature by encouraging children to try again, and as shyness increased, gave the most supportive feedback of supplying the word across children. These results add to a growing body of research on child X environment models of development (e.g., Cairns, Elder, & Costello, 1996; Magnusson & Hakan, 1998; Sameroff, 1993), suggesting that children produce different reactions in their environment that may affect their development.

As was earlier noted, one suggested reason for young shy children’s poorer language scores is that their behavioural inhibition restricts the amount of practice they have with language. The same may be true in the initial stages of their becoming independent readers. The present study shows that shy children more frequently ask for help and that the help received is more likely to consist of parents giving the word. Thus shyer children have less opportunity to try out their developing skills and receive coaching on how to read novel words. At school this may be exacerbated by teachers, who sensitive to children’s shyness, may desire to minimize their stress and embarrassment by less often having these children read out loud to them. It may be further exacerbated by the documented tendency of teachers to view shy children as less intelligent and less academically competent (Coplan, Gavinski-Molina, Lagace-Seguin, & Wichmann, 2001; Gordon & Thomas, 1967; McBryde, Ziviani, & Cuskelly, 2004; McCroskey & Daly, 1976). Knowing of these findings in combination with the present results will hopefully encourage parents and teachers to empathetically wait a little longer for shy children’s responses. This in itself will help them to better understand children’s skill level, to make it clear to the children that it is okay to make mistakes as it is part of trying out new knowledge, and to provide on-line coaching in word recognition during oral reading to encourage children’s reading development. Further suggestions for interacting with shy young children are provided in Coplan and Arbeau (2008), and Evans (2001, 2010).