Letter Teaching in Parent–Child Conversations

Abstract

Talk about letters is an important part of the home literacy environment. Such talk has been studied primarily through questionnaires, but these are limited in the amount of information they provide. Here we analyzed conversations between 55 U.S. children and their parents who were visited in their homes every 4 months when the child was between 1.2 and 4.8 years old. We examined the aspects of alphabet knowledge that parents and children discussed, the materials they used, and how these varied with the age of the child and the socioeconomic status of the family. Children primarily focused on identifying letters, while parents also emphasized letter writing and spelling. Talk about the associations between letters and sounds, which is critical in learning to read and write, was less common than anticipated based on the results of questionnaire studies. Teachers should thus not overestimate the knowledge of letter sounds that children acquire at home.

Learning to read and write within an alphabetic writing system requires learning a variety of precursor skills. These include oral language skills, phonological awareness, and—the ability on which we focus here—knowledge about the letters of the alphabet (National Early Literacy Panel [NELP], 2008). Given that alphabet knowledge in preschool and kindergarten is one of the strongest predictors of later literacy performance (e.g., Lonigan, Burgess, & Anthony, 2000; Storch & Whitehurst, 2002), educators typically provide children with systematic instruction about letters. Children do not come to such instruction as blank slates, though, for they typically have opportunities to learn about the alphabet at home. Indeed, Bronfenbrenner’s bioecological systems theory (e.g., Bronfenbrenner & Morris, 2006) emphasizes the importance of the frequent and extended interactions between children and their parents, their earliest teachers. Similarly, Vygotsky’s theory of development, which has played a major role in early childhood education (e.g., Berk & Winsler, 1995, Vygotsky, 1978), sees interacting with more knowledgeable others as critical to a child’s development. To help inform instruction, it is important to understand the opportunities for alphabet learning that parents provide at home. To do this, the present study analyzed everyday conversations between U.S. parents and their young children that touched on letters of the alphabet. We used longitudinal data from 55 families who were visited every 4 months when their children were between 1.2 and 4.8 years, allowing us to investigate changes across the early years.

Components of Alphabet Knowledge

Alphabet knowledge is not a unitary skill (e.g., Piasta & Wagner, 2010). Children must learn about several aspects of letters in order to read and write, and conversations may focus on different aspects. By studying the aspects of alphabet knowledge that parents and children discuss, we can get a better understanding of what parents do and do not teach children about the alphabet and what children are most interested in learning.

Our study focused on four aspects of letters. The first is that letters represent sounds. A father who tells his child that S makes the sound /s/ is giving his child an opportunity to learn about this aspect of alphabet knowledge, which we refer to as the sound aspect. Knowledge about the correspondences between letters and sounds is critical for decoding written words and constructing spellings of words that have not been stored in memory (e.g., Caravolas, Hulme, & Snowling, 2001; Ehri, 2015). Another aspect of a letter is its visual appearance. Learning to distinguish among letters’ shapes is important for learning to read and write (e.g., Huang, & Invernizzi, 2014; Treiman & Kessler, 2014). Talk about identifying letters in the environment, as when a parent points out an S in a storybook, can promote this aspect of alphabet knowledge, which we call identification. The third aspect of letter knowledge studied here is that letters are arranged in sequences to form words. A mother who tells her child that sun is spelled with S, U, and N is focusing on this aspect of letter knowledge. Although this statement does not explain that the use of these specific letters is motivated by the sounds in the spoken word, it indicates that words are composed of letters and provides information about the letters in a specific word. We refer to this aspect of letter knowledge as spelling. The final aspect of alphabet knowledge studied here is production: the construction of letter shapes. A parent may promote this aspect of alphabet knowledge by helping a child write a letter, and a child may show an interest in this aspect by asking about how to do so. Production is important because the ability to write letters accurately and without much effort is important for spelling and composition, and practice in letter production appears to help children with letter identification (Alves et al., 2016; Longcamp, Zerbato-Poudou, & Velay, 2005). In the present study, we analyzed talk about each of these aspects of letter knowledge by both parents and children.

Questionnaire Studies of Parent Alphabet Teaching

Researchers have often assessed children’s opportunities to learn about letters at home by asking parents how often they teach their children about the alphabet. For example, parents may be asked whether they engage in such teaching every day or a few times a week, once a week, less than once a week, or never (e.g., Chen, Pisani, White, & Soroui, 2012). A weakness of such studies is that they do not distinguish among different aspects of letter knowledge. Some questionnaire studies have attempted to do this by asking parents about the specific aspects of letters that they (or other family members) discuss with their children. In the studies of Haney and Hill (2004) and Martini and Sénéchal (2012), which were conducted with U.S. parents of 3-to 5-year-olds and Canadian parents of 5-year-olds, respectively, at least 70% of parents reported that they taught children to identify letters by name. Letter production also appeared to be a point of emphasis, with 45% of the U.S. parents and 72% of the Canadian parents reporting teaching about letter printing. Teaching of letter sounds was reported by 65% of the U.S. and 79% of the Canadian parents, suggesting that the majority of North American children have the opportunity to learn about the sound aspect of letters at home. The results of questionnaire studies also reveal some talk about words’ spellings, at least the spelling of the child’s name. For example, North American parents report that they help children write their names or that their children do so on their own (e.g., Martini & Sénéchal, 2012; Puranik, Phillips, Lonigan, & Gibson, 2018).

