Using strategic pauses during shared reading with preschoolers: Time for prediction is better than time for reflection when learning new words

Abstract

Preschoolers can learn vocabulary through shared book reading, especially when given the opportunity to predict and/or reflect on the novel words encountered in the story. Readers often pause and encourage children to guess or repeat novel words during shared reading, and prior research has suggested a positive correlation between how much readers dramatically pause and how well words are later retained. This experimental study of 60 3- to 5-year-olds compared the effects of placing pauses before target words to encourage predictions, placing pauses after target words to encourage reflection, or not pausing at all on children’s retention of novel monster names in a rhymed storybook. Children who heard dramatic pauses that invited prediction before the monsters were named identified more at test than children who heard either post-target pauses or the story read verbatim. In addition, there was an interaction between pre- vs. post-target pausing and whether the pauses were silent or replaced with an eliciting question, such that silent pauses were more effective before the target words, while eliciting questions were more effective after. Overall, dramatic silent pauses before new words in a story were found to best help children attend to and remember those new words.

Keywords

Dialogic reading shared book reading strategic pauses word learning

Reading with young children is a common practice at home and in the preschool classroom that provides rich support for language learning, and strengthening vocabulary (e.g., DeBaryshe, 1993; Elley, 1989; Farrant & Zubrick, 2013; Flack, Field, & Horst, 2018; Robbins & Ehri, 1994; Sénéchal & Cornell, 1993). While children can learn vocabulary passively from being merely exposed to words in context during reading (e.g., Elley, 1989; McLeod & McDade, 2011), highlighting new words during shared reading can make them especially memorable. Individual words can be highlighted during reading in several ways, including both their placement in the text and the way they are presented or even commented upon by the adult reader. For example, words at the end of a sentence receive additional stress (e.g., Clark, 2010) and therefore may draw more attention; and, if a text is in verse, words that complete a rhyming stanza are especially memorable (e.g., Read, 2014; Read, Macauley, & Furay, 2014; Read & Quirke, 2018). In addition, words that appear at the end of a page are more likely to be explained by parents reading with their young children than equally novel words earlier on the page (Evans, Reynolds, Shaw, & Pursoo, 2011). Thus, before the reader even picks up a book to share with a child, the placement of vocabulary within it can make it more or less memorable.

Of course, new words can also be highlighted by the reader independently of the text structure. Repetition may be the simplest way to highlight vocabulary, and young children typically remember more new words when an adult reader repeats the words either within a single story or across multiple readings of the same story (e.g., Horst, Parsons, & Bryan, 2011; McLeod & McDade, 2011; Sénéchal, 1997). Going a step further in an attempt to get the child to be a more active participant, the reader might highlight new words by eliciting the child’s repetition of them directly (e.g., Ewers & Brownson, 1999; Sénéchal, 1997). For example, in Ewers and Brownson’s (1999) investigation of a shared book activity, the adult would read aloud a sentence with a new word such as, ‘He is wearing his favorite fedora’ while pointing to the illustration, and then immediately elicit the child’s repetition of that word with a question, e.g., ‘What was he wearing?’ thus drawing attention to the word and reminding the child of the mapping they just learned. This kind of active participation on the part of the child resulted in better word retention. In many other learning scenarios, active participation appears to produce better learning and memory. In the study of children’s basic memory processes there are mixed results around the possibility of a ‘production effect,’ such that when children actively verbalize words they might remember (and potentially learn) them more readily. Icht and Mama (2015) found that when 5-year-old children looked at pictures and said their labels aloud this resulted in better memory for both familiar and new words than simply looking at the pictures and listening to them labeled. Alhough Zamuner, Strahm, Morin-Lessard, and Page (2018) actually found a reverse effect for children this age in a similar task, such that children remembered more novel words when they concentrated on just listening to them labeled rather than having to repeat them aloud as well. Zamuner et al. suggest that when task complexity is increased, the production effect may be reversed because of cognitive load. So while repetition of new vocabulary can be helpful for highlighting the words that we want children to attend to and learn, their own repetitions of those words aloud may be useful in some cases but not all.

Aside from repetition there are other direct ways the reader can highlight vocabulary during shared reading such as giving simple explanations or explicit definitions of new words (e.g., Beck & McKeown, 2007; Houston-Price, Howe, & Lintern, 2014), or commenting on or asking questions about new words as they are encountered in a story (e.g., Ard & Beverly, 2004; Ewers & Brownson, 1999; Sénéchal, 1997; Sénéchal & Cornell, 1993; Walsh & Blewitt, 2006). These direct teaching strategies can help build vocabulary, though Walsh and Blewitt (2006) found that while stopping and asking questions about a word was helpful for later retention, there was no difference between ‘eliciting’ questions such as ‘What is this?’ while the investigator pointed to a pagoda compared to ‘non-eliciting’ questions (e.g., ‘What color is the pagoda?’) that simply repeated the novel word but did not require the child to use it in their answer. Therefore, adding to the mixed findings on production effects, it does not appear that the child’s own verbal production of the words necessarily produces better retention. Further, a recent shared reading study with 3- to 6-year-olds by Jimenez and Saylor (2017) found that interjecting questions or additional information during the story rather than before or after the story can potentially overtax young listeners’ memory, meaning that especially for younger learners, or those with weaker memory and attention skills, too much questioning or explaining of vocabulary during a story might actually make learning the new words harder.

