Abstract
This study examines the potential benefits of rhyme on young children’s word retention during shared reading. In two experiments, 2- to 4-year-old children heard their parent read either a rhymed or non-rhymed version of the same animal story, and were then tested on how many animal names they subsequently recognized from the story in Experiment 1 and could correctly identify in Experiment 2. In both experiments, children performed better in the rhyme condition across the age range despite differing levels of word familiarity. While there were no other differences between conditions in parents’ reading styles or the emphasis placed on the animal names, parents’ dramatic pausing just before reading animal names may have promoted children’s ability to anticipate animals before they were initially named. These findings supported the hypothesis that rhyme combined with parental behavior can facilitate active prediction on the part of children, which in turn may contribute to their word retention and learning from the storybooks.
Keywords
Reading with young children has undeniable benefits for their early language development. Among many other positive outcomes, shared book reading supports young learners in building vocabulary (Arterberry, Midgett, Putnick, & Bornstein, 2007; Blewitt, Rump, Shealy, & Cook, 2009; DeTemple & Snow, 2003; Farrant & Zubrick, 2013; Senechal, LeFevre, Hudson, & Lawson, 1996); and we know from empirical research that learning vocabulary from storybooks is mediated by individual characteristics of children as listeners (e.g., Fletcher & Reese, 2005; Laakso, Poikkeus, Eklund, & Lyytinen, 2004; Senechal, Thomas, & Monker, 1995), the amount of elaboration and interactivity in the reading styles of parents (e.g., Ard & Beverly, 2004; Reese, Cox, Harte, & McAnally, 2003; Whitehurst et al., 1998) and even by the amount of repetition of the same story over and over (e.g., Horst, Parsons, & Bryan, 2011). Recent work has shown how differences in book genre (narrative or didactic books) can differentially affect parents’ reading styles (Nyhout & O’Neill, 2013). Beyond that, however, we know very little about the role of the books themselves in this process (e.g., Fletcher & Reese, 2005; van Kleeck, 2003), specifically about how the style of the text might influence word learning.
Children’s books are beloved for many reasons, but one common irresistible feature is the song-like quality of the language that comes from rhythm, repetition and often rhyme. In a local survey of 165 parents of 2- to 4-year-old children, parents reported that on average rhyming books make up 38% of their home libraries, and 20 of the ‘Top 100 children’s books of all-time’ (2013) for children aged 0–5 are in verse. Thus rhyme is ubiquitous in the modern child’s experience of shared storybook reading, and for a reason – children enjoy it. In Hayes, Chemelski, and Palmer’s (1982) experiments, which pitted rhymed vs. non-rhymed versions of stories against one another for the purpose of testing children’s recall, regardless of how well children remembered the stories, they consistently liked the rhyming versions of the stories more. This seems unsurprising as it fits anecdotally with the experience of parents, teachers and children’s book writers, but is there some other advantage to rhyme over prose besides the delight in it?
Theorists have posited a broad view of the benefits of rhyme, such as Kenney (2005) who wrote, ‘The more rhymes [children] know, the more ideas they have to think about’ (p. 28). And experience with rhymes, whether gleaned from storybooks or simply recited orally, does correlate positively with some wide-ranging measures of language development. There is a positive link between nursery rhyme knowledge and familiarity and later reading outcomes which is assumed to proceed through the heightened phonological sensitivity that rhyme offers to young listeners (e.g., Bryant, Bradley, Maclean, & Crossland, 1989; Dunst, Meter, & Hamby, 2011; Hayes, 2011). But, our own query about the potential benefits of rhyme is more specific – could rhymed stories help foster language development because they make the vocabulary within them stand out for memory and learning? Amassing a strong and diverse vocabulary is a central task of early language development, and storybooks provide exposure to words not often heard in daily life. When storybooks are rhymed, are those words potentially more memorable?