A concern about questionnaire studies is that it can be difficult for parents to respond accurately to detailed questions about the aspects of letters that they discuss with their children. Moreover, because literacy has been the focus of a number of well-publicized policy initiatives, parents might exaggerate how often they teach their children about letters and other literacy-related matters in an attempt to portray themselves as good parents (e.g., Sénéchal, LeFevre, Hudson, & Lawson, 1996; see van de Mortel, 2008, for a general discussion of social desirability biases). Questionnaire studies may thus not provide a complete and accurate picture of what parents teach their children about the alphabet before formal literacy instruction begins. And questionnaire studies have not solicited information from young children themselves about their knowledge of and interest in different aspects of letters.

Observational Studies of Parent–Child Conversations About Letters

A different way to study how parents and children discuss letters, and the one we adopt here, is to examine their conversations. Several studies have transcribed and analyzed conversations between parents and children during the course of daily activities, asking how often the conversations touch on letters of the alphabet and what aspects of alphabet knowledge are discussed. A study using a subset of the transcripts analyzed here found that about 3% of everyday talk between U.S. parents and their one- to four-year-old children is part of a conversation related to letters (Treiman, Decker, Robins, Ghosh & Rosales, 2018), with the percentage being higher for four-year-olds than for one-year-olds. Parents and children have a myriad of topics to talk about, and it is interesting that they talked about letters, especially given that the researchers did not ask them to discuss literacy-related matters and did not provide them with materials (such as alphabet books) that might encourage such discussions. Although previous studies did not break down talk about letters into the categories used here, the results suggest that letter identification is a major focus of parents’ and children’s letter-related conversations. For example, parents ask questions that encourage children to name letters that are in view and children ask their parents to provide this information (Robins, Treiman, & Rosales, 2014). References to the spellings of children’s names and other words are also common (Robins, Ghosh, Rosales, & Treiman, 2014; Robins, Treiman, & Rosales, 2014). These results broadly agree with those of questionnaire studies, but transcript-based studies have found less discussion of letters’ sounds than expected based on the results of questionnaire studies. For example, Robins, Treiman, and Rosales (2014) found that parents rarely asked children to provide letters’ sounds and that children rarely requested such information.

Other researchers have observed parents and children when they are asked to perform a literacy-related task, such as writing an invitation to a birthday party (e.g., Bindman, Skibbe, Hindman, Aram, & Morrison, 2014; Skibbe, Bindman, Hindman, Aram, & Morrison, 2013). The parents in these studies often told children the word or the sequence of letters to write (e.g., “write Rachel” or “write R A C H E L”), providing information about the spelling and production aspects of letters. Parents did not usually discuss the connections between the letters and the sounds in the word, a result that might reflect a focus on producing a completed invitation.

Materials Used in Conversations About Letters

According to the bioecological perspective mentioned earlier (Bronfenbrenner & Morris, 2006), an important way in which adults structure a child’s environment is through the provision of learning materials. When teachers provide alphabet instruction, they typically use materials that are designed for this purpose, such as workbooks. In contrast, the Canadian parents surveyed by Martini and Sénéchal (2012) reported that they most often taught their children about letters using storybooks. These parents also reported using environmental print, including letters on street signs and labels on household objects. In the present study, we asked whether observational data show the same patterns.

It is important to examine the materials that parents use because their nature may influence children’s learning. For example, materials that are designed for letter teaching typically present visually salient letters in isolation, as with a block with a red B imprinted on it or a large letter on a page of a workbook. A letter in a storybook, in contrast, is likely to be small, black, and near a picture (Treiman, Rosales, & Kessler, 2016). Studies show that young children spend little time looking at letters when being read to (Evans & Saint-Aubin, 2005) and that parents primarily talk about the narrative and pictures rather than the print (e.g., Hindman, Connor, Jewkes, & Morrison, 2008; Price, Kleeck, & Huberty, 2009). For these reasons, learning about letters from storybooks may not be all that effective.

Differences across Time and Families

Vygotsky’s (1978) theory of development suggests that assisting children during learning activities can promote learning. According to this theory, parents should assess what children can do on their own and what they may be able to accomplish with help and then work within this zone of proximal development. This theory leads us to hypothesize that parents will engage in more complex talk about letters as children get older. For example, parents may increasingly focus on the sound aspect of letters rather than the simpler aspect of identification. As another example, parents may be increasingly likely to provide all of the letters of a word’s spelling rather than just one letter. One study that observed U. S. parents and children in writing tasks found that parents provided more explanations about links between letters and sounds when children were about 5 ½ years old than they were a year younger (Skibbe et al., 2013). In the study of Robins, Treiman, and Rosales (2014), however, parents’ questions to children about letters’ sounds did not increase significantly in frequency across the age range of 1.0 to 5.0 years. Here we tracked parents’ and children’s letter talk between child ages of 1.2 and 4.8, examining possible changes in the aspects of letters that were discussed and the materials that were used.