A less overt, potentially less demanding, way to highlight vocabulary during shared reading is through pausing either before or after a target word. While there is little empirical evidence that simply pausing after a word will make it more memorable, intuitively it seems that allowing a child time for reflection after hearing a new word should be advantageous for learning. In their observational study, Evans et al. (2011) suggest that parents define and highlight challenging words that are at the end of the page more often than other words because they could be taking advantage of the natural pause in the book that comes at a page turn, but these researchers did not actually test whether that natural pause was in itself helpful. Research demonstrates that repetitions of a new word after it has been introduced are common among young children (Clark, 2007) and may be a kind of spontaneous post-target reflection. And as described above, questions or discussions of new words after they are heard in a story are in some cases helpful (e.g., Ard & Beverly, 2004; Ewers & Browson, 1997). However, the question that remains unanswered is whether simply pausing silently after a word would also create a space for reflection that could make the words more memorable.

Pausing can also be a source of highlighting when it occurs before a target word. In particular, a long pause before a word often suggests to the child that they can supply the next word. For example, in a study by Read et al. (2014), parents might read the lines of a storybook, ‘My floppy ears might look quite funny / if I were a hopping …’ and pause dramatically before turning a storybook page and revealing ‘bunny!’ In that pre-target pause children produced guesses like bunny more than 30% of the time. While parents have been observed to do this kind of dramatic pausing during read-alouds for decades (e.g., Moerk, 1972) and early literacy promotion programs have recommended this as a useful dialogic reading technique to parents and educators (e.g., Trelease, 2013; Whitehurst & Pearson Learning Group, 2006; Yopp & Yopp, 2000), it has also not been empirically tested as a way of highlighting new vocabulary for the purpose of later retention. Prior research has, however, established that there is a positive relationship between the average duration of parent readers’ dramatic pausing before target vocabulary and how well children remember and identify that vocabulary after a story (e.g., Read, 2014; Read et al., 2014). For example, parents who paused before revealing the predictable word bunny in the above example from Read et al. (2014) had children who remembered the animals from the story with a higher success rate. However, even this finding was only correlational – we do not yet know that it was the pausing that increased memorability of target words, or whether another third factor was responsible for the association. For example, parents who know that their children have greater prior knowledge of a word or story might be more confident in using dramatic pausing than those whose children are less familiar with the vocabulary or less attentive in general, so from prior work we cannot be sure that it is the pausing in and of itself that benefits word recognition.

The current study examines the potential benefits to children’s vocabulary retention during storybook reading of pausing both before and after target words compared with reading without pauses. Pauses in either position may draw attention to target words, but pre-target pauses may give children an opportunity to predict the words, while post-target pauses may give children opportunities to reflect on those words. In an effort to understand the role of explicitly encouraging prediction and reflection, we also measure the effects of both pre- and post-target pauses that are silent compared with those that include an eliciting question. While pausing to allow completions and asking a child to repeat the information that they have just heard are both shared reading techniques advocated by many descriptions of good practices in dialogic reading with young children (e.g., Whitehurst & Pearson Learning Group, 2006; Yopp & Yopp, 2000), to our knowledge there has been no experimental research testing the potential benefits of these practices while controlling for other types of interactive reading. This study aims to garner a clearer understanding of how intentional pausing on the part of readers can benefit how preschoolers attend to and remember novel vocabulary from shared storybook reading.

Method

Participants

Sixty-three children participated in this study; three children failed to complete the full protocol due to inattentiveness, and therefore were excluded from the analysis. The remaining 60 participants (M_age = 47 months, age range: 32–65 months) were randomly assigned to three conditions: in the prediction condition there were 12 males and 8 females (M_age = 49 months); in the reflection condition there were 12 males and 8 females (M_age = 46 months); and in the control condition there were 8 males and 12 females (M_age = 47 months). All of the participants were from an on-campus preschool program for the children of faculty and staff of the university. Children were from ethnically diverse households, some of which included languages other than English spoken in the home, but all children were learning English as their primary language. All children experienced shared storybook reading in English at least several times per week.

Materials

This study used the same ‘monster rhyme’ researcher designed stories that have been used for previous experimental research and that children in this age range have demonstrated they understand and enjoy (e.g., Read, 2014). In the story, children are introduced to a set of eight new monster friends. The pages of each story were created as individual slides using PowerPoint© software. A novel monster was featured on each of the eight pages on a white background, with one rhyming stanza of text describing the monster. At the end of this four-line stanza was the monster’s name; the last word of each line rhymed with the name, and the name rhymed with a prominent feature of the monster (e.g., the ‘groze’ had a large nose) to ensure the novel word (the monster’s name) was maximally predictable and memorable (e.g., Read, 2014). The monster pictures were presented as bold-color, cartoon-style drawings of monsters described by children as ‘cute’ and designed by researchers to appear friendly and not scary. Each name was one syllable, began with one of three consonant clusters, and ended in a common rhyme. This naming pattern ensured the novel names rhymed with words common even to children as young as two and were not too difficult for young children to produce. The full text of all the rhymes and monster names used in each condition are featured in Appendix 1.

Eight identification test slides followed the storybook pages, each with a side-by-side pair of monsters from which participants were asked to choose the target after hearing a prompt (e.g., ‘Which one is the Groze?’). Here, the pairings of monsters were pseudo-randomized, monsters with the same initial consonant cluster were never paired together, the same pair was never repeated, each monster was used as both a foil and a target once, and the target monster was located equally often on either the left or right side of the slide. Another eight slides were created for the production test in which monster pictures were each presented alone with a white background, so that the child could be asked to produce a name for each. The order of the monsters in the production task was also randomized.