For adults in experimental settings rhyme makes word pairs (Bower & Bolton, 1969), exact words (Tillman & Dowling, 2007) and story events (Hayes et al., 1982) easier to remember than when pairs or passages are written in prose. What about children? Despite the ubiquity of rhyme in the life of a preschooler and the intuitions of parents and teachers, only a small amount of empirical research has looked at whether children’s books in verse provide children with any extra help in remembering what they learn from the story. There is a lot to be remembered from a story – the moral, setting, sequence of events, vocabulary, pictures, etc. – and those researchers in the last 40 years who have directly compared prose vs. rhyming stories found different things. Hayes and colleagues uncovered some of the negative impacts of rhyme on memory for 4- to 6-year-olds when asked about the events of a story (Hayes, 1999; Hayes et al., 1982; Johnson & Hayes, 1987). However, Sheingold and Foundas (1978) found that just as many details of a story were recalled by 6-year-olds when they heard rhyming vs. prose versions, but that the rhymed stories gave children an advantage in remembering the sequence of events. In addition, children recalled more of the specific rhyming words of the story compared to other details (Hayes et al., 1982) and demonstrated better verbatim recall (Johnson & Hayes, 1987) despite decrements in overall paraphrase abilities. Sheingold and Foundas (1978) wrote of the ways rhyme stories might be more interesting or memorable for children, but also how rhyme could potentially detract from the story’s overall narrative. To explain these mixed findings, Hayes (1999) focused on how deeply children process information when they encounter rhyme, and how processing at one level (e.g., remembering the words verbatim) may detract from the other (e.g., remembering the gist of the story). When it comes to vocabulary learning, however, a child must pay attention to the ‘shallow level’ phonology of a specific word in order to connect it to the ‘deeper level’ meaning. The two levels are connected – if you are trying to remember the word porcupine then it matters that you don’t say ‘squirrel.’
Rhyme might help children remember exact words, and thus potentially aid in strengthening less familiar words or even outright learning of novel words. This may not be solely because of the emphasis placed on the sounds of individual words, but also because stories written in verse increase the amount of predictability in the phrases children hear, which gives children an edge in language comprehension and language learning. When words are predictable, for example when they are placed in frequent frames, common phrases like brush your… teeth (Arnon & Clark, 2011; Fernald & Hurtado, 2006) or contextually specific phrases like the pirate buried the… treasure (Borovsky, Elman, & Fernald, 2012) then those words can be anticipated and identified more rapidly (Borovsky et al., 2012; Fernald & Hurtado, 2006), and produced correctly more often (Arnon & Clark, 2011). Rhyme, when added to other story cues like the narrative or illustrations, can also enable a child to anticipate words at the end of a line even before they are read. In a naturalistic study in family homes, Moerk (1972) even observed mothers taking advantage of this kind of predictability with young children to make storybook reading more interactive – mothers would read rhymes but deliberately leave off the last word in order to allow their young child to fill it in. Could this type of predictability, the kind that rhyme adds like an extra layer of context, make words more memorable? If a child’s anticipation of a word is confirmed, she might not only feel pleased in the moment, but might better remember that word for use in the future.
The current study looks directly at this question by testing children’s recognition (Experiment 1) and identification (Experiment 2) of nine animals presented in a natural shared storybook reading context in which the story was either rhymed or not. Because of our interest in how rhyme can strengthen predictability and aid in word retention we deliberately chose a book that supported predictability, and did not ask children to remember or learn anything else about the story than simply the names of the animals that they heard. Just as in real life, children were asked to remember precise words like porcupine, and thus we hypothesized that the ‘shallow’ processing for verbatim recall that rhyme might trigger combined with the predictability of the form would be an effective combination for making familiar and even novel vocabulary words from the story more memorable.
Experiment 1
Method
Participants
Twenty-four 2- to 4-year-old children participated (Mage = 40 months). Twelve of the participants were girls. All were learning English as their primary language without any reported language delays or disabilities. Each participant brought one parent with them (three fathers, the rest mothers) who had been recruited through parenting groups and an on-campus preschool in Santa Clara, CA. Participants tended to be from homes in which parents were college-educated and of moderate to high income levels. All children in the study were read to daily by the parent who accompanied them.
Materials
Two ‘laptop stories’ – one for each condition – were created for the study based on the commercially available book For just one day (Leuck, 2009). The original story is a 32-page rhymed narrative story about nine children who each imagine themselves as an animal for a day. The original book builds suspense with the rhyme structure as well as narrative clues before revealing each animal. For example in the couplet, ‘If I could have a spiky spine/I’d get to be a/[page turn] porcupine!’ the illustration of a girl as a porcupine and the word porcupine itself are not revealed until a page is turned. Each page of the original story was adapted so that it could be easily presented to parents on a laptop. For the rhyme condition, the story text, page breaks and illustrations were not changed. For the non-rhyme condition one or two words were changed in each couplet of the text to undo the rhymes from the original version while still preserving the content of the story, the dramatic page turns and the illustrations. (See Appendix 1 for target text of both conditions.)