Studies of parents’ and children’s talk about letters need to consider possible differences not only as a function of the age of the child but also the socio-economic status (SES) of the family. According to the bioecological view of development (Bronfenbrenner & Morris, 2006), this and other aspects of the cultural context can influence the learning environment that families provide. The home literacy environment is widely seen as richer in higher-SES families than lower-SES families, and it is thought that these differences help to explain SES-related differences in the alphabet knowledge that children bring to school and their academic achievement (e.g., Neumann, 2016; Strang & Piasta, 2016). Consistent with this view, questionnaire studies suggest that reading books to children is a more frequent activity in higher-SES than lower-SES homes (e.g., Chen et al., 2012; Kuo, Franke, Regalado, & Halfon, 2004; Schaub, 2015; Yarosz & Barnett, 2001). Surprisingly, however, Chen et al. (2012) found that U.S. parents of different income levels reported similar frequencies of teaching the alphabet to their 2-to-5-year-old children. Similarly, the transcript-based research of Treiman et al. (2018) did not find differences as a function of SES in the overall frequency of letter talk between parents and children. Here we asked whether a finer-grained analysis of letter-related conversations would reveal differences related to SES in the aspects of alphabet knowledge that parents and children discuss.

Also of interest was whether there were SES-related differences in the materials that parents use in teaching their children about letters. Chen et al. (2012) hypothesized that lower-SES parents, owning fewer storybooks, are less likely to use them in letter teaching than higher SES-parents. Instead, Chen et al. hypothesized, lower-SES parents tend to use print on everyday household objects. This hypothesis has not been directly tested. One study found that parents with lower levels of literacy were more likely than parents with higher levels of literacy to state that materials such as flashcards and alphabet blocks were important (Fitzgerald, Spiegel, & Cunningham, 1991), but that study did not examine the materials parents actually used.

The Present Study

We analyzed parent–child conversations that were collected as a part of the Chicago Longitudinal Language Project (Goldin-Meadow et al., 2014). In this project, a diverse group of families in the Chicago area was visited in their homes every 4 months when the target child was between 1.2 and 4.8 years. At each of the 12 home visits, the parent and child were videotaped while carrying out their ordinary daily activities. The conversations were transcribed and supplemented with information about the activities in which the parents and children were engaged and the objects that were present while they spoke. Data from the Chicago Longitudinal Language Project have been used to study many aspects of language development (see Goldin-Meadow et al., 2014), including parent–child conversations about literacy-related topics (Treiman et al., 2015; 2018). An advantage of using these data for the present analyses is that the data collection procedures were consistent across families. Other studies of parent–child conversations about literacy-related matters (Robins, Ghosh, et al., 2014; Robins & Treiman, 2009; Robins, Treiman, et al., 2014; Robins, Treiman, Rosales, & Otake, 2012) have used data from the Child Language Data Exchange System (MacWhinney, 2000), which includes studies that vary in their procedures. Importantly, the parents in the Chicago Longitudinal Language Project did not know that their literacy-related talk would be studied. Researchers did not bring books or other literacy-related materials to families’ homes and did not ask parents to engage in specific activities with their children. Thus, parents had no particular motivation to increase their rate of literacy-related talk. As mentioned previously, social desirability biases are a potential concern in some other types of studies, including questionnaire studies and studies in which parents and children are asked to perform specific literacy-related tasks.

The purpose of the present study was to characterize, more precisely than in previous research, how parents and children discuss letters in the home. Our specific research questions may be summarized as follows:

• What aspects of alphabet knowledge do parents and their 1.2 to 4.8-year-old children discuss, and does this change as children get older? We hypothesized that conversations involving older children would increasingly focus on more advanced aspects of letter knowledge and less on simple identification. We also expected that talk about the sound aspect of letters would not be as common as suggested by the results of questionnaire studies.

• What materials are used in discussions of letters, and does this change as children get older? A previous questionnaire study (Martini & Sénéchal, 2012) reported that parents most often used storybooks to teach about letters. However, given that the results of questionnaire studies do not always agree with those of in-home observations, we hypothesized that there may be some differences here as well.

• Are there variations in the aspects of alphabet knowledge that are discussed and the materials that are used to do so as a function of the SES of the family? Given the paucity of previous work on these topics, we sought to describe the patterns in the data. We also tested a specific hypothesis put forward by Chen et al. (2012): that lower-SES families are more likely to use environmental print and less likely to use storybooks in discussions of letters.