Procedure

Children heard one of three versions of the same story based on condition read twice by an experimenter. In the prediction condition, the experimenter either took a 3-second-long silent pause, or an eliciting pause asking ‘What’s he called?’ before reading aloud the monster’s name on each page. The monsters were always presented in the same order in all versions, but version 1A used silent pauses before monsters 1, 4, 6, and 7 and eliciting pauses before monsters 2, 3, 5 and 8, while version 1B used the reverse assignment of pause types before each monster. In the reflection condition, the same story was read with the pauses (either 3-second silences, or eliciting ‘What’s he called?’ questions) occurring after the monster names were read aloud, and also with two sub-versions 2A and 2B, which assigned the pause types alternatively to each monster. In the control condition the story was read without any pauses.

The study was designed to feel like a story and game for preschool-aged participants, and took no longer than 25 minutes per participant (including time to play and ask questions). Both parental consent for the child to participate and the child’s own verbal assent were obtained before we began the session. In the study, children sat in either a quiet, child-friendly playroom with their caregiver present, or in the staff room at their preschool with an assistant teacher present. The story and test slides were presented to children page-by-page as a full-screen PowerPoint© presentation on a laptop. The story was read from a laptop by a female researcher who was trained to time her pauses and elicit questions accurately and consistently, while still read in a natural, child-friendly manner. Whether the researcher paused before or after the target words or not at all, was dependent on the condition and sub-version of the story to which each child was randomly assigned. Researchers did not engage in any other side commentary or questions during the reading, and if a child asked questions or made comments, researchers’ responses were friendly but minimal (e.g., ‘mhm, let’s see what’s next’). The researcher asked the first set of identification questions after the first read-through of the story, then repeated the story and asked the identification questions again followed by the production questions. With written parental consent, a video recording was taken of the reading session using the laptop’s built-in camera so children’s spontaneous guesses, repetitions, and comments as well as their responses to retention tasks could be coded after the session in an effort to minimize distraction for the child.

Measures

Novel word retention

Storybook retention was measured through children’s accuracy in novel monster identification and production. To measure identification, children were shown eight slides with pictures of two monster pairs and asked to point to a target monster, (e.g., ‘Which one is the Smooze?’). Each monster was targeted once during the eight trials after the first storybook reading and again after the second reading. We derived a proportion correct score out of eight trials following each storybook reading for each participant. To measure production, after only the second reading of the story, children were shown eight slides, each with a picture of a single monster, and asked ‘What’s this monster called?’ Children received a point for each monster correctly named, and a half-point if they gave a monster name that was a close approximation (e.g., ‘schmooze’ instead of ‘smooze’) but no points if they offered a known word that was a feature present in the illustration (e.g., ‘shoes’ for ‘smooze’). This yielded a possible production score of 0–8 for each child. Two separate trained coders independently scored all the videos, and inter-rater reliability for these two measures was above 90%. In any cases of inter-rater disagreement, the primary investigator resolved the points given to a child by close observation of their video response.

Guesses and repetitions

Spontaneous or prompted guesses each participant made aloud before hearing the named monster during both storybook readings were recorded and counted, regardless of whether the guesses were correct. The number of spontaneous or prompted repetitions of monster names children made after hearing a monster named were also recorded and counted. This yielded two scores for each child in all three conditions – total number of guesses and total number of repetitions. Two separate trained coders independently scored all the videos, and inter-rater reliability for both these measures was 100%.

Results

Age effects

The ages of children were balanced across the between-subjects conditions; there were no significant differences between the ages of children in the prediction (M = 49 months, SD = 10), reflection (M = 46 months, SD = 7), and control (M = 47 months, SD = 13) conditions, F(2, 59) = .32, p = .727. However, because age is often an important predictor of language skill in preschool children, it was included as a covariate in all of the following analyses.

Novel word retention

Identification

In order to compare children’s correct novel monster identifications across the three conditions, both the first and the second time they heard the story and were tested, we conducted a 3 (condition) × 2 (time) analysis of variance with age of the children as a covariate (ANCOVA). This analysis indicated that there was both a significant main effect of condition, F(2, 56) = 4.55, p = .015, with a large effect size of η_p² = .14 and a significant main effect of the covariate age, F(1, 56) = 4.80, p = .033, η_p² = .08. However, there was no main effect of time, F(1, 56) = 2.83, p = .098, η_p² = .05; no interaction effect between time and condition, F(2, 56) = 0.75, p = .479, η_p² = .03; and no interaction effect of time and age, F(1, 56) = 3.09, p = .084, η_p² = .05.

Because of these findings, in further analyses we collapsed across time, using children’s total correct identifications out of all 16 times that they had been tested as a measure of their total proportion correct. Additionally, we continued to include age as a covariate in analyses of variance, but also directly tested the effects of age and any potential interaction effects of age and condition with a more direct regression analysis (see below).

When we analyzed children’s total correct identification across time, we again found a main effect of condition, F(2, 56) = 4.47, p = .016, η_p² = .14, and a main effect of the covariate age, F(1, 56) = 4.76, p = .033, η_p² = .08. Post-hoc analyses showed that the condition effect was driven by higher mean correct identification scores by children in the prediction condition (M = .70, SD = .17) compared to scores in the reflection condition (M = .57, SD = .18), p = .022, d = .73, and compared to children in the control condition (M = .54, SD = .17), p = .005, d = .94, though mean identification scores in the latter two conditions did not significantly differ from one another, p = .579, d = .18. In addition, when comparing children’s mean correct identification scores to a chance level of responding (.50, given the two-alternative test), it was only in the prediction condition that children identified the novel monsters significantly more often than chance, t(19) = 5.20, p < .001, d = 1.18. In the reflection condition children’s identification was marginally greater than chance, t(19) = 1.83, p = .083, two-tailed, d = .39, but in the control condition, children’s correct identification was not different from chance, t(19) = 1.14, p = .268, two-tailed, d = .24.