Procedure
The study was conducted in a small quiet playroom meant to be inviting to children but relatively free of distractions. At the beginning of the session there was a short play period where children were invited to play with a puzzle while the procedure and consent form were explained to the parent. Afterwards, each child was asked if he or she would like to listen to their parent read a story ‘on the computer.’ Children and their parents sat at a small table with a 15 inch MacBook Pro® laptop. The video camera built into the laptop was used to record the story reading and child’s responses. The stories were displayed using PowerPoint© software, so that the parents could advance through the book using the arrow key and easily read page-by-page at their own pace as they would with a conventional print storybook. Each parent was told to read the story naturally, the way they would at home, and each child was told that afterwards the experimenter would ask some questions about the animals in the story. The last page of both versions of the story was an identical short rhyme about enjoying pretending to have animal features like ‘jagged jaws and furry paws,’ but ‘just for one day because/the other days I want to be/that special someone that is me.’ This was included so that there would be a short break between naming all the animals and testing the children’s retention.
Children were randomly assigned to either the rhyme or the non-rhyme condition, and depending on their assigned condition, the parent was given either the rhyming version of the book or the non-rhyming version.
Once the parent had read the story, the experimenter then asked the child if he or she could help her remember some of the animals. With the child’s assent, the experimenter proceeded with the retention test by asking the child the same-framed question ‘Was there a ____ in the story?’ for each of the animals. The test had 19 such questions: one ‘warm-up’ item easy for children to say ‘no’ to (e.g., ‘Was there a dinosaur in the story?’) followed by nine correct targets and nine distractors in one of two semi-randomized orders. Distractors were similar types of animals to the targets and were included to prevent children from simply developing a ‘yes’ bias and correctly identifying the target animals by default. The experimenter asked about all the animals in the same neutral, child-friendly tone and never gave feedback about whether the child was correct or incorrect, only positive encouragement. Parents were seated behind children during this testing period so as not to give any cues (intentional or unintentional) about the animals in question. The entire procedure took approximately 20 minutes.
Measures
Children’s accurate recognition
The main measure of Experiment 1 was a simple test of children’s animal name retention assessed with the ‘Was there a ____ in the story?’ questions asked by the experimenter at the end of the story. Each child’s accurate recognition score was a count of the total number of target animals they correctly indicated were in the story (out of a possible nine). If a child responded to the target questions with ‘yes,’ or clearly nodded, then they were given credit for a correct response; if they responded ‘I don’t know,’ ‘no,’ or by shaking their head, then they were not. In addition, one child answered the questions with either ‘probably’ or ‘no’ and for this child we counted ‘probably’ as a ‘yes’ response.
While we were primarily interested in the effect of condition on how well children retained the animal names that they heard in the story, we took additional measures in order to capture some of the individual variations that each parent and child brought to the storybook reading experience and to investigate those variables’ impact on retention.
Children’s guessing
We took two measures of children’s reaction to the story as they were listening to it: first, a count of how many times each child spontaneously said the name of an animal (made a guess) before the parent read the target; and second, the proportion of those guesses which were correct with respect to what the parent would then read (e.g., crocodile, but not alligator). While children’s prediction may have occurred without such overt verbal behaviors, measuring guesses was a window into how predictable the animal names were, as well as an indicator of how interactive children were with the story. They also enabled us to check for effects of condition on guessing, as well as whether there was any correlation between guessing out loud and later retention.
Parents’ reading
We also took three measures of how the parents read the stories in each condition. For each parent, we took two measures using just the audio of the story reading session converted into Audacity® sound editing software files for precision. First, we measured to the hundredth of a second the duration of the pause that preceded naming each target animal. Pause duration was an average for each parent of pause lengths across the nine targets. This measure was meant to investigate whether the amount of time parents allowed a child to anticipate an upcoming animal differed by condition and/or was correlated with later retention. Second, we measured the duration of each target label, in order to investigate whether the emphasis parents placed on the names of the animals differed based on rhyme or non-rhyme condition and/or was correlated with later retention. And third, we measured parents’ extra-textual talk by counting the total number of utterances that each parent made between starting the first stanza and starting the final ‘filler’ page. Extra-textual talk included comments such as, ‘Chimpanzees like bananas,’ questions such as, ‘What is that guy doing?,’ and responses to children’s questions, but not self-corrections or comments directed to the experimenter. This measure was meant to capture the total amount of conversation that was inspired by the book as an indicator of the parents’ reading style (e.g., Nyhout & O’Neill, 2013; Reese et al., 2003) and whether that differed by rhyme vs. non-rhyme condition and/or correlated with later retention.