Method

Participants

Families were recruited to the Chicago Longitudinal Language Project via mailings to families living in targeted zip codes and advertisements in a free monthly magazine for parentsInterested parents were interviewed about their backgrounds, and 64 families who were representative of the greater Chicago area in ethnicity and income and who spoke English at home as the primary language were selected. The present study included data from 55 of the families. We excluded data from 3 families in which the children were later diagnosed with a disorder that could have impacted their language or literacy development and data from 6 families that missed more than one of the 12 home visits. Eight families that missed one visit were included. The primary caregiver was the mother in 48 of the 55 families and the father in one; the remaining families were dual caregiver families. The children included 29 boys and 26 girls, 38 of whom were reported to be White, 11 African American, and 6 of two or more races. Six children were reported to be Hispanic.

Information about the education level of the caregivers and the family’s income was collected categorically in a questionnaire that was given at or before the first home visit. Each category for education was assigned a value equivalent to years of education (e.g., 12 for completion of high school). We used the value for the primary caregiver when one parent was the primary caregiver and the average value for the two parents for dual-caregiver families. The categories for family income, which ranged from less than $15,000 to over $100,000 per year, were transformed into a scale by using the midpoints of the incomes in each category except the highest, which was coded as $100,000. According to these scales, the mean number of years of education of the caregivers was 16.29 (SD = 2.94) and the mean family income was $60,500 (SD = 31,998). Education and income were positively correlated (r = .38). As in several previous studies using data from the Chicago Longitudinal Language Project (Goldin-Meadow et al., 2014; Treiman et al., 2015, 2018), we used principal components analysis to combine education and income into a composite measure of SES with a mean of 0 and a standard deviation of 1.0.

Procedure

Transcription of talk during home visits

We analyzed data from home visits that took place every four months, beginning in 2002, from when each child was approximately 1.2 years of age until the child was approximately 4.8 years. At each visit, a research assistant videotaped the parent–child dyad for approximately 90 minutes. Because the goal was to obtain a picture of typical parent–child interactions, the research assistant did not bring toys or books but asked parents to interact with their child as they normally would. The activities in which parents and children engaged varied, but typical sessions included activities such as playing with toys and eating. All caregiver speech to the child and all child speech in the videotaped sessions was transcribed. Caregiver speech to the child’s siblings, if any, was also transcribed and included in our analyses. The unit of transcription was the utterance, which was defined as a sequence of words that was preceded and followed by a pause, a change in a conversational turn, or a change in intonation pattern. For a random 20% of transcripts, a second person transcribed 10% of the utterances. Interrater agreement for the transcription was at or above 95%.

Coding of letter talk

For the present analyses, we searched the transcripts to locate utterances by children and parents that included names of letters. We only coded utterances that included the name of a specific letter because it was impossible to know if utterances such as “That has three lines” referred to a letter. Utterances that used the article a and the pronoun I were not counted as letter name utterances, nor were those in which a letter name was part of a word, such as TV or ABC soup.

We coded a letter name utterance as referencing identification if it allowed the letter to be identified or recognized, either because the letter name was said while the letter was present or because a letter’s appearance was described without the letter present, as in “I has a dot.” Utterances in which a child or parent spelled a word or provided one letter of a word, such as “You wrote B O Y,” or “A is for alligator,” were placed in the spelling category. An utterance in this category did not have to explain why the word was spelled as it was. The letter and the word that it spelled were not required to be in the same utterance. Thus, each utterance in the series of five utterances “C”, “A”, “M”, “E”, and “L” counted in the spelling category. Production utterances referred to letters that a parent or child was writing or described the production of a letter shape, such as a child’s “I wrote this M.” Utterances that mentioned the sound of a letter were coded as referencing sound.

Some utterances referenced multiple aspects of alphabet knowledge, and in these cases each aspect was coded. For example, a parent’s “That’s a K for kuh Kevin” in the presence of a visible K was coded as referencing identification, spelling, and sound. Other letter name utterances, such as “We could play A P B D,” did not reference any of the aspects of interest. Reliability of our coding was assessed by having a second individual code the data from two randomly selected families from each session. Inter-rater agreement was 95% for identification, 98% for production, 100% for sound, and 99% for spelling.

We carried out additional coding of utterances that referenced spelling for which the specific word(s) being spelled could be identified. We coded each word being spelled as the child’s first name, another personal name (other first names, last names, words such as mom or dad), a content word other than a personal name (e.g., cat, bowl), or a function word (e.g., the grammatical word of). We excluded from these analyses the 39 utterances (1% of the spelling utterances that referenced an identifiable word) where the word being spelled did not clearly fit one of the four categories (e.g., yes) and the 180 utterances (6%) where the word being spelled was unknown or where there were multiple possibilities, as with “What is G for?” We also coded whether the reference to spelling involved just the first letter of a multi-letter word, as in “A is for ant.” or more than one letter, as in “N U T spells nut.” (There were no spelling utterances involving single-letter words where the first letter of a word would be the whole word.) Interrater agreement, calculated in the same manner as for aspects of letter knowledge, was 100% for the coding of both word category and first letter or more than one letter.