Because age significantly correlated with total correct identifications, r(58) = .29, p = .024, and was a significant covariate in the prior analyses, we also conducted a multiple linear regression analysis to predict children’s proportion of correct scores based on condition, age, and the interaction of the two. A significant regression equation was found, F(3, 59) = 4.72, p = .005 with an R² of .45. Both condition (β = .34, p = .006) and age (β = .26, p = .045) were significant predictors, with condition being the stronger of the two, but the interaction between age and condition (β = −.05, p = .680) was not a significant predictor of children’s correct identification.

In sum, while older children were better than younger children at choosing the correct novel monsters overall, the effect of condition was a unique predictor of their performance. This effect of condition was driven by higher proportion correct scores in the prediction condition than in either the reflection condition or the control (see Figure 1).

Figure 1.

Children’s proportion of correct monster name identifications across both story readings by condition. The x’s represent mean values, horizontal lines within the boxes represent median values, and the ‘whiskers’ represent upper and lower quartiles of the proportion of correct scores in each condition.

In order to look more closely at the possible effects of whether the pauses before and after the novel monster names were silent or eliciting, we conducted a pause type analysis using a 2 (pause type) × 2 (condition) analysis of variance of children’s total proportion correct identification responses again including age as a covariate (ANCOVA). This analysis focused only on the prediction and reflection condition responses because in the control condition there was no pausing, and thus no way to include pause type as a factor. The ANCOVA revealed that there was no main effect of pause type, F(1, 37) = .85, p = .362, η_p² = .02, no main effect of age, F(1, 37) = .61, p = .440, η_p² = .02, and no pause type and age interaction, F(1, 37) = .79, p = .362, η_p² = .02. However, there was a main effect of condition, F(1, 37) = 4.57, p = .039, η_p² = .11, and a significant pause type by condition interaction, F(1, 37) = 7.10, p = .011, η_p² = .16.

Follow-up tests showed that when the data were split by condition, children in the prediction condition had marginally higher mean correct identification scores for monsters that were preceded by silent pauses (M = .74, SD = .21) than those preceded by an eliciting ‘What’s he called?’ prompt (M = .66, SD = .19), t(19) = 1.85, p = .079, d = .39. In the reflection condition the pattern was reversed – children had higher correct identification scores for monsters that were followed by an eliciting question (M = .63, SD = .21) than those that were presented and then followed by a silent pause (M = .52, SD = .21), t(19) = −2.17, p = .043, d = .50 (see Figure 2). Looking at these results another way, split first by pause type, we found that children who heard silent pauses before the target monsters were named identified significantly more of them than children who heard silent pauses after targets were named, t(38) = 3.38, p = .002, d = 1.05, but that when eliciting questions were used there was no difference in correct identification of monsters for children who heard the eliciting question before or after the targets were named, t(38) = .488, p = .628, d = .15. Finally, when comparing all four proportion correct identification means to chance performance on identification (.50), we found that in the prediction condition, silent pauses resulted in better than chance performance, t(19) = 5.13, p < .001, d = 1.14, as did eliciting question pauses, t(19) = 3.60, p = .002, d = .82. In the reflection condition, however, eliciting pauses resulted in better than chance performance, t(19) = 2.65, p = .016, d = .60, while silent pauses did not, t(19) = .40, p = .691, d = .09.

Figure 2.

Children’s proportion of correct monster identification scores by condition (prediction or reflection) and by type of pause (silent or eliciting). The x’s represent mean values, horizontal lines within the boxes represent median values, and the ‘whiskers’ represent upper and lower quartiles of the proportion of correct scores in each condition.

In sum, while children performed better overall on novel monster identification when they had heard pauses occurring before those target words in the story, the beneficial effect of this pre-target pausing was greatest when the pauses were silent rather than filled with an eliciting question. On the contrary, silent pausing after a target word produced the worst results on later novel word identification – no difference from chance performance.

Production

In addition to testing children’s ability to identify the newly learned monster names in a forced-choice task, we also asked children to produce the monster names (e.g., showing a child each monster picture and asking, ‘What is this monster called?’). The number of correct monster names that a child produced yielded a production score between from 0 to 8 for each participant. In all three conditions, these scores were decidedly low, averaging less than one monster named out of eight tries (M_prediction = 0.68, SD = 1.03, M_reflection = 0.70, SD = 0.83 and M_control = 0.55, SD = 0.86), and negatively skewed because many participants could not correctly produce the names of any of the monsters when asked to do so in free recall. In order to compare children’s correct novel monster identifications across the three conditions, we conducted a one-way analysis of variance with condition as a between-subjects factor and age of the children as a covariate. This revealed a significant main effect of age, F(1, 54) = 7.14, p = .010, η_p² = .12, but no significant effect of condition, F(2, 54) =.48, p = .621, η_p² = .02, and no significant interaction between age and condition, F(2, 54) =.45, p = .639, η_p² = .02.

