|Home | About | Journals | Submit | Contact Us | Français|
In six analyses using CHILDES (MacWhinney, 2000), we explored whether and how parents and their 1.5 to 5-year-old children talk about writing. Parent speech might include information about the similarity between print and speech and about the difference between writing and drawing. Parents could convey similarity between print and speech by using the words say, name, and word to refer to both spoken and written language. Parents could differentiate writing and drawing by making syntactic and semantic distinctions in their discussion of the two symbol systems. Our results indicate that parent speech includes these types of information. However, young children themselves sometimes confuse writing and drawing in their speech
Written language is glottographic: It symbolizes the sounds of speech (Sampson, 1985). For example, the written word dog symbolizes a dog through its relation to a spoken word in English. In order to learn how to read and write, children need to understand that the printed word dog represents a spoken word, which is itself a symbol.
Young children do not always appreciate writing’s glottographic nature, often treating writing as if it were much like drawing. Between the ages of two and four, children’s attempts at writing and drawing look quite similar and are difficult for even the children’s own parents to distinguish (Levin & Bus, 2003). There are also mechanical similarities in how young children create the basic elements of each system (Adi-Japha& Freeman, 2001). Although children improve in their ability to produce distinct markings for writing and drawing throughout the preschool years, their writings continue to contain pictorial elements. For example, three to five year olds sometimes assume that printed words should look like what they represent, using surface features of a word such as its size, color, or location on the page to determine which word it is (Bialystok, 1991; Levin & Bus, 2003; Levin & Tolchinsky Landsmann, 1989; Lundberg & Tornéus, 1978). Young children’s understanding of drawing is better developed, with three year olds showing a grasp of the intentional nature of pictures (Bloom & Markson, 1998; Gelman & Ebeling, 1998). Between the ages of three and four, children see the meanings of pictorial representations as relatively fixed, but they have difficulty understanding that printed words have stable meanings (Apperly, Williams, & Williams, 2004; Bialystok, 2000; Bialystok, Shenfield, & Codd, 2000). The developmental process by which children come to understand of the glottographic nature of written language is long and complex (Bialystok, 1992; Olson, 2002).
What experiences help children to understand the nature of written language? Formal instruction in school is one such experience. In the United Sates, explicit instruction about the nature of print often begins in kindergarten (around age 5), when children are taught the sounds of letters. Teaching children that m makes the sound/m/ or that s makes the sound/s/ can help them to understand that writing represents speech. Much research has focused on formal instruction, asking which types of explicit instruction best help children learn about the nature of writing (e.g., Snow & Juel, 2005).
Although instruction in school helps children understand the nature of writing, children begin to learn about some aspects of writing before formal instruction begins. These facets of development are often referred to as emergent literacy (e.g., Whitehurst & Lonigan, 1998). Many features of written language are visible in the child’s environment. For example, writing is generally presented on lines, adjacent symbols are not usually identical, and the marks do not resemble the objects they depict. Children learn about some of these surface properties of writing from exposure to print around them (Ganopole, 1987; Lavine, 1977). Although this exposure helps children become familiar with the general features of written language, it does not always result in knowledge about specific units of print. For example, children’s identification of the label on a soft drink bottle may remain unchanged even when essential features of the word, such as letters or letter order, have been changed (Masonheimer, Drum, & Ehri, 1984).
Children are exposed to writing in various situations. For example, they see labels for commercial products, as mentioned above, and writing in books. A number of studies have focused on storybook reading as a contributor to emergent literacy, assessing the home literacy environment by examining the frequency of book reading. However, preschool children spend more time looking at the pictures than at the print in books (Evans & Saint-Aubin, 2005; Justice, Skibbe, Canning, & Lankford, 2005). This means that storybook reading may play a relatively small role in fostering print knowledge. Activities that focus on writing and letters themselves may be stronger predictors of children’s emergent literacy skills. For example, North American children are encouraged to sing the alphabet song and to identify letter shapes. Time spent engaging in such letter-related activities predicts children’s later reading and spelling skills (Evans, Shaw, & Bell, 2000). Children’s literacy-related interactions with parents are not confined to routine activities such as singing the alphabet song and reading stories before bedtime (Aram & Levin, 2002). Throughout the day, parents may interact with their children as the children attempt to scribble words while playing or to identify printed words. Such interactions have been shown to influence children’s emergent literacy skills (Aram, 2006; Aram & Levin, 2004; Sénéchal, 2006; Sonnenschein, Baker, Serpell, Scher, Truitt, & Munsterman, 1997).
Previous findings provide a host of evidence about children’s emergent literacy in the preschool years and present a variety of ways in which the home literacy environment may be measured. However, the studies may have missed some of the everyday interactions that inform children about writing. In the present study, we explore the hypothesis that parents talk to even young children about writing and that the ways in which they do so provide evidence about how print works. This may happen even when writing is not the focus of attention. For example, in one conversation that we examined, a father tells his 2-year-old son that he’ll be ready to play as soon as I’m done writing my name (Kucjaz, 1976). The father is not teaching his son how this or other names are written, nor is he intervening in his son’s attempt to write. However, the father is conveying the idea that writing is a way of representing names. If conversations that touch on writing are common, they could help children learn about the nature of print. Investigating the ways in which parents and children talk about language may also allow us to detect knowledge about written language that young children have but that is missed in studies that require them to perform more complex tasks.