To examine the materials used in discussing letters, we coded where the letter under discussion was in the environment for utterances that referenced identification. We excluded from these analyses the 172 utterances (3% of the identification utterances) where the information about context that was provided in the transcripts was not sufficient to indicate the material being used or where a speaker did such things as describing an imagined letter that was not present or identifying an apple slice or an arm movement as a letter. The storybook category was used for letters in storybooks, as when a child identified letters on a page of a book. The environmental print category included letters on objects that have real-life functions other than that of literacy instruction, such as letters on a coffee can. The writing category included letters that parents or children were writing on materials that were not explicitly designed to teach about letters, such as letters from a child’s name that a parent was writing for a child on a pad of paper. Letters in the manipulative category were those on puzzles, blocks, cards, magnets, stickers, and the like—materials that were specifically designed to teach about letters. Letters in the electronic teaching material category were those in computer or tablet games intended to teach letters or other literacy skills, and letters in the paper-based teaching material category were those in alphabet books, workbooks, letter coloring sheets, activity sheets, word searches, and crosswords. Although it would have been possible for a letter name utterance to refer to letters in more than one category, this did not occur. Inter-rater agreement for this coding was 95%.

Data analysis procedures

We used negative binomial regression models to analyze the data. Although standard Poisson regression is usually appropriate for count outcome variables, our data do not meet the assumptions of this approach in that the variances exceeded the means. Also, Poisson regression assumes that each occurrence is independent of previous occurrences, which is unlikely to be the case for the letter name utterances studied here. Negative regression models do not make these assumptions of equidispersion and independence and so are appropriate for our data (see Gardner, Mulvey, & Shaw, 1995, and Sturman, 1999 for discussion). We conducted one analysis for each aspect of alphabet knowledge, another analysis for utterances that referenced more than one aspect of alphabet knowledge, and another for utterances that did not reference any of the aspects. For each of these models, the dependent variable was the number of letter name utterances that referenced the aspect of interest. The offset variable, which is the number of times the event could have occurred, was the number of letter name utterances by the speaker in the session. Family number was included as a random factor to characterize variation due to differences across families. All models included the fixed factors speaker (child or parent, coded as 1 and 0, respectively), SES (the composite measure described earlier), the mean-centered linear and quadratic effects of child age (in days), and all possible interactions except those involving interactions between linear age and quadratic age. The analyses were conducted using the package glmmADMB (Skaug, Fournier, Bolker, Magnusson, & Nielsen, 2016) in R version 3.1.3 (R Core Team, 2016). We used a similar approach to examine the coded characteristics of utterances that referenced spelling. Here, the offset variable was the number of letter name utterances in which an identifiable word was spelled. For the analyses dealing with the materials used, the offset variable was the number of letter name utterances that referenced identification where the material could be identified.

The conversation transcripts, coding materials, and data analysis scripts may be found at https://osf.io/g9gs3/?view_only=c176a2ef7c4f4190a15baeddd4b257c9

Results

References to Different Aspects of Letters

Table 1 shows the proportion and number of letter name utterances by parents and by children that referenced each of identification, spelling, production, and sound. This and other tables show the results for four age groups. As mentioned earlier, however, we treated age as a continuous variable in the statistical analyses.

Table 1.

Proportion of Letter Name Utterances That Referenced Identification, Spelling, Production, and Sound as a Function of Child Age (Number of Utterances Referencing Each Aspect Out of Total Number of Letter Name Utterances in Parentheses)

	Identification		Spelling		Production		Sound
Child Age	Parents	Children	Parents	Children	Parents	Children	Parents	Children
1;2 – 1;10	.68 (400/587)	.64 (106/165)	.34 (202/587)	.02 (4/165)	.12 (69/587)	.01 (1/165)	.01 (7/587)	.00 (0/165)
2;2 – 2;10	.82 (864/1049)	.75 (506/675)	.37 (386/1049)	.18 (119/675)	.07 (72/1049)	.04 (30/675)	.02 (23/1049)	.00 (2/675)
3;2 – 3;10	.74 (1035/1394)	.74 (1077/1465)	.46 (639/1394)	.30 (441/1465)	.21 (288/1394)	.08 (117/1465)	.02 (29/1394)	.02 (25/1465)
4;2 – 4;10	.81 (1433/1766)	.76 (1118/1465)	.37 (646/1766)	.29 (421/1465)	.19 (333/1766)	.09 (126/1465)	.04 (63/1766)	.03 (50/1465)
Total	.78 (3732/4796)	.74 (2807/3770)	.39 (1873/4796)	.26 (985/3770)	.16 (763/4796)	.07 (274/3770)	.03 (122/4796)	.02 (77/3770)

Identification was by far the most commonly referenced aspect of alphabet knowledge, as Table 1 shows. The model for identification showed no significant main effect of age or speaker and no significant effects involving SES. However, there were significant interactions between the linear effect of age and speaker (b = 0.20, SE = 0.06, p = .001) and the quadratic effect of age and speaker (b = −0.11, SE = 0.05, p = .032). In this and other cases where there were significant interactions of linear and quadratic age with speaker, we ran separate negative binomial regression models for parents and children with linear and quadratic age as fixed factors. The model for children found significant linear (b = 0.24, SE = 0.05, p < .001) and quadratic effects of age (b = −0.15, SE = 0.04, p < .001). As Table 1 shows, the likelihood that a child’s letter name utterance would involve identification increased as the child got older, but the increase slowed after around age 2. The likelihood that a parent’s letter name utterance would reference identification did not change significantly as children got older.