Each of these measures was double-coded by trained research assistants. Interrater reliability was greater than 90%. Whenever there was disagreement, a discussion among the raters and the primary investigator settled the final measure.
Results
The descriptive statistics for each of our measures by condition can be found in Table 1. As can be seen from the table, the age of children did not differ by condition, nor did the gender distribution, as both the rhyme and non-rhyme conditions included six girls and six boys.
Summary of reading and recognition measures by condition.
Note: CI = confidence interval, means in
When we tested the effect of condition on the main measure of the study, children’s accurate recognition, we found significantly higher recognition in the rhyme condition compared to the non-rhyme condition, t(22) = −2.23, p = .04 (two-tailed). Thus, on average children remembered more of the animals when they heard the rhymed story. However, inspection of the data revealed that performance in the rhyme condition was not normally distributed and indicated a ceiling effect in which 11 out of the 12 participants in the rhyme condition got seven or more of the nine animals correct. This ceiling effect caused differing amounts of variability between the two conditions, and so we also conducted a non-parametric Mann–Whitney U test for independent samples, which also revealed a significantly higher mean for children’s accurate recognition in the rhyme condition compared to the non-rhyme condition, p = .05 (one-tailed).
Children’s guessing
Children were equally likely to make spontaneous guesses about upcoming animals regardless of condition. However, there was a significantly higher mean proportion of guesses which were correct for the seven children who made guesses in the rhyme condition compared to those eight children who made guesses in the non-rhyme condition, t(13) = −1.83, p = .045 (one-tailed) and the size of the condition effect on correct guessing was moderate, Cohen’s d = .65. It should be noted that coding the ‘correctness’ of guesses was straightforward – those children who verbalized guesses produced clear animal names. The largest variation from a correct target that we allowed was ‘pokey-pine’ for porcupine, while even near neighbors such as monkey for chimpanzee were counted as incorrect.
Parents’ reading
When we looked at the measures of how parents were reading to their children in each condition, we found no differences between conditions in pause duration, target duration, or extra-textual talk. The lack of differences between conditions in target duration or extra-textual talk suggests that parents were not more emphatic or discursive in one condition over the other. However, the measure of pause duration did show a marginal increase in the rhyme condition, t(20) = 1.51, p = .07, one-tailed, and the practical size of the effect of rhyme on pause duration was moderate, Cohen’s d = .45, suggesting that parents reading rhymed stories may have allowed more time for their child to predict the upcoming animal at the end of a rhyme couplet than parents in the non-rhyme condition. Additionally, when we looked closely at the types of extra-textual talk used by parents, we noticed in some cases parents would elicit a prediction from the child before revealing the animal’s name, e.g., ‘What animal do you think it is?’ This kind of prediction elicitation was used at least once by four out of the 12 parents in the rhyme condition (33%) but only by one parent in the non-rhyme condition (8.3%).
Correlation measures
Overall there were no significant correlations between age, parents’ reading measures, children’s guessing and accuracy. However, when we looked at the rhyme and non-rhyme conditions separately, we found a significant correlation in the non-rhyme condition alone. Children’s recognition of the animals from the story was positively associated with their age, r = .62, p = .03, two-tailed (compared to r = −.27, p = .39 in the rhyme condition), meaning that within the non-rhyme condition older children were more accurate in their recognition of the animals than younger children, but for the rhyme condition there was no relationship between age and accuracy. This was despite no differences between the two conditions in mean age of the children, but may have been affected by the ceiling effects observed for children in the rhyme condition.
Discussion of Experiment 1
The main finding of Experiment 1 was that children remembered more animal names when they heard them in the context of a rhyming story compared to an almost identical non-rhyming story. The differences between rhymed and non-rhymed target vocabulary did not seem to stem from differences in the way parents emphasized the words or from differences in extra-textual talk. Some parents said very little outside of the text of the story, while others used a great deal of extra-textual talk, but within-condition variance was greater than between-condition variance. In fact, the only differences that we found between the two conditions, besides how well children remembered the animals, were marginal differences in how long parents paused before naming the animal, how often they elicited a prediction, and how accurate children were at guessing the upcoming animals. These findings suggest a difference between conditions in how much parents might have encouraged prediction, as well as how easy it was for children to anticipate the animals. While correlations were not found between these latter measures and accurate recognition, likely because of ceiling effects in the rhyme condition, it may be that the improved predictability in the rhyme condition over the non-rhyme condition was helping to make the animals more memorable.
Experiment 1 gave us a glimpse into the connections between rhyme, prediction and word retention, and these results initially support our hypothesis – rhyme can facilitate memory for specific words in a story, and this boost in word recognition may be due in part to the predictability of rhyme. However, there were some clear limitations of this first study, which motivated the changes to come in Experiment 2.
First, in Experiment 1 we had no way of knowing just how familiar all the children were with the target words, and there may have been some harder (e.g., porcupine, chimpanzee) vs. easier vocabulary (e.g., bunny, butterfly). This presented us with a problem in interpreting our findings – an uncertainty about whether children were simply remembering familiar animals better from the rhymed story, or whether for some of the younger or lower vocabulary children who may have been unfamiliar with the animals before the story, some initial learning of their names might be taking place.
Another limitation was the ceiling effect found in Experiment 1, as it may have masked important potential relationships between parents’ reading, children’s guessing and what children remembered. The ceiling effect demonstrated that the yes/no recognition task may have been too easy, especially for older children who presumably had better verbal working memories in addition to stronger vocabularies, and especially for children who had received a boost from the rhymed books.
Finally, the working interpretation of Experiment 1 presupposes that parents may believe that their child can predict the upcoming animals. However, we had no record of parents’ expectations of their children’s level of understanding or ability to predict the targets. Qualitatively, it seemed that some parents paused longer to allow their children to guess, but we had no way of knowing whether these parents were thinking about whether their children could guess the upcoming animals or if for some parents their longer pausing in response to the rhymed stories was just an individually variable and reflexive response.
Thus, in Experiment 2 we made three significant changes to address some of these limitations: (1) we tested a younger and narrower age range; (2) we took a parent report measure of children’s prior knowledge of the target words; and (3) we used a picture-pointing task as a measure of word identification, rather than a yes/no recognition task, as our main dependent measure. Testing a younger and narrower age range and asking parents to report on their children’s knowledge of the target words would allow us to focus on children who may not already be familiar with all of the animals in the story. In combination with parents’ report of their children’s word knowledge, using an identification rather than a recognition task would allow us to examine whether children’s comprehension of the animal names was strengthened to a greater extent during the rhyme story as compared to the non-rhymed story. Collecting parents’ report measures prior to reading the stories could also give us insight into whether parents considered their own child’s knowledge when reading the story with them. And, finally, lowering the age range and using an identification rather than recognition task was also intended to reduce the possibility of ceiling effects in a new sample.
Experiment 2
Method
Participants
Twenty-two 20- to 35-month-old children participated (Mage = 26 months) in Experiment 2. Fifteen of the participants were girls. All were learning English as their primary language without any reported language delays or disabilities. Each participant brought one parent with them (two fathers, the rest mothers) who had been recruited through parenting groups and an on-campus preschool in Santa Clara, CA. Participants tended to be from homes in which parents were college-educated and of moderate to high income levels. All children in the study were read to at least daily by the parent who accompanied them. One additional participant was dropped from the study due to inattention during the story and refusal to participate in the testing.
Materials
The stories used in Experiment 2 were the same stories used in Experiment 1 with the addition of nine testing slides at the end of each. On each testing slide four animals were presented, one randomly placed in each quadrant: (a) the target animal (e.g., grizzly bear) (b) a ‘familiar’ distractor which was another animal from the story (e.g., chimpanzee) that would be a target on another trial, (c) a ‘near neighbor’ distractor (e.g., panda bear) and (d) an ‘unrelated’ distractor (e.g., penguin). The illustrations for the test trial animals were edited versions of animals taken from the inside cover of the original storybook by the same illustrator so that they would be of the same style as those seen within the story.
Procedure
The procedure used in Experiment 2 was the same as in Experiment 1, except for two changes. First, parents were given a questionnaire along with the consent form which asked them to report how often they read with their child, how much their child seemed to enjoy rhymed stories over non-rhymed stories on a Likert scale of 1–5, and for a list of 32 basic vocabulary words whether their child ‘understands’ and ‘says’ each word. The 32 words included the nine target animals that would be heard in the story along with 23 common nouns that norming data from the CLEX database (Jorgensen, Dale, Bleses, & Fenson, 2010) indicate upwards of 80% of 24-month-olds produce (e.g., dog, car, shoe) meant to serve as distractors. Parents were given an example before filling out the checklist of how a child might ‘understand’ some words but not ‘say’ them, and parents clearly understood how to complete it. The checklist provided both a parent-report measure of children’s comprehension and production of the specific animal names we would later test them on, and was also intended to prime parents to think about their children’s word knowledge as they were reading the story.
The second change to procedure was in the testing portion of Experiment 2. At the end of the story presentation, nine identification questions were asked of each participant instead of the retention questions asked in Experiment 1. Once the parent had read the story, the experimenter then asked the child if he or she could help her remember some of the animals. With the child’s assent, the experimenter proceeded with the identification test by asking the child the same-framed question, ‘Can you show me which one is a ____?’ for each of the target animals on each of the nine test slides. Children were asked to touch the animals on the screen to make their responses most clear. The participants clearly enjoyed this mode of response. The experimenter asked about all the animals in the same neutral, child-friendly tone and never gave feedback about whether the child was correct or incorrect, only positive encouragement. Parents were seated behind children during this testing period so as not to give any cues (intentional or unintentional) about the animals in question.
Measures
Children’s accurate identification
The main measure of Experiment 2 was a test of children’s animal name retention and identification assessed with the ‘Which one is the ____?’ questions asked by the experimenter at the end of the story. Each child’s accurate identification score was a count of the total number of target animals they correctly pointed to or touched when asked (out of a possible nine). Only children’s first choices were counted.
Parents’ reading
As in Experiment 1, in Experiment 2 we took three measures of how the parents read the stories in each condition: pause duration, target duration and total number of extra-textual comments.
Children’s guessing
Also as in Experiment 1, in Experiment 2 we recorded the number of times each child made a guess about an upcoming animal before hearing the animal’s name read during the storybook, and of those guesses how many each child got correct.
Parents’ report of target word knowledge
Additionally, because of the data provided by the parents in the initial vocabulary checklist, we calculated how many of the target items each parent reported their child to ‘understand’ (comprehend) and ‘say’ (produce).
Results
The descriptive statistics for each of our measures by condition can be found in Table 2. As can be seen from the table, the age of children did not differ by condition, nor did the gender distribution (rhyme: 8 girls, 3 boys; non-rhyme: 7 girls, 4 boys). Information obtained before testing from the parent questionnaires also showed no difference between conditions in how frequently children were read to (all parents in both conditions reported reading daily with their children), how much their child seemed to enjoy rhyme on a Likert scale from 1 to 5 (Mrhyme = 3.2, Mnonrhyme = 3.1), or, as shown in Table 2, how many of the target words parents reported that their child understood or said before hearing the story.
Summary of measures by condition.
Note: CI = confidence interval; means in
Children’s accurate identification
When we tested the effect of condition on the main measure of the study, children’s correct animal identification, we found significantly higher scores in the rhyme condition compared to the non-rhyme condition, t(20) = 2.05, p = .05 (two-tailed) and a moderate effect of rhyme on correct identification, Cohen’s d = .62. Thus, on average children correctly identified more of the animals from the story when they heard the rhymed version.
Parents’ reading
As in Experiment 1, when we looked at the three parent reading measures in Experiment 2, we found no differences between conditions in parents’ total extra-textual utterances and the average durations of target words (see Table 2 for means and 95% confidence intervals). As in Experiment 1, in Experiment 2 parents tended to pause longer before target words in the rhyme condition than in the non-rhyme condition. However, here the effect of rhyme was significant, t(18) = 2.27, p = .04 (two-tailed) and moderate, Cohen’s d = .72. It should be noted that we removed two clear outliers (one from each condition) from both the analysis of extra-textual utterances and pause duration because on both measures each of these outlier parents were well beyond three standard deviations above the mean (for extra-textual utterances: > 90 utterances, and for pause duration: mean pause durations > 4000 ms). Additionally, when we looked closely at the types of extra-textual talk that parents used, we noticed, as in Experiment 1, that in some cases parents would elicit a prediction from the child before revealing the animal’s name, e.g., ‘What animal do you think it is?’ This kind of prediction elicitation was used at least once by eight out of the 11 parents in the rhyme condition (73%) and by six of the 11 parents in the non-rhyme condition (55%).
Children’s guessing
Children’s spontaneous guessing of the upcoming animal before it was named was less common for the younger children of Experiment 2 than it had been in Experiment 1, with only four children in the rhyme condition ever making a guess and only two in the non-rhyme condition. Despite the small number of guessers, there were still more correct guesses in total in the rhyme condition (10 of 21) than in the non-rhyme condition (2 of 6).
Correlations
In Experiment 2 when we investigated bivariate correlations among the variables we measured, and as in Experiment 1, we found no relationship between children’s age and their correct identification of the target animals (r =.35, p >.10). However, children’s correct identification of animals was significantly correlated with (1) the duration of parents’ pauses before reading the target animal names (r =.45, p < .05); (2) parents’ report of their child’s understanding of the target words (r =.45, p < .05); and (3) parents’ report of their child’s production of the target words (r =.63, p < .01). The measures of parents’ reports of their children’s word knowledge were also correlated with average pause duration (both r’s > .45, p’s < .05). Taken together, this demonstrated that parents who thought their child knew more of the target words were also more likely to pause for a longer time before revealing them during the story and their children were also more likely to correctly identify more targets during testing, though these correlations alone do not indicate a causal relationship from parent expectations to lengthened pausing to better word identification as rhyme was the only variable systematically varied.
When we looked at the correlations within each condition (rhyme vs. non-rhyme) separately, we found that the positive relationship between how many target animals a child was reported to produce and how many they identified correctly was significant within both conditions (r’s >.60, p’s <.05) and for both conditions the relationship between correct identification and target words reported to be understood were marginally significant (r’s > .45, p’s < .08, one-tailed), but the positive relationship between parents’ reports of their children’s target animal knowledge and the duration of their pauses was only significant within the rhyme condition (r’s > .66, p’s < .05), not within the non-rhyme condition (r’s < .45, p’s > .15). Knowing more of the animals beforehand, and especially being able to produce them, was always related to being more successful at identifying them at test. However, it was only in the rhyme condition that children’s animal name knowledge prior to the story was related to parents pausing before revealing the animal names within the story – a differential response to the rhymed version of the story on the part of the parents.
Items and error analysis
While some animals were harder than others for children to correctly identify at test, ranging from 45% correct for chimpanzee, grizzly bear and rattlesnake to better than 75% correct for bunny, butterfly and crocodile, there were no animals that stood out individually as markedly easier or harder than the rest to identify. And, likewise, while parents had a range of responses as to whether their children might comprehend or produce each animal, for some animals they tended to overestimate their children’s ability (e.g., grizzly bear) while for others they tended to underestimate (e.g., porcupine) without any specific targets standing out in either direction. In general, the nine animals in this study represented a good range of words with which children in this sample and age range had varying degrees of familiarity. In addition, when we looked at the types of errors children made in the identification test, more of them were the choice of the ‘near neighbor’ distractor (49%) than the ‘familiar’ (31%) or ‘unrelated’ (20%) distractors, but the pattern did not differ by condition.
Discussion of Experiment 2
The main finding of Experiment 2 was that, similar to Experiment 1, children performed better after hearing a rhymed version of a story compared to a near identical non-rhymed version. In Experiment 2 children’s performance was measured with an identification test instead of a simple memory test. While success on the identification test could very well be aided by the children’s enhanced memory for the target animals, it also suggests that children in the rhyme condition were strengthening their knowledge of the animal names through the story, especially as we consider that before hearing the stories read, children in the rhyme and non-rhyme conditions did not differ in their knowledge of the animals according to parental report.
In Experiment 2, we also found that the rhymed stories supported the prediction of target animals, first by encouraging more out-loud guessing of the correct animal names before they were read by parents, and also indirectly by one important difference in the way parents themselves read the rhymed vs. non-rhymed stories. We found that parents tended to pause more just before revealing the target animal in the rhymed stories compared to the non-rhymed stories, giving children almost three-quarters of a second longer to anticipate the target when it would fit into a rhyme. In Experiment 1, we found a marginal effect of condition on pausing, but here in Experiment 2 with even younger children the effect was robust. Moreover, the length of parents’ pre-target pauses in the rhymed story was related both to how many of the target animals they believed their child to know, and also to how accurately their child identified the target animals at test.
General discussion
Taken together, both experiments show how rhyming stories can strengthen young children’s vocabulary. In Experiment 1 children remembered more animal names when they heard them in the context of a rhyming story compared to an almost identical non-rhyming story, and in Experiment 2 younger children, who may have been less familiar with the animals, nonetheless correctly identified more of them after hearing the rhymed version. In neither experiment did the differences between rhymed and non-rhymed target vocabulary stem from differences in the way parents emphasized the words or differences in parent elaborations in the two conditions. Importantly, the additional differences that we did find between the two conditions, besides how well children remembered or identified the animals, reflected how long parents paused before naming the animal, how often they elicited a prediction, and the amount of correct guesses children were able to make before they heard them.
These findings initially support our hypothesis that rhyme can benefit word retention and even word learning from shared storybook reading, and that there is a connection between the predictability that rhyme affords and the way that both parent readers and child listeners interact with that predictability to make the words more memorable.
This study cannot yet support a causal interpretation of the connection between the dramatic pausing that parents do more of when reading the rhymed stories and children’s improved word retention from rhymed stories, and future research will have to manipulate pausing and prediction elicitation in a more deliberate way to investigate their direct effects on word retention and learning. It may also be of interest to examine whether parents and children are more likely to have other kinds of interactions specifically around these pauses before the terminal word in rhymes. Evans, Reynolds, Shaw, and Pursoo (2011) found that, in general, while parents reading to kindergartners tended not to take up opportunities to explain unfamiliar words, they were more likely to discuss a new word when it was at the end of a page. Evans et al. suggest that the natural pause in storybook reading that occurs before a page turn might afford space for a parent and child to reflect and discuss, and they recommend exploiting this as a strategy for vocabulary development. In the case of our findings, the potentially helpful pause happens before the target vocabulary rather than just after, but still it provides a space for a ‘testing and feedback’ interaction that may highlight vocabulary. Future research will hopefully shed light on whether this kind of pausing within rhymed stories could also be used for effective word learning interventions. However, even just these observations of the existence of parents’ spontaneous and natural variation in dramatic pausing during storybook reading, and their differential use of it between rhymed and non-rhymed stories with otherwise very homogeneous groups of young listeners is a novel and thought-provoking finding.
While we cannot directly compare the age groups of the two experiments reported here, it is clear that children across the preschool age range have varied vocabulary size and amount of familiarity with words like the animal names used in this study. This may mean that in the course of a shared storybook reading different processes are taking place for children with different levels of familiarity with the highlighted words – for some children the story may boost simple short-term verbal memory for the familiar words it contains, other children might benefit from the story acting as a reminder of words they have heard but have a weaker representation of, and for some children the story may provide a chance to initially map an unfamiliar animal name to a referent. In each case, though, there is strengthening of words and that is at least the beginning of vocabulary learning. Here we have shown that this strengthening is happening more robustly when children and parents read a rhymed version rather than an unrhymed version of the story.
Our findings that rhyme can boost memory and recognition may not be surprising given that many people already view rhyme as an aide to memory of surface forms. For example, in his discussion of rhyme in oral tradition, Rubin (1995) makes it clear that there is a memory advantage from rhyme because of how it makes the exact sound of words important, therefore more easily recalled. We would add that young children engaged in storybook reading with an adult is a form of oral tradition, and for young children the exact words are important – they are fleeting for the pre-literate listener who does not have access to the text, and so when rhyme makes them more memorable, that is an advantage for the vocabulary of the young learner. This may be especially helpful when a child is not only hearing familiar words but learning and practicing new words. As Cook (2000, p. 27) hypothesized: [V]erse may help the child gain control over difficult new language and ideas. … The very regularity and predictability of the verse creates a necessary counterbalance to the extreme unpredictability – and consequently the slipperiness in memory – of the language and meanings which it carries.
Does this mean all children’s stories should be written in verse? No, clearly there is more to be gained from shared storybook reading than simply the highlighted vocabulary in a book, and some stories lend themselves better to rhyme than others. However, it is clear from the present results that rhyme would serve a child well when vocabulary is the focus of what a parent or teacher is trying to share through storybook reading, especially if that parent or teacher is allowing the rhyme to support a child’s prediction and active engagement with the vocabulary.
From Mary’s little lamb to green eggs and ham, rhyme is everywhere in the world of young children and they enjoy it, but perhaps not just because of its sound – as one parent in our study observed of her 3-year-old daughter, ‘She’s able to memorize our rhyming books and I think it makes her feel smart/confident to be able to complete the sentences.’ That feeling of confidence, the ability to predict and participate in the shared book reading because of the support of the rhyme, may make the experience more enjoyable and more beneficial to children.
Footnotes
Appendix 1
Acknowledgements
We would also like to acknowledge the help and feedback received from research assistants Jennifer Coleman, Briana Mitchell Andrew Weaver, and Madeline Regan.
Funding
This research was not financially supported, but we are indebted to the many children and parents who freely volunteered and participated in this study.