Research in other domains has shown that parent speech is a rich source of information that children can use to understand aspects of the world around them. For example, parents’ generic noun phrases such as birds lay eggs and dogs have fur provide information to young children about the nature of animal categories (Gelman, 2003). Although these are not explicit statements about the essential features shared by members of a species, they imply that the property in question refers to the category as a whole rather than to just some individuals. Parent speech also provides children with information regarding number words (Bloom & Wynn, 1997), category membership (Waxman & Markow, 1995), parts of speech (Brown, 1957), proper names (Hall, Lee, & Belanger, 2001), and the distinction between count and mass nouns (Hall et al., 2001; Soja, 1992).
To test the idea that parents speak to their young children about writing in ways that could provide information about the nature of print, we examined parent–child conversations in the CHILDES database (MacWhinney, 2000) that involved children between the ages of 1;6 (years; months) and 5;0. This database contains transcripts of parent–child conversations that were collected by various researchers in service of diverse research questions. Using this database is advantageous for addressing our research question, as it provides conversations across a range of generations, experimental conditions, social contexts, and everyday situations. Surveying this broad range of circumstances fits our exploratory purposes. We begin by asking whether parents and their young children talk about writing. If they do, then we can study the structure of this speech and what information it reveals about the nature of written language.
We report six specific analyses. The first two analyses explored the idea that adults might talk about writing in some of the same ways that they talk about spoken language. For example, an adult may tell a child that the printed word dog says dog, just as they tell a child that a person says dog. Such speech patterns could convey to children the similarity between speech and writing. To examine these issues, Analyses 1 and 2 explored parents’ use of the words say, name, and word in reference to spoken and written language. The other analyses addressed the question of whether parents speak about writing differently than drawing. Speaking about the two systems in distinct ways could reinforce the idea that writing and drawing are different symbol systems. Analyses 3 to 6, therefore, examined syntactic and semantic ways in which parent speech about writing may differ from parent speech about drawing.
In each of the six analyses, we also examined children’s speech in an attempt to gain insights into their developing knowledge of writing. Children’s understanding of the nature of writing has previously been assessed by examining their production and identification of various symbols. But young children have limited motor skills and do not always cooperate with or understand experimental tasks. By looking at spontaneous interactions between parents and their young children, we may gain new insights into children’s understanding of writing. For example, we can ask whether young children confuse writing and drawing in their talk about the two systems in the same way that they often produce similar products when asked to write and when asked to draw (Levin & Bus, 2003). We do not have sufficient data on each parent–child pair to determine whether differences among parents in their speech correlate with or predict children’s speech. Also, because CHILDES does not provide information on the children’s reading and writing skills, we cannot examine whether parent speech predicts children’s later literacy skills. Our initial exploration of whether and how parents and children speak about writing, however, should set the stage for future studies of these issues.
Parents might indicate that writing is a code for speech by using the same words to designate written and spoken language. The word say is an example of this. Parents may use say to refer to something that is spoken, as in Don’t say that, or something that is written, as in What does that sign say? Such speech could provide information about the similarity between print and speech. We examined use of say for written and oral language by parents of children aged 1;6 to 5;0, as well as children’s usage in this age range. Specifically, we asked whether parents use say for both types of language and whether children do the same.
This and the following analyses were conducted using transcripts of conversations from the CHILDES database (MacWhinney, 2000). English-language transcripts from both the U.S. and U.K. corpora that were available on the CHILDES website in August 2005 were considered for analysis. 1 The 34 corpora included were originally collected for analyses that are reported in publications spanning a broad date range, from one in 1929 to one in 2001. Aside from these two extremes, all other corpora are from publications between 1965 and 1998, the majority between the late 1970s and the mid 1990s. They thus reflect interactions between parents and children interactions from the latter part of the 1900s. The majority of the transcripts (approximately two-thirds) involved children from middle- to upper-middle class families, although one-fifth involved working-class families. The length and number of conversations available for any one child varied according to the purposes and methods of the original researchers. A number of the children were recorded on one occasion and others were recorded more than once.
To be included in the analysis, the target childhad to be recorded during the period of 1;6to 5;0 and the child had to be considered to be developing normally. A total of 607 children met the requirements and were included in the analyses. Transcripts were grouped by the child’s age. In Analysis 1 we used seven groups representing six-month periods across the 1;6–5;0 age span. The number of children included in each age group was as follows: 1;6–2;0: 119, 2;0–2;6: 147, 2;6–3;0: 250, 3;0–3;6: 230, 3;6–4;0: 40, 4;0–4;6: 22, 4;6–5;0: 60. This total (868) is higher than the 607 total children because some children were recorded more than once and had transcripts that fell into more than one age group. The reason that there are fewer transcripts in the older groups than in the younger ones is probably that many of the original researchers were interested in the beginning stages of spoken language development, particularly in early syntax. However, the transcripts from older children tended to be longer than those from younger ones. Given differences in the amount of data available at different age ranges, most of our analyses involve proportions of utterances of various kinds rather than raw numbers. To address the issue of development for children who had transcripts in more than one age group, we identified seven parent–child pairs who had files that fit into in at least three age groups and used these pairs for additional, longitudinal analyses. Of these, six children were from middle- to upper-middle class families and one was workingclass.