Spelling was the second most commonly referenced aspect of alphabet knowledge, and utterances that referenced spelling were significantly more common among parents than children (b = −0.39, SE = 0.13, p = .003). There was also a significant main effect of SES (b = 0.20, SE = 0.09, p = .031). A letter name utterance was less likely to reference spelling when the speaker was of lower SES than of higher SES. To illustrate, the proportion of letter name utterances that referenced spelling was .27 (1066/3906) for families that were below the median in SES and .38 (1770/4660) for families that were above the median. Although there was no main effect of age, there were significant interactions between linear age and speaker (b = 0.85, SE = 0.14, p < .001) and quadratic age and speaker (b = −0.42, SE = 0.11, p < .001). A separate model for children found significant linear (b = 0.81, SE = 0.14, p < .001) and quadratic effects of age (b = −0.46, SE = 0.10, p < .001). As children got older, there was an increase in the likelihood that their letter name utterances would reference spelling, but the increase slowed during the last year of the study. For parents, the likelihood that a letter name utterance would reference spelling did not change significantly as a function of the child’s age.

Production, like spelling, was more often referenced by parents than children (b = −0.81, SE = 0.25, p = .001). The only other significant effects in the model for production were the interactions between speaker and linear age (b = 0.87, SE = 0.28, p = .002) and between speaker and quadratic age (b = −0.42, SE = 0.20, p = .032). A separate model for children found significant linear (b = 1.01, SE = 0.28, p < .001) and quadratic effects of age (b = −0.43, SE = 0.19, p = .024). As Table 1 shows, the likelihood that a child’s letter name utterance would reference production increased as the child got older, but the effect of age flattened out by around age 3½. A similar analysis for parents did not find significant effects of child age.

Sound was the least commonly referenced aspect of alphabet knowledge, as Table 1 shows. There was a significant main effect of speaker, such that parents were more likely than children to discuss the sounds of letters (b = −1.68, SE = 0.77, p = .030). There was also a significant main effect of linear age (b = 0.68, SE = 0.27, p = .011). Although the main effect of quadratic age was not significant, there were significant interactions of linear age and speaker (b = 4.93, SE = 1.73, p = .004) and quadratic age and speaker (b = −2.64, SE = 0.97, p = .007). A separate model for children found significant linear (b = 2.95, SE = 0.86, p < .001) and quadratic effects of age (b = −1.57, SE = 0.54, p = .003). The model for parents found a smaller but significant linear effect of age (b = 0.58, SE = 0.22, p = .008). The likelihood that a parent’s or child’s letter name utterance would reference the letter’s sound increased as children got older. In children, the increase with age slowed during the last year of the study.

Some utterances were complex in that they referenced more than one aspect of alphabet knowledge, and Table 2 shows the proportion of such utterances. A model like those described above showed that parents were more likely than children to produce complex letter name utterances (b = −0.65, SE = 0.15, p < .001). There was also a significant main effect of linear age (b = 0.19, SE = 0.06, p = .002). Although there was no main effect of quadratic age, there were significant interactions of linear age and speaker (b = 0.69, SE = 0.16, p < .001) and quadratic age and speaker (b = −0.32, SE = 0.12, p = .007). A separate model for children found significant linear (b = 0.83, SE = 0.15, p < .001) and quadratic effects of age (b = −0.37, SE = 0.11, p < .001). The model for parents found a smaller but significant linear effect of age (b = 0.17, SE = 0.05, p = .001). As Table 2 shows, the likelihood that a letter name utterance would reference more than one aspect of alphabet knowledge increased as children got older. In children, the effect of age flattened out by around age 4.

Table 2.

Proportion of Letter Name Utterances That Referenced More than One Aspect or No Aspects of Alphabet Knowledge as a Function of Child Age (Number of Utterances Referencing Each Aspect Out of Total Number of Letter Name Utterances in Parentheses)

	More than one aspect		No aspect
Child Age	Parents	Children	Parents	Children
1;2 – 1;10	.28 (163/587)	.01 (2/165)	.12 (72/587)	.34 (56/165)
2;2 – 2;10	.33 (351/1049)	.12 (80/675)	.05 (55/1049)	.15 (98/675)
3;2 – 3;10	.45 (625/1394)	.21 (304/1465)	.04 (561394)	.09 (128/1465)
4;2 – 4;10	.41 (718/1766)	.23 (331/1465)	.03 (48/1766)	.06 (92/1465)
Total	.39 (1857/4796)	.19 (717/3770)	.05 (231/4796)	.10 (374/3770)

Data on utterances that included letter names but that did not reference any of the aspects of letter knowledge considered here are shown in Table 2. The only significant effect in the statistical model was the main effect of speaker (b = 0.63, SE = 0.28, p = .023): Children were more likely than parents to produce idiosyncratic letter name utterances that did not reference any of the four aspects of alphabet knowledge.