In order to test whether there was any pause type (silent or eliciting) effect on children’s correct productions, we conducted a 2 (pause type) × 2 (condition) analysis of variance of children’s correct production responses including age as a covariate (ANCOVA) as we had done for children’s identification responses (described above). This ANCOVA showed that there was a main effect of age, F(1, 37) =5.23, p = .028, η_p² = .12, but no main effect of condition, F(1, 37) =.20, p = .659, η_p² = .01, or pause type, F(1, 37) =.32, p = .577, η_p² = .01, and no interaction between age and pause type, F(1, 37) =.35, p = .559, η_p² = .01. The interaction between pause type and condition was also not significant, F(1, 37) = 2.63, p = .114, with a medium effect size, η_p² = .07.

Verbalized guesses and repetitions

Because children often spoke up during the reading of the story when there was a long pause or when they were prompted, and even occasionally without prompting (e.g., repeating the monster name after it was labeled in the prediction condition), we were able to measure the number of verbalized guesses children made (whether correct or incorrect) before the target monster names were read aloud, and also how many repetitions children made directly after hearing the monsters named during the reading of the stories. While guesses and repetitions happened frequently in both the prediction condition (M = 7.20 verbalizations) and reflection condition (M = 10.89 verbalizations), in the control condition only one child out of 20 repeated a monster name, and this child only did that once out of 16 presentations (on the last trial). Thus, we did not include the control condition in our analyses of guesses and repetitions.

In order to look more closely at guesses and repetitions in the prediction and reflection conditions, we first conducted a 2 (condition) × 2 (pause type) × 2 (response type) mixed-design analysis of variance on children’s overall amount of verbalized responses with condition (prediction or reflection) as a between-subjects factor, and pause type (silent or eliciting) and response type (guess or repetition) as within-subjects factors, and again included age as a covariate. This mixed ANOVA revealed no main effects of condition, F(1, 36) = 3.44, p = .072, with a medium effect size, η_p² = .09, pause type, F(1, 36) = 1.32, p = .259, response type, F(1, 36) = .67, p = .417, or age, F(1, 36) = 3.22, p = .081, nor any interaction effects of age with the other factors, all F’s < 1.00, p’s > .350. There was no difference overall in how much verbalizing there was between children in the prediction condition (M = 7.20, SD = 7.90) and in the reflection condition (M = 10.89, SD = 6.28), t(37) = −1.61, p = .116, d = .52. There was, however, a significant interaction between condition and pause type, F(1, 36) = 13.57, p = .001, η_p² = .27, such that children in the prediction condition produced responses equally as often when there was a silent pause (M = 3.45, SD = 4.01) or a prompt (M = 3.75, SD = 4.01), t(19) = −.97, p = .343, d = .22, but children in the reflection condition produced more responses when there was a prompt (M = 6.95, SD = 3.50) than a silent pause (M = 3.95, SD = 3.39), t(18) = −4.59, p < .001, d = 1.05. There was also a significant interaction between condition and response type, F(1, 36) = 51.06, p < .001, η_p² = .59, such that children in the prediction condition produced more guesses (M = 5.00, SD = 5.60) than repetitions (M = 2.20, SD = 3.81), t(19) = 2.31, p = .032, d = .52, but in the reflection condition children produced more repetitions (M = 10.90, SD = 6.28) than guesses (M = 0.00, SD = 0.00), t(18) = −7.56, p < .001, d = 1.73. Finally, there was also a significant three-way interaction effect of condition, pause type, and response type, F(1, 36) = 24.14, p < .001, η_p² = .40. In the prediction condition, guesses were marginally more common after an eliciting question (M = 2.75, SD = 2.88) than after a silent pause (M = 2.23, SD = 2.83), t(19) = −2.032, p = .056, d = .49, while for children in the reflection condition repetitions were more common after an eliciting question (M = 6.95, SD = 3.50) than after a silent pause (M = 3.95, SD = 3.39), t(19) = −4.592, p < .001, d = 1.06 (see Figure 3).

Figure 3.

Children’s total responses by type (guesses or repetitions) by condition (prediction or reflection) and by type of pause (silent or eliciting). Error bars represent standard error of the mean.

In sum, children’s spontaneous responding occurred largely as we would expect based on the pragmatics of the situation – children guessed the name of an upcoming monster more often when the reader paused and invited a prediction before revealing the monsters’ names and children repeated the monsters’ names aloud more often when they were asked ‘What’s he called?’ right after the monster names were heard.

Lastly, in order to compare how verbalizing guesses and producing repetitions aloud during the storybook reading might relate to children’s later retention of the new monster names, we conducted a multiple linear regression analysis to predict children’s proportion correct scores on the identification task based on age, total number of guesses produced, and total number of repetitions produced by the children. There was not a significant regression equation found, F(3, 38) = .27, p = .847 with an R² of .02. Neither age (β = .12, p = .485) nor guesses (β = .03, p = .869) nor repetitions (β = −.07, p = .708) were significant predictors of children’s downstream correct identification. We also conducted a multiple linear regression analysis to predict children’s proportion correct scores on the production task based on age, total number of guesses produced, and total number of repetitions produced by the children. Again, there was not a significant regression equation found, F(3, 38) = 2.47, p = .078 with R² = .18. While age (β = .35, p = .040) was a significant predictor, neither children’s guesses (β = .08, p = .653) nor repetitions (β = .18, p = .266) were significant predictors of children’s downstream production of the newly learned monster names.

Thus, while children were clearly verbalizing guesses and repetitions especially under the conditions in which they were given time and prompting, the extent to which they made guesses or repetitions aloud did not relate to how well they were able to remember and identify the novel monster names when tested.

General discussion

Taken all together, these findings demonstrate that in shared storybook reading with children, a pause right before a novel word is read helps children later retain that word better than either a pause after a novel word is read or verbatim reading without pauses. If pitted against one another, pausing pre-target is more helpful than pausing post-target, though whether it is helpful to ask an eliciting question during a pause or not depends on whether it comes before or after a target word – silent pauses are more effective than eliciting questions before novel vocabulary, but eliciting questions are more effective than silent pauses after novel vocabulary. It is also of note that actually verbalizing predictions or repetitions of the target words during pauses does not appear necessary for better retention. Thus, we would argue that overall silent dramatic pausing before novel vocabulary during shared book reading is a useful strategy for helping children attend to and remember that vocabulary.

We found that there were observable benefits for this simple, yet subtle reading strategy. In this study, children were able to initially map never-before-heard monster names to their referents after only hearing them presented in a story twice, and the benefits of pausing before the monster names were measurable even without any further elaboration or repetition of the new words. This may be because the text of the story itself already presented the words in a clear and salient way. The novel monsters were the singular focus of each page both in the text and in the illustration (e.g., Flack & Horst, 2018). Also, the monster names were both the last word presented on each page and were the completions of rhymed stanzas (e.g., Read, 2014). Thus, each monster name was already both semantically and prosodically highlighted. This kind of highly structured rhyming text is not uncommon in the shared reading experiences of children in our age group, as rhyming books have been estimated to make up as much as 30% of families’ home storybook collections (Read, 2014). However, if new words in a story were not already so prominently placed or central to the narrative, simply pausing before introducing them may not be enough to make them more memorable. When Rice, Buhr, and Oetting (1992) tested the effect of pausing before novel high-level vocabulary words in a study with children with specific language impairment, they did not find any benefit of pre-target pauses on children’s word learning. However, in that study the story text did not rhyme, and did not consistently introduce a single class of novel words in the same position on every page, and so may not have ‘set up’ the vocabulary as predictably as in the more highly constrained stories that we used. Furthermore, in the Rice et al. study, children heard recordings rather than a live reader, which could have affected the pragmatics of the pausing and demands of the task – for example, when a live reader pauses and glances at the listener it may indicate that it is the listener’s turn to contribute, but it is not clear that children would respond the same way to a pause within a recording. The comparison of our findings to those of Rice et al. indicates some limits to the generalizability of these findings. It is not simply that pausing is a universally beneficial strategy for vocabulary teaching, but that it is a highlighting cue that can be used by a reader to complement other helpful cues already present within the book.

We have also found in this study that not all pausing is alike – pausing before a word is pragmatically different than pausing after a word, and thus, whether an eliciting prompt is necessary to actively engage a child appears to differ. When pausing before a novel word it is best to have a silent pause, and when pausing after a novel word, it is best to have an eliciting pause. Closely watching the video recordings of each storybook reading session helps explain the pragmatics of each type of pause and better understand the outcomes of these four conditions. Silently pausing before reading a monster’s name seemed to signify to the listeners that an important word was about to be said and appeared to act as an attention-getter for young listeners. Most children looked back and forth between the page and the reader during this pre-target pause, seemingly waiting to hear what the monster was called, and children verbalized guesses of the monster’s name 28% of the time, without being directly prompted. However, when the silent pause occurred after a monster’s name, the children seemed to be confused or distracted, possibly wondering why the reader was not moving on to the next page; this silent post-target pause seemed to hinder the story’s momentum. On the other hand, when an eliciting pause occurred after a monster’s name, children seemed encouraged to reflect on and repeat the novel names, and did so 87% of the time, but an eliciting pause occurring just before the monster’s name seemed to have a more distracting effect especially the first time that children heard the story. While children did offer up guesses of the monster names, or rhymes (e.g., ‘Pooz’ for Smooze) or descriptors of the monster (e.g., ‘Big Nose’ for Groze) with eliciting pauses 34% of the time, some children appeared to be confused as to why they were being asked ‘What’s he called?’ before they had heard the monster’s name revealed, and occasionally responded during the eliciting pause with a simple ‘I don’t know.’

Given the pragmatics of the situations, it is not surprising that when examining the guesses or repetitions that children made during the storybook reading, more guesses and repetitions were made by children when they were directly elicited, than when there was a silent pause (see Figure 3). By asking just a simple question, such as ‘What’s he called?’ as in this experiment, readers can elicit more responses from children, and potentially get them more actively engaged in the story (e.g., Wasik & Bond, 2001; Whitehurst et al., 1988). However, in this study there was no correlation between verbal responses and the likelihood that children remembered the novel word after finishing the story. Where we might have expected a production effect (e.g., Icht & Mama, 2015), we did not find a relationship between children actually saying the monster names aloud and recognizing or producing those names better at test. This could be because of the challenging nature of the task of making many new word–picture mappings in one session (e.g., Zamuner et al., 2018), or because the highlighting of the words in the context of the story was already sufficient without having to say them out loud (e.g., Ard & Beverly, 2004). In either case, we found that children do not have to actually verbalize their guesses or reflections in order to benefit from the space to make them.

An important question to ask with respect to our interpretation of the findings here is whether the pre-target pause really is encouraging active prediction or whether it is just helping draw attention to the word it precedes. The pre-target pauses in this study did appear to encourage children to take a turn, and make a prediction. We see in this study and others (e.g., Moerk, 1972; Read et al., 2014) that children take the pause as an opportunity to guess aloud, but future work will test if children use pre-target pauses within story reading to make silent predictions as well by tracking their anticipatory looking patterns. We already know that even without pauses children in this age range are able to rapidly anticipate words at the ends of sentences using many available types of constraining cues such as semantics (e.g., Borovsky, Elman, & Fernald, 2012) and prosody (e.g., Gambi, Gorrie, Pickering, & Rabagliati, 2018), and so given extra time and a consistent pattern of introducing attention-getting novel monster names at the end of each stanza in this book, why wouldn’t children be in a state of anticipation during a pre-target pause in this study?

Because in this study correct pre-target guesses were exceedingly rare, we have no evidence whether correct or incorrect predictions led to better retention – children did not have to be right (or even say anything) to benefit from the pause. Thus, in this case it is not likely that the benefit of a pre-target pause comes from making a lexical prediction and then subsequently observing a match or mismatch between that prediction and the word that is heard. Rather, we would argue that the pause allowed the time and encouraged the effort to make predictions when it occurred before the monster names, and it was the effort to predict and how that effort may have heightened children’s attention that made the new names that followed a pause more memorable. In a recent review of educational media supports for teaching vocabulary, Neuman, Wong, Flynn, and Kaefer (2019) categorized the use of a pause before the introduction of a target vocabulary item as an attention-directing cue. Rice et al. (1992) also suggested that pauses before novel words could be potentially helpful for making words easier to segment from the stream of speech. We would argue that the pre-target pauses used in this study within the highly constrained monster rhymes that were used both oriented children’s attention to the upcoming word (e.g., Ferreira & Chantavarin, 2018), and also invited active lexical prediction. The pre-target pauses used in these books may have given children an opportunity to gather cues to a monster’s identity (e.g., the distinctive features that rhymed with the upcoming name) and prepare to integrate those cues with the upcoming novel name, even as they also tried to predict it. Thus, attentional orienting and active prediction may work together to improve retention of these new vocabulary items.

One surprising finding from this research was that there was no effect of whether it was children’s first or second time hearing the story on their novel word recognition. We would have expected that after hearing the story twice and having already been questioned about the novel monsters once in between readings that children would have improved in their recognition of the novel monster names on the second test, but this was not the case in the current study. Unlike other researchers (e.g., Horst et al., 2011; Robbins & Ehri, 1994), we may not have seen improved novel vocabulary retention with story repetition because the number of exposures to the new words in our study was still quite limited or there may not have been enough time in between readings for children to consolidate their new vocabulary learning. In future research, we would like to investigate whether when the story is repeated more often and over more extended periods of time, as a book would typically be read in the home, this might increase the benefits of pausing. More repetition of the story might also improve the pragmatics of pausing – stopping to allow a child to make a guess would be more natural if the reader and the child were more confident that because the child had heard the book several times, he or she could correctly guess an upcoming word. In fact, in the current study we did see more guessing aloud in the prediction condition during the second story reading (M = 2.00 guesses, SD = 2.73 during the first story compared to M = 3.00 guesses, SD = 3.13 the second time they heard the story, t(19) = −2.52, p = .021, d = .57). While this increased guessing aloud did not affect how many monsters children correctly remembered at test after just two repetitions of the story, there is still the possibility that this kind of engagement over a longer period of time could bolster learning.

A third avenue of future research would pursue the combination of useful reading strategies for vocabulary retention. The design of the current experiment limits us from investigating how the combination of a silent pause before a novel word with an eliciting pause after the same novel word may affect language development among young listeners. Further research should attempt to determine the effects of this combination: Would the use of these two techniques together cumulatively benefit children’s vocabulary retention, and also would these pausing techniques used with other more overt dialogic reading strategies be beneficial? One could imagine reading a new vocabulary word within a story with a dramatic pause used before its introduction as a way to elicit a prediction, and then an explicit post-target discussion of the meaning of the word and the relationship of the target word with alternatives that the child might have guessed prior to hearing it. For example, an adult might read, ‘my floppy ears might look quite funny / if I were a hopping …’ and pause dramatically to let the child guess the next word, should the child guess ‘kangaroo’ the reader could read ‘bunny,’ elicit a repetition, but also follow up with a comment on the difference between bunnies and kangaroos. These vocabulary highlighting techniques could potentially be used in concert to help a child remember new words, connect them with what they already know, and engage with the story.

In sum, research indicates that by incorporating simple techniques during storybook reading, adult readers can help children learn novel vocabulary words that they find within the text. The present study found that by incorporating pre-target pauses into storybook reading, readers can support a child’s new word identification. This experiment further emphasizes the importance of giving children the opportunity to interact during storybook reading. By giving children the space to prepare for an upcoming word and even predict that word, the reader ensures that they are reading with a child instead of reading at a child.

Footnotes

Appendix

Acknowledgements

We are indebted to the many children and parents who freely volunteered and participated in this study and the partnership of Kids on Campus Child Development Center at Santa Clara University. We would also like to acknowledge the help and feedback received from research colleagues in the Read Lab.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Kirsten Read

References

Ard

L. M.

Beverly

B. L.

(2004). Preschool word learning during joint book reading: Effect of adult questions and comments. Communication Disorders Quarterly, 26, 17–28.

Beck

I. L.

McKeown

M. G.

(2007). Increasing young low-income children’s oral vocabulary repertoires through rich and focused instruction. The Elementary School Journal, 107, 251–271.

Borovsky

Elman

J. L.

Fernald

(2012). Knowing a lot for one’s age: Vocabulary skill and not age is associated with anticipatory incremental sentence interpretation in children and adults. Journal of Experimental Child Psychology, 112, 417–436.

Clark

E. V.

(2007). Young children’s uptake of new words in conversation. Language in Society, 36, 157–182.

Clark

E. V.

(2010). Adult offer, word-class, and child uptake in early lexical acquisition. First Language, 30, 250–269.