Searches for this and the following analyses were conducted using the CLAN program (MacWhinney, 2000). All utterances of say and its inflected forms (says, saying, said) by the parents and child were included. Mother and father utterances were collapsed as parent utterances. Utterances from researchers, other children, or unspecified speakers were excluded from this and the following analyses, as were exact repetitions of previous utterances.
Each utterance of say, by the parent or child, was analyzed to determine whether it referenced spoken or written language. For example, Daddy says I can have some was coded as spoken because the word say here refers to the father’s spoken language. What does that letter say?was coded as written because say here refers to writing. In this and the following analyses, the searches included nine lines of text both before and after the relevant utterance to aid in interpretation of its meaning. If the use of say was unclear from this context, the item was coded as ambiguous. For example, Wendy at age 2;0 uttered I be say, which was never clarified in context, and so the utterance was coded as ambiguous (Warren-Leubecker & Bohannon, 1984).
For child utterances of say in reference to spoken and written language, we further asked whether the child or the parent initiated the utterance. For example, if a mother asked What does the book say? and her child replied It says penguins, then the utterance was coded as cued because the child’s use of say was arguably prompted by the mother’s initial usage. If the child him- or herself, without prompting, said something like The card says my name, then the utterance was coded as spontaneous. The parent utterance did not have to be a direct question in order to be considered a possible cue. For example, Daddy says you cannot go outside followed by Why is Daddy saying that?was considered cued. We coded a child’s utterance as cued if the prompt occurred within the three prior lines of text. Sometimes a parent and child used say back and forth several times in one conversation. In these instances, even when the adult utterance was in the right position to be considered a prompt for the child’s speech, it was only recorded as such if the child did not have a previous utterance in the conversation that was coded as spontaneous.
In this and all subsequent analyses, a second researcher performed duplicate searches of either five or six different corpora to test how reliably the appropriate utterances were being found. The searches by the two researchers always found the same utterances, yielding a search agreement of 100% in this and all subsequent analyses. The second researcher also coded the utterances found during this search, which ranged across analyses from 8% to 31% of the total utterances. For Analysis 1, the two coders agreed 96% of the time on the coding of spoken, written, or ambiguous. Agreement on judgments of cued versus spontaneous utterances was 95%. Where there was disagreement, the codings used in the analyses were taken from the first and more experienced researcher.
Table 1 shows the data on utterances of say referring to written and spoken language by parents and children. Parents occasionally used say for written language from the earliest age analyzed. For example, one mother was reading a book to her daughter when she was 1;7 and explained It says drink your milk!(Bernstein-Ratner, 1984). Children in the 1;6–2;0 age group never used say in reference to written language. The right column of Table 1 shows the proportion of utterances of say referring to written language out of the total utterances in reference to both written and spoken language. Although this proportion was slightly higher for parents than children, the difference was not significant by a chi-square test pooling across age groups, χ2 (1, N = 1931) = 1.13, p> .30.
Across the seven age groups, as Table 1 shows, the children always had more spontaneous than cued utterances of say for spoken language. However, there were some changes in the proportion of spontaneous use of say for written language across the age groups. There were no utterances of say for written language in the 1;6–2;0 age range, as mentioned above. In the 2;0–2;6 age range, children had more cued than spontaneous utterances of say for written language. The proportion of cued utterances was relatively high in the 2;6–3;0 age range as well. Chi square tests revealed that the proportion of cued utterances of say was significantly higher for written language than for oral language in these two age ranges, χ2 (1, N = 493) = 52.38, p< .001, and χ2 (1, N = 884) = 35.44, p< .001, respectively. In the remaining age ranges, there were more spontaneous than cued utterances and the differences between written and oral language in the proportion of cued utterances were not significant by chi square tests. Thus, children between 2;0 and 3;0 sometimes responded with an utterance of say for written language, as when asked Can you read what this says? However, they did not often initiate such utterances on their own until after age 3;0.
For the seven children selected for longitudinal analysis, we asked whether the parents used say for written language before the children did. The first column of data in Table 2 shows these children’s ages at the first use of say for written language by parents and children. This was calculated as the child’s age in months when the utterance was recorded. For children, the relevant utterance had to be spontaneous. The parents’ first use of say for written language occurred on average when the children were 2;5, and a Wilcoxon signed ranks test revealed that the parents used say earlier than the children did, T (6) = 28, p < .01.
The results of Analysis 1 show that parents talk about written words with children even when the children are quite young, interpreting writing for the children in terms of what print says. Parent use of say for written language as well as spoken language may provide a clue about the glottographic nature of writing. By 3;0–3;6, children sometimes use say on their own to talk about writing, even though theyprobably could read or write few if any words at this age. This result is striking given that previous research has shown that young children have difficulty grasping the meaning of communicative verbs such as say (Papafragou, Cassidy & Gleitman, 2007). Although parent use of say for written language provides information as to the glottographic nature of writing, parents used say much less often for written language than for spoken language throughout the age range studied. Given this, Analysis 2 investigated other words that parents may use more frequently in discussions of writing.