Characteristics of Utterances that Referenced Spelling

Table 3 provides further information about the utterances that involved spelling. Overall, as the table shows, content words other than personal names were the most commonly spelled type of word. Parents were more likely than children to spell such words, as shown by a significant main effect of speaker (b = −0.47, SE = 0.15, p = .001). There were also interactions between the linear effect of age and speaker (b = 0.55, SE = 0.13, p < .001) and the quadratic effect of age and speaker (b = −0.21, SE = 0.10, p = .038). A separate model for children found significant linear (b = 0.54, SE = 0.13, p < .001) and quadratic effects of age (b = −0.22, SE = 0.10, p = .025) that appeared to reflect the emphasis on names in the earlier years that is documented in the analyses to be discussed next. A similar analysis for parents did not find a significant linear or quadratic effect of child age.

Table 3.

Proportion of Words Spelled in Spelling Utterances of Specific Types as a Function of Child Age (Number of Words of Each Type Out of Total Number of Words Spelled in Spelling Utterances in Parentheses)

	Content Word		Child’s Name		Other Name		Function Word
Child Age	Parents	Children	Parents	Children	Parents	Children	Parents	Children
1;2 – 1;10	.83 (153/185)	.50 (2/4)	.11 (21/185)	.00 (0/4)	.05 (10/185)	.50 (2/4)	.01 (2/185)	.00 (0/4)
2;2 – 2;10	.59 (213/359)	.63 (74/117)	.20 (72/359)	.17 (20/117)	.21 (75/359)	.21 (24/117)	.00 (0/359)	.00 (0/117)
3;2 – 3;10	.47 (272/578)	.32 (137/424)	.28 (162/578)	.38 (160/424)	.25 (146/578)	.30 (127/424)	.00 (0/578)	.00 (0/424)
4;2 – 4;10	.65 (373/578)	.51 (201/394)	.18 (103/578)	.36 (140/394)	.16 (93/578)	.13 (50/394)	.02 (11/578)	.01 (2/394)
Total	.59 (1011/1700)	.44 (414/939)	.21 (358/1700)	.34 (320/939)	.19 (324/1700)	.22 (203/939)	.01 (13/1700)	.00 (2/939)

The second most common type of word spelled was the name of the child. Although there was no main effect of age or speaker for this type of word, there were significant interactions between the linear effect of age and speaker (b = 1.79, SE = 0.52, p < .001) and the quadratic effect of age and speaker (b = −0.74, SE = 0.31, p = .016). A separate model for children found significant linear (b = 1.43, SE = 0.34, p < .001) and quadratic effects of age (b = −0.75, SE = 0.22, p < .001), such that the likelihood that children would spell their own name increased with age, with the effect of age flattening out by around age 4. The model for parents found only a quadratic effect of age (b = −0.20, SE = 0.10, p = .047). The likelihood that a parent’s spelling would be of the child’s name began decreasing when the child was around 3. There was also a significant main effect of SES (b = 0.44, SE = 0.20, p = .028). A letter name utterance was less likely to spell the child’s name when the speaker was of lower SES than when the speaker of higher SES. To illustrate, the proportion of letter name utterances that referenced the child’s name was .20 (197/997) for families that were below the median in SES and .29 (481/1642) for families that were above the median.

For spelling of personal names other than that of the child, there were main effects of linear (b = 0.33, SE = 0.21, p = .004) and quadratic age (b = −0.25, SE = 0.09, p = .009). As Table 3 shows, the likelihood that a parent or child would spell a personal name other than that of the child increased as children got older, but the effect of age flattened out by around age 3½.

Function words were rarely spelled, as Table 3 shows, and we did not conduct a statistical analysis of this category.

Table 4 provides information about the proportion of spelling utterances that were incomplete in that they mentioned just the first letter of a multi-letter word. There were main effects of linear age (b = −0.27, SE = 0.09, p = .005) and quadratic age (b = −0.19, SE = 0.08, p = .025), such that speakers were more likely to provide just the first letter of a word when children were younger than when they were older. In addition, linear age interacted with speaker (b = 0.61, SE = 0.20, p = .003). A separate model for parents that included the effect of linear age found a significant effect (b = −0.23, SE = 0.08, p = .006). Although there was a numerical decrease with age for children, the effect of age was not statistically significant.

Table 4.

Proportion of Words Spelled in Spelling Utterances That Mentioned Just First Letter of Word (Number of Words Spelled with Just the First Letter Out of Total Number of Words Spelled in Spelling Utterances in Parentheses)

Child Age	Parents	Children
1;2 – 1;10	.59 (109/185)	.50 (2/4)
2;2 – 2;10	.48 (171/359)	.46 (54/117)
3;2 – 3;10	.26 (149/578)	.16 (66/424)
4;2 – 4;10	.19 (109/578)	.14 (55/394)
Total	.32 (538/1700)	.19 (177/939)

Materials Used in Discussing Letters

We turn now to the materials that were used in letter name utterances that referenced a visible letter. The data, shown in Table 5, are pooled across parents and children because the statistical analyses to be discussed showed no significant effects of speaker. Strikingly, over three quarters of the letter name utterances that referenced a visible letter involved materials that were intended to teach literacy skills: manipulatives, such as blocks; paper-based teaching materials, such as workbooks; or electronic letter teaching materials, such as tablet games.