DeBaryshe

B. D.

(1993). Joint picture-book reading correlates of early oral language skill. Journal of Child Language, 20, 455–461.

Elley

W. B.

(1989). Vocabulary acquisition from listening to stories. Reading Research Quarterly, 24, 174–187.

Evans

M. A.

Reynolds

Shaw

Pursoo

(2011). Parental explanations of vocabulary during shared book reading: A missed opportunity. First Language, 31, 195–213.

Ewers

C. A.

Brownson

S. M.

(1999). Kindergarteners’ vocabulary acquisition as a function of active vs. passive storybook reading, prior vocabulary, and working memory. Journal of Reading Psychology, 20, 11–20.

10.

Farrant

B. M.

Zubrick

S. R.

(2013). Parent–child book reading across early childhood and child vocabulary in the early school years: Findings from the Longitudinal Study of Australian Children. First Language, 33, 280–293.

11.

Ferreira

Chantavarin

(2018). Integration and prediction in language processing: A synthesis of old and new. Current Directions in Psychological Science, 27, 443–448.

12.

Flack

Z. M.

Field

A. P.

Horst

J. S.

(2018). The effects of shared storybook reading on word learning: A meta-analysis. Developmental Psychology, 54, 1334–1346.

13.

Flack

Z. M.

Horst

J. S.

(2018). Two sides to every story: Children learn words better from one storybook page at a time. Infant and Child Development, 27, e2047.

14.

Gambi

Gorrie

Pickering

M. J.

Rabagliati

(2018). The development of linguistic prediction: Predictions of sound and meaning in 2- to 5-year-olds. Journal of Experimental Child Psychology, 173, 351–370.

15.

Horst

J. S.

Parsons

K. L.

Bryan

N. M.

(2011). Get the story straight: Contextual repetition promotes word learning from storybooks. Frontiers in Psychology, 2, 17.

16.

Houston-Price

Howe

J. A.

Lintern

N. J

. (2014). Once upon a time, there was a fabulous funambulist…: What children learn about the ‘high-level’ vocabulary they encounter while listening to stories. Frontiers in Psychology, 5, 75.

17.

Icht

Mama

(2015). The production effect in memory: A prominent mnemonic in children. Journal of Child Language, 42, 1–23.

18.

Jimenez

S. R.

Saylor

M. M.

(2017). Preschoolers’ word learning and story comprehension during shared book reading. Cognitive Development, 44, 57–68.

19.

McLeod

A. N.

McDade

H. L.

(2011). Preschoolers’ incidental learning of novel words during storybook reading. Communication Disorders Quarterly, 32, 256–266.

20.

Moerk

(1972). Principles of interaction in language learning. Merrill-Palmer Quarterly of Behavior and Development, 18, 229–257.

21.

Neuman

S. B.

Wong

K. M.

Flynn

Kaefer

(2019). Learning vocabulary from educational media: The role of pedagogical supports for low-income preschoolers. Journal of Educational Psychology, 111, 32–44.

22.

Read

(2014). Clues cue the smooze: Rhyme, pausing, and prediction help children learn new words from storybooks. Frontiers in Psychology, 5, 149.

23.

Read

Macauley

Furay

(2014). The Seuss boost: Rhyme helps children retain words from shared storybook reading. First Language, 34, 354–371.

24.

Read

Quirke

(2018). Rhyme and word placement in storybooks support high-level verb mapping in 3- to 5-year-olds. Frontiers in Psychology, 9, 889.

25.

Rice

M. L.

Buhr

Oetting

J. B.

(1992). Specific-language-impaired children’s quick incidental learning of words: The effect of a pause. Journal of Speech, Language, and Hearing Research, 35, 1040–1048.

26.

Robbins

Ehri

L. C.

(1994). Reading storybooks to kindergartners helps them learn new vocabulary words. Journal of Educational Psychology, 86, 54–64.

27.

Sénéchal

(1997). The differential effect of storybook reading on preschoolers’ acquisition of expressive and receptive vocabulary. Journal of Child Language, 24, 123–138.

28.

Sénéchal

Cornell

E. H.

(1993). Vocabulary acquisition through shared reading experiences. Reading Research Quarterly, 28, 360–374.

29.

Trelease

(2013). The do’s and don’ts of read-aloud. In Trelease

(Ed.), The read-aloud handbook (7th ed., pp. 73–79). New York, NY: Penguin.

30.

Walsh

B. A.

Blewitt

(2006). The effect of questioning style during storybook reading on novel vocabulary acquisition of preschoolers. Early Childhood Education Journal, 33, 273–278.

31.

Wasik

B. A.

Bond

M. A.

(2001). Beyond the pages of a book: Interactive book reading and language development in preschool classrooms. Journal of Educational Psychology, 93, 243–250.

32.

Whitehurst

G. J.

Falco

F. L.

Lonigan

C. J.

Fischel

Debaryshe

Valdez-Menchaca

M. C.

Caulfield

(1988). Accelerating language development through picture book reading. Developmental Psychology, 24, 552–559.

33.

Whitehurst

G. J., &

Pearson Learning Group. (2006). Read together, talk together: A dialogic reading program for young children. Parsippany, NJ: Pearson Early Learning.

34.

Yopp

H. K.

Yopp

R. H.

(2000). Supporting phonemic awareness development in the classroom. The Reading Teacher, 54, 130–143.

35.

Zamuner

T. S.

Strahm

Morin-Lessard

Page

M. P. A.

(2018). Reverse production effect: Children recognize novel words better when they are heard rather than produced. Developmental Science, 21, e12636.