Parents sometimes talk to their children about words. Word can refer to any independent unit of language. If parents focus on words as written objects, this could imply that all sorts of things can be conveyed in writing, including actions, people, places, ideas, and even grammatical words. If parents instead focus on names, this might convey the idea that writing is a symbol system used mainly for labeling people. In Analysis 2, therefore, we compared use of word and name for written language. Although word is more general than name, children’s own names are very important to them (Justice, Pence, Bowles, & Wiggins, 2006; Mandel, Jusczyk, & Pisoni, 1995; Treiman, Cohen, Mulqueeny, Kessler, & Schechtman, 2007). Many children become proficient at writing their own names before they can write other words(Levin, Both-De Vries, Aram, & Bus, 2005). Therefore, parents and children might devote much of their discussion of writing to names, implying that written language is primarily used to symbolize names.
Analysis 2 was conducted using the same age groups as Analysis 1.
All utterances of word and name by the target child and his or her parents were initially examined. The three coding levels used in Analysis 1 (spoken, written, and ambiguous) were applied to this analysis; utterances of word and name coded as written were selected for further analysis. For example, Do you know how to write your name?was coded as a written utterance because it refers to the printed form of the child’s name. For children’s utterances of word and name, we distinguished cued and spontaneous utterances as in Analysis 1. A second researcher searched for and coded utterances of word and name as described earlier. The two coders agreed on 95% of their judgments, including the spoken, written, or ambiguous judgment and the cued or spontaneous judgment.
Table 3 shows data on use of name and word for written language by parents and children. Both parents and children used name for written language more often than word. However, differences in parent’s and children’s proportional use of name occurred after age 3;0. Prior to this age, the name/word ratios for parents and children were statistically indistinguishable, χ2 (1, N = 483) = 3.17, p > .10. After 3;0, children favored name over word for written language more strongly than their parents did, χ2 (1, N = 543) = 36.45, p< .001. This result suggests that, starting around their third birthday, children become especially preoccupied with the written representation of names. They focus on this type of writing even when their parents are talking more about the written representation of words in general.
The proportion of children’s cued uses of word and name for written language is also shown in Table 3. The ratio of cued to spontaneous utterances was not significantly different for word and name across the age ranges, χ2 (1, N = 439) = .50, p> .10. Collapsing over word and name utterances, the ratio of cued to spontaneous utterances was higher for children of 1;6–2;0 than for older children, χ2 (1, N = 439) = 43.84, p< .001. That is, many of young children’s uses of name and word to refer to written language are prompted by parents.
Data from the same seven children identified for longitudinal study in Analysis 1 were used to look at the use of word and name longitudinally. We asked whether parents used name and word for written language before their children did, and also whether parents and children tended to use these words in a particular order. The right columns of Table 2 show the children’s ages at the first use of word and name for written language by parents and children. This was calculated as in Analysis 1. When the target word was never used in the files, the first use was recorded as one month after the final file for that child. Pooling over word and name, a Wilcoxon signed ranks test indicated that parents used these words to refer to written language significantly earlier than their children did, T (10) = 97, p< .001. Also, name was used significantly earlier than word, pooling over parents and children, T (11) = 57, p< .05. The finding that name is used first, by both parents and children, suggests that written names are salient for both.
Parents’ use of word and name to refer to both spoken and written language may provide information about the relation between speech and print. Specifically, the fact that these words are used for both systems may be a clue that writing is quoted speech. However, the emphasis on names in discussions of writing may imply that writing is primarily a system for symbolizing names. Both parents and children emphasized names, and children did so especially after their third birthday. Children’s attentiveness to names may attract them to print, while at the same time delaying their understanding of the deeper properties of written language.
In this and the following analyses, we examine whether parents differentiate between writing and drawing in their speech and whether children do the same. People may use different words in discussions of the two systems, suggesting that the systems are distinct. Research mentioned earlier has indicated that syntactic distinctions in adult language provide information about differences among semantic categories. For example, the distinction between count and mass nouns is marked in English by the presence or absence of the indefinite article a or an; count nouns such as table are often preceded by an indefinite article whereas mass nouns like water generally are not. Three- and four-year-oldchildren appear to attend to this syntactic clue with nonwords (e.g., This is a zav versus This is zav) and use it to decide whether a nonword refers to a discrete object or an uncountable substance (Hall et al., 2001).
A similar pattern occurs with quoted forms. Normally, quoted speech is presented without any determiner. For example, one might report She said it is raining, but not She said a it is raining. Draw is like most verbs in English, in that its argument is not a quoted form. Thus determiners are permitted and, in the case of singular count nouns, required. For example, we say Draw a kitten, but not normally Draw kitten. The verbs write and spell, by contrast, require quoted forms as their arguments. One says Write kitten and Spell kitten but not usually Write a kitten or Spell the kitten. This syntactic difference, if present in speech to young children, would indicate not only that writing and drawing are different but also that references to writing lack a determiner because they are quoting speech. We asked whether parents make such a syntactic distinction between writing and drawing and whether children of various ages do so.
Files were sorted into three year-long age groupings (2;0–3;0, 3;0–4;0, 4;0–5;0) because relatively few utterances met the criteria for inclusion. The 1;6–2;0 range was excluded because there was only one relevant utterance in this range.
All utterances of draw, write, and spell, including forms such as drawed, writing, and spells, by the child and his or her parents were included in the analysis. For present purposes, only uses of draw, write, and spell with singular count nouns were considered. Utterances of write and spell were collapsed for comparison to utterances of draw. With these nouns, a judgment was made about the presence or absence of a determiner. Thus, four coding options were possible: write/spell without determiner, write/spell with determiner, draw without determiner, and draw with determiner. A second coder analyzed a portion of the utterances, following the procedures mentioned earlier. The percentage of agreement for assigning utterances to the four categories was 98%.
Table 4 shows data on parent and child use of determiners with the target verbs. We were interested in whether parents and children used draw, write and spell appropriately—that is, with no determiner for write and spell and determiner for draw—across the age range studied. In categorizing utterances this way, we assumed that someone who says write a cat means for cat to be written, not a cat. The proportion of utterances with appropriate determiner use was calculated for both parents and children. Parents almost always used determiners appropriately with draw, write, and spell across the three age ranges. Parents always used a determiner for singular, concrete objects with draw. For example, Adam’s mother (Brown, 1973) asked him at age 2;3, Are you drawing a kitty? and at age 4;3 Why don’t you draw me a whale? Parents’ use of determiners with write and spell was appropriate 99% of the time. There was only one parent error for write when Adam’s mother said, Why don’t you write a tomato for me?when Adam was aged 2;7. There were no significant differences in parent’s appropriate use of determiners with draw and with write and spell. Parents’ clear differentiation between write and draw with children between two and three years of age is notable given that children of this age make few if any distinctions between writing and drawing in their own productions (Levin & Bus, 2003) and given that, as we show below, children of this age do not make the appropriate distinction in determiner use in their own speech.
For children, we first asked whether they used determiners at different rates for draw than for write and spell. Between ages two and three, children used determiners for some of their draw, write, and spell utterances (draw: 59%, write and spell: 73%). However, children used determiners significantly more with write and spell than with draw, χ2 (1, N = 266) = 4.27, p< .05, the opposite of the conventional pattern. At this young age, children seem to know that determiners are sometimes needed in these contexts but are unclear as to when. The lack of appropriate differentiation between write/spell and draw shows how children confuse the two systems at this age. Between three and four years of age there were few analyzable utterances. Children aged 4;0 to 5;0 used determiners 90% of the time for draw, as compared to only 21% for write and spell. The difference was significant, χ2 (1, N = 74) = 36.13, p< .001, and its direction accords with the conventional pattern. Thus, children between four and five do not use a determiner very often with write and spell, following a convention about determiner use with these verbs and suggesting some ability to differentiate between writing and drawing.
We further asked whether children’s ability to use determiners appropriately differed for draw and for write and spell. To address this question, we did a chi square test comparing the appropriate use of determiners with draw and with write and spell across the entire age range. The rate of appropriate determiner use was higher for draw than for write and spell, χ2 (1, N = 361) = 16.22, p< .001. Thus, even as children improved in their use of determiners, they made more mistakes with write and spell than with draw. Even four to five year olds, who had settled on mostly appropriate determiner usage, made determiner errors with write and spell around 20% of the time. For example, Sarah at age 4;6 (Brown, 1973) said Write a carriage. Children’s higher error rate with write and spell than with draw suggests that they continue to have some difficulty understanding the nature of writing, sometimes conceiving of it as the drawing of specific two-dimension shapes (Levin & Bus, 2003).
People may differentiate writing and drawing with the verbs and direct objects used in talk about each system. We explore this possibility in Analysis 4 through an examination of the verbs that parents and children used with the common direct objects of each symbol system. Certain direct objects, such as picture and drawing, are only appropriately used with draw; other direct objects, such as word, name, and letter, are only appropriately used with write or spell. For example, it is appropriate to say Draw a picture and Write a word but it is very odd to say the reverse. Certain neutral verbs, like do and make, may be used for both symbolic systems. For example, a child could say Do a drawing or I’m making my letters. The purpose of Analysis 4 was to investigate whether parents and children used write-only verbs with the common objects of writing and draw-only verbs with the common objects of drawing. We were also interested in the use of neutral production verbs with the common objects of both systems.
Files were sorted into three, year-long age groupings (2;0–3;0, 3;0–4;0, 4;0–5;0) for the same reasons as in Analysis 3.
All utterances that included name, letter, word, sentence, picture, or drawing were considered. Only those utterances that used one of the target words as a direct object with one of the target production verbs were included. That is, the utterance had to refer to actively producing a written word, picture, etc. The target verbs were divided into three categories. Write-only verbs were write and spell. A child might use one of these to say such things as Spell my name. Draw-only verbs were draw, color, and paint. For example, a child could say I am coloring a picture. Two neutralverbs, make and do, were also included because children can use them to refer to objects in either system. The two coders agreed on 96% of their judgments.
Table 5 shows information about the verbs that parents and children used with the target direct objects. Parents never used a write-only verb with a drawing object. That is, no parent utterances like Spell a drawing or Write a picture were found. However, parents sometimes used draw-only verbs with writing objects throughout the age range analyzed. For example, parents occasionally said Draw your letters or Color your name. Children sometimes used write-only verbs with drawing objects and draw-only verbs with writing objects, saying things like I am writing a picture and I am drawing my name. All utterances like the former, with write-only verbs and drawing objects, came from children. This difference between parent and child use of write-only verbs was statistically significant, χ2 (1, N = 216) = 14.26, p< .001. Additionally, children used reliably more draw-only verbs with writing objects than parents did, χ2 (1, N = 325) = 5.10, p< .05. In other words, children produced utterances like draw my name more often than parents. These results suggest that children have some difficulty using the verbs appropriately even when parents provide them with appropriate examples. Both parents and children were more likely to use draw inappropriately than write and spell inappropriately, χ2 (1, N = 541) = 16.99, p< .001, collapsing across parents and children.
Collectively, these results suggest that parents are strict in their talk about writing: They never suggest to children that writing can be the same as drawing. However, parents speak about drawing with some looseness, occasionally saying things like Draw your name. In these instances, parent speech may be thought to imply that writing is a type of drawing that involves sequences of shapes rather than a distinct system for symbolizing language. Still, parent statements like draw your name were much less common than statements like write your name. This finding is noteworthy considering that the written names and letters produced by children at the younger end of the studied age range tend to be unrecognizable scribbles that look quite similar to the scribbles they produce when drawing (Levin & Bus, 2003).
Both parents and children used neutral verbs with writing and drawing objects throughout. A chi square test revealed that children used more neutral verbs than draw-only or write-only verbs when talking about drawing objects, whereas the reverse pattern held for parents, χ2 (1, N = 442), p< .01. There was no significant difference between parent and child use of neutral verbs versus specific verbs for writing, χ2 (1, N = 407) = .45, p> .10. Both parents and children used specific verbs more than neutral verbs when talking about writing, and neutral verbs use was significantly more common with draw-only objects (e.g., make a picture) than with write-only objects (e.g., do my letters), χ2 (1, N = 849) = 88.79, p< .001. Despite these differences between writing and drawing, the fact that parents sometimes used neutral verbs for both writing and drawing may allow children to think of the two symbol systems as similar. This generic usage persisted across the age range studied, and it could potentially contribute to children’s difficulty in grasping the distinction between writing and drawing.
The verbs that parents use with the common objects of writing and drawing provide information about the distinction between the two systems. However, parents occasionally speak of writing as a form of drawing, and this may provide less straightforward information about the nature of print. Analysis 5 further investigated parent speech about writing and drawing by exploring the direct objects used with the common verbs of both systems.
People may differentiate writing and spelling from drawing by using different types of direct objects with the respective verbs. Although many words and phrases can be the object of all three verbs, words such as picture can only be used with draw. Other words, such as word, name, and letter can only be used with write. We asked whether parents made a distinction between the direct objects used with the verbs draw, write, and spell and whether children of various ages made this distinction in their own speech.
Files were sorted into three, year-long age groupings (2;0–3;0, 3;0–4;0, 4;0–5;0) for the same reasons as in Analyses 3 and 4.
The target words for Analysis 5 were forms of draw, write, and spell. Every utterance of draw, write, and spell that featured a direct object (e.g., I can draw a kitty, I want to write cat) was included. Color and paint were not used in this analysis, unlike in Analysis 4, because there were few relevant utterances. The corresponding direct object was classified into one of three categories. If the object was one that could be used with either symbol system (e.g., this, that, kitty), then it was coded as both. If the object was one that it would only be appropriate to draw (e.g., pictures), then it was coded as a draw-only object. If the object was one that it would only be appropriate to write (e.g., letter, name, word), then it was coded as a write-only object. The percentage of agreement between the two coders was 97%.
Table 6 shows the data on parent and child use of draw, write, and spell with direct objects. Parents had no inappropriate uses of write or spell. That is, a parent never said something like Can you write a picture? However, parents did use draw inappropriately on a few occasions. For example, parents occasionally said Can you draw your name? These inappropriate utterances were found throughout the age range studied. The proportion of appropriate utterances by parents was significantly higher for write and spell than for draw, χ2 (1, N = 470) = 14.30, p< .001. Children made some inappropriate utterances in both directions, using draw for write-only direct objects and write for draw-only direct objects. These errors were found throughout the period studied. Children had a significantly lower proportion of appropriate utterances of draw than for write or spell, χ2 (1, N = 364) = 35.36, p< .001. They more frequently said something like Draw my letters than Write a picture. Thus, both parents and children tend to use draw more loosely than write and spell.
Although there was a significant difference in the appropriate use of draw versus write and spell, pooling over all the verbs children did show a significant increase in the proportion of appropriate use over time, χ2 (1, N = 362) = 15.32, p< .001. Children’s appropriate utterances of draw, write, and spell went from 75% between 2;0 and 3;0 to 91% between 4;0 and 5;0. There was no significant increase in the proportion of appropriate use for parents across the three age ranges, χ2 (1, N = 449) = 1.45, p> .10. Pooling over draw, write, and spell, parent use was between 97% and 99% appropriate throughout the age range studied.
Although children displayed some confusion as to the objects that can be symbolized by writing and drawing, there were some signs that older children and their parents began talking about certain types of objects that are only properly symbolized by writing. There are two examples of this. First, there were five instances for parents and seven instances for children when nonpicturable function words were used as the objects of write or spell. These words were at, is, no, not, of, and with. All use of write and spell with these objects, for both parents and children, occurred after child age of 3;6. Although there were few examples of this sort with write and spell, there were no occurrences with draw. That is, neither parents nor children ever said things like draw at or draw with. Second, there were nine instances where parents and children used a stretch of quoted speech as the direct object of writing. For example, at age 4;1 Abe (Kucjaz, 1976) was making a card and instructed his mother to write from Abe to Rob and Rich and Mike. Phrases words never occurred as the object of draw; Parents and children only used lines of quoted speech as direct objects of write or spell. Together, these examples suggest that some parents give children at least occasional information that writing is a system for symbolizing speech by the end of the children’s preschool years.
Names constituted a large proportion of the direct objects that parents and children used with draw, write, and spell (161 utterances of name and specific personal names for draw, 368 utterances of name and personal names for write and spell). Children often used a draw-only verb with name, as in Draw my name or Draw your name. Although children made such errors with several pronouns (your, mine, their), adults only used draw with name when they were saying to the child Draw your name. They did speak about writing names other than the child’s, but they only used draw when referencing the child’s own name. Thus, parent use of draw with name as a direct object seems to be limited to discussion of the child’s name and is not a way that parents discuss names in general. Parent utterances such as draw your name may mislead children, in that they simultaneously invite children to see writing as pictorial representation and as a system specialized for symbolizing names.
In the final analysis, we examined the feedback that parents give their children concerning the child’s ability to write and draw. Parents rarely provide young children with explicit feedback regarding the grammaticality of their spoken utterances (Brown & Hanlon, 1970). What feedback, if any, do parents provide about children’s ability to write and their ability to draw? If parents are more positive about children’s drawing abilities than about their writing abilities, this may convey the idea that writing is a difficult and complex system that requires special abilities.
Files were grouped into six-month periods across the 1;6–5;0 age span.
Analysis 6 included all utterances of write, draw, and spell. Write and spell were collapsed for comparison to draw. We asked what information the utterances contained about the child’s ability to draw and write. Two distinctions were coded. First, a statement could accept or reject a person’s ability to draw or write. For example, That’s a pretty drawing was considered an acceptance of a person’s ability to draw, whereas You cannot write was a rejection of the person’s writing ability. Second, the subject of the ability judgment was categorized as the child, an adult (parents, investigator, or other), an animate but inappropriate reference (e.g., a dog, cat, or cartoon character), or an unspecified or ambiguous other. The percentage of agreement between the two coders was 98%.
There were 6253 utterances analyzed for judgments of ability, 2815 from children and 3438 from adults. Across all verbs, ages, and speakers, there were only seven statements of rejection, all of which were for write and spell. Rejections for spell all involved individual words. For example, Nina’s mother (Suppes, 1974) informed her You don’t know how to spell your name yet when Nina was 3;1. Even when a parent rejected the child’s ability to write, the statement was sometimes misleading. For example, Adam’s mother (Brown, 1973) informed him at age 2;3 that he couldn’t write by saying You can’t write and by telling the investigator He [other child] can write, but Adam can’t yet. A little later in the same conversation, though, she asked Adam What are you writing? Thus, parent speech does not seem to provide much direct information regarding differences in their child’s ability to draw and write. Instead of criticizing children’s abilities, parents encourage children to be interested in writing and drawing.
In a few instances, children attributed the ability to write and draw to objects that cannot have these abilities. This occurred 11 times across age range studied. Adam (Brown, 1973) credited tractors and dogs with the ability to write at age 2;3 and Sarah (Brown, 1973) said at age 3;3 that Bugs Bunny can write. Parents occasionally participated in this way of speaking. There were four such utterances, all in response to child utterances. For example, Adam said at age 2;3Doggie writing and his mother replied What did the doggie write? Thus, in trying to interest children in writing, parents may sometimes speak about writing in loose ways.
The results of Analysis 6 suggest that parents in the U.S. and the U.K. give their children little or no direct feedback on their ability to write and draw, just as they give children little or no direct feedback on their spoken syntax (Brown & Hanlon, 1970). Instead, parents encourage their children to engage in the activity of writing, regardless of the children’s ability. By encouraging children to write, parents may foster the attention to print that is necessary for children to understand the nature of writing.
In order to master reading and writing, children must grasp that written language symbolizes the sounds of spoken language. Children receive some information about the nature of written language before they enter school through such activities as being helped to print words. In the present study, we investigated an additional way in which children may learn about writing. We found that parents speak to their children about written language beginning when the children are quite young, and that they do so in ways that provide information about the glottographic nature of written language. Parent speech provides helpful information that writing is quoted speech by invoking similarities between speaking and writing. Parents used say, word, and name for both spoken and written language. Allowing the same linguistic items to feature in speech and print provides evidence of the glottographic nature of written language. Additionally, parents indicate that writing is different from drawing by making distinctions in their discussions of the two systems. With singular count nouns, parents almost always use a determiner with draw and almost never with write or spell. These results suggest that parent speech provides children with information about the nature of written language before the children enter school.
However, some parent speech may be misleading as to the glottographic nature of print. When discussing writing with their young children, parents talk about names more often than words. The emphasis on names may hinder a full understanding of the relation between speech and print. Children need to understand that anything that can be spoken can be written. Some children appear to understand this by the age of five, saying things such as write of, but the emphasis on names might make this realization more difficult.Further, parents sometimes speak of writing as if it were a form of drawing, suggesting that children draw their names. Speaking of writing in this way may make writing and drawing seem less different than they are. We also found that parents support young children’s attempts at writing, rarely saying that children are unable to write. The lack of negative feedback may allow children to believe that it is appropriate to understand writing as a form of drawing.
There are, of course, reasons why parents speak this way. Parents have an incentive to focus on the aspects of writing that are most interesting to children, such as the children’s own names. Linking writing to drawing may also encourage children to engage in the process of learning to write, as drawing is an enjoyable activity for most young children. Wanting to interest children in writing, parents have little motivation to criticize children’s abilities. The technical difficulties in learning to write may also influence parents’ speech. Forming the small and detailed symbols of writing is hard for young children, and so there is a focus on the surface, mechanical aspects of writing. This may make writing seem more like drawing than it actually is. Viewing writing as a mechanical process may be helpful when learning to form specific letters. However, it may delay children’s understanding of the deeper nature of writing. Children may think that the ability to write is purely a mechanical achievement and that the stumbling block they face is learning to make the symbols, not learning the deeper nature of the system.
Collectively, our results suggest that, as in many other domains (essentialism: Gelman, 2003; category membership: Waxman & Markow, 1995; count/mass noun distinction: Hall et al., 2001), parent speech provides information to young children about the nature of the world around them. A good deal of information about written language is available to children in literate societies prior to their attending school and receiving formal instruction. As previous studies show, young children can learn about some of the graphic features of writing (Ganapole, 1987; Treiman et al., 2007) the alphabet (Evans, Shaw, & Bell, 2000), and even specific words (Aram, 2006; Aram & Levin, 2004) through such experiences. The current findings add that children receive information about the symbolic function of writing through conversations with their parents. Implicit information in parent speech indicates that writing works differently than drawing, and even that it symbolizes speech, even before children are explicitly taught how it is that writing does so. Whether children are able to use this information, and whether they do in ways that influence their later literacy skills, are not issues we could address in the present study. Even if children do recognize and use this information, it would not suffice for comprehending the nature of print. A full understanding of how writing works, of course, requires children to grasp exactly how writing goes about symbolizing spoken language. This further understanding may not happen until children receive formal instruction in school. Thus, although some aspects of literacy learning may require explicit instruction, children receive information about how writing works from informal interactions with their parents.
Our results also shed light on the ways in which young children understand writing. Their talk about writing changes across the preschool period, suggesting a respective set of changes in how they understand the glottographic nature of print. Previous studies have assessed children’s understanding of writing by asking whether they to produce writings that are distinguishable from drawings and whether they explicitly recognize the difference between words and pictures. Although success at these tasks indicates knowledge about print, failure does not necessarily imply a lack of knowledge. The nature of our study allowed us to examine very young children who cannot easily participate in most production and recognition tasks. We found that children make some distinctions between writing and drawing from a young age but that the process of distinguishing the two systems is not complete even as children begin to do things such as produce their written names. This is shown, for example, by 4 and 5 year olds’ statements such as draw my name.
Our results suggest that studies which focus on features of homes and families such as number of books in the home or time spent reading stories or teaching about letters have missed much of the richness of literacy-related activities that occur in homes. In particular, such studies have missed many of the conversations that mention writing. Our results show that such conversations occur sporadically across a host of contexts beginning when children are quite young and that they may provide children with useful information. In future research, it will be important to determine whether children take advantage of this information by examining early parent–child conversations about writing in relation to children’s later literacy skills. This was not possible in the current study, but our work sets the stage for future work by providing information about the types of parent speech that may be important.
Our use of the CHILDES data, while allowing us insight into literacy-related discussions that occur in homes, had certain other limitations. Although we had a large sample, there were certain age ranges for which only small amounts of data were available. The longitudinal analyses that we included mitigated the effects of this limitation to some extent, but not totally. Further, our analyses were confined to English. It would be valuable to examine data from other languages and cultures to provide a more complete understanding of language-specific and language-universal cues. It is also important to examine parent–child conversations about writing as a function of social class and parent education. The CHILDES sample is skewed toward middle- and upper-middle-class families, as mentioned earlier, but differences as a function of social class and education would be anticipated on the basis of previous findings (e.g., Hart & Risley, 1995). In the present study, we did not have enough data to allow for separate analyses of different subgroups. In other ways, however, the variability among the corpora—in date collected, context, and family background— were advantageous for our exploratory purposes, allowing us tosurvey the extent to which young children and their parents discuss written language across a broad range of conditions.
Although much work remains to be done, our results show that children are exposed to information about certain aspects of writing in conversations with their parents even before formal instruction in reading and writing begins and even before they produce anything resembling conventional writing. Parent speech indicates similarities between spoken and written language, as well as differences between writing and drawing. The present findings thus offer insight into the environmental input that may help children develop an understanding of the nature of written language. The findings also suggest that children’s speech about writing offers insight into their understanding of writing’s nature, supplementing the information about children’s understanding of writing that is available in other ways.
We thank Elizabeth Schotter, Suzanne Schechtman, and Lindsey Clasen, who performed the reliability coding, and the members of the Reading and Language Lab at Washington University in St. Louis for helpful comments on an earlier draft. Research for this paper was supported in part by NIH Grant HD-051610.
1CHILDES corpora included in analyses: Bates, Belfast, Bernstein-Ratner, Bliss, Bloom 1970, Bloom 1973, Bohannon, Brown, Clark, Cornell, Cruttenden, Demetras Working, Feldman, Fletcher, Gleason, Haggerty, Hall, Higginson, Howe, Kucjaz, Manchester, Morisset, Nelson, New England, Post, Sachs, Snow, Suppes, Tardif, Valian, Van Houten, Van Kleeck, Warren-Leubecker, Wells.
Sarah Robins, Department of Philosophy, Washington University in St. Louis.
Rebecca Treiman, Department of Psychology, Washington University in St. Louis.