Table 5.

Proportion of Identification Letter Name Utterances Referencing Different Materials as a Function of Child Age (Number of Utterances Referencing Each Type of Material Out of Total Number of Identification Letter Name Utterances in Parentheses)

	Intended for Letter Teaching			Not Intended for Letter Teaching
Child Age	Manipulative	Paper-based	Electronic	Environmental Print	Writing	Storybooks
1;2 – 1;10	.35 (172/492)	.36 (176/492)	.00 (0/492)	.25 (125/492)	.01 (7/492)	.02 (12/492)
2;2 – 2;10	.64 (868/1366)	.13 (184/1366)	.04 (57/1366)	.12 (162/1366)	.01 (19/1366)	.06 (76/1366)
3;2 – 3;10	.41 (820/2024)	.24 (493/2024)	.15 (297/2024)	.11 (230/2024)	.06 (131/2024)	.03 (53/2024)
4;2 – 4;10	.30 (744/2485)	.36 (900/2485)	.13 (313/2485)	.13 (320/2485)	.04 (105/2485)	.04 (103/2485)
Total	.41 (2604/6367)	.28 (1753/6367)	.10 (667/6367)	.13 (837/6367)	.04 (262/6367)	.04 (244/6367)

The negative binomial regression models showed no significant effects related to speaker, age, or SES for manipulatives, paper-based teaching materials, or environmental print. (The apparent peak in references to manipulatives around age 2 was due in part to three families that spent a large part of a session at this age playing with blocks with letters on them.)

For the electronic teaching material category the significant effects were the linear (b = 0.57, SE = 0.20, p = .004) and quadratic effects of age (b = −0.36, SE = 0.14, p = .008). As Table 5 shows, the likelihood that a letter name utterance would reference a letter in electronic teaching materials increased as children got older, with the effect of age flattening out around age 3½.

The only significant effect in the model for the writing category (i.e., letters in a word that a parent or child was writing on materials that were not specifically designed to teach about letters) was that of linear age (b = 0.79, SE = 0.35, p = 0.026). The likelihood that a letter name utterance would reference a letter in writing increased as children got older.

The model for the storybook category found only a significant main effect of SES (b = 0.99, SE = 0.46, p = .032). A letter name utterance was less likely to reference letters in storybooks in a lower-SES family than in a higher-SES family. To illustrate, the proportion of letter name utterances that referenced letters in storybooks was .02 (65/2983) for families that were below the median in SES and .05 (179/3384) for families that were above the median.

Discussion

Parents are children’s first teachers, and developmental theories stress the importance of understanding the learning experiences that they provide and the ways in which they scaffold children’s learning (e.g., Berk & Winsler, 1995; Bronfenbrenner & Morris, 2006; Vygotsky, 1978). The present study examined an aspect of the home environment that is important for literacy development—talk about letters of the alphabet. Rather than using questionnaires, as in many previous studies, we studied transcripts of everyday conversations between parents and their 1.2- to 4.8-year-old children. We found that U.S. parents and children discuss some aspects of letters more than others and that there are some differences in the aspects of alphabet knowledge that are discussed as a function of the age of the child and the SES of the family. These differences would be missed if different aspects of alphabet teaching were not distinguished, as in some previous questionnaire studies (e.g., Chen et al., 2012). Importantly, the picture of the home literacy environment that we obtain by observing and analyzing letter talk in the home does not fully match the picture obtained from questionnaire studies. Our results have a number of implications for educators, as we discuss in what follows.

References to Different Aspects of Letters

Consistent with the results of questionnaire studies (Haney & Hill, 2004; Martini & Sénéchal, 2012), the parents and children in our study often focused on identifying letters in the environment. This was by far the most commonly discussed aspect of alphabet knowledge, showing that linking letter shapes to their names is a major focus of parents and young children. The finding that identification was the only aspect of letter knowledge that children referenced as often as parents points to children’s interest in learning about letters as names for visual shapes.

Although parents and children predominantly discussed letters as labels for shapes, they sometimes talked about the letters or letter sequences in specific words and about the production of letter shapes. Talk that explicitly linked the sound of a letter to its name was rare, although it did become more common as children got older, consistent with the results of Skibbe et al. (2013). Although parents sometimes spelled out the letters in a word, they did not usually talk about the letters’ links with sounds as a way to explain why the word included these particular letters.

The low rates of letter sound talk in the current study and in other observational studies (Bindman et al., 2014; Robins, Treiman, et al., 2014; Robins et al., 2012; Skibbe et al., 2013) are surprising given that most parents in questionnaire studies report that they teach their preschoolers about letters’ sounds (Haney & Hill, 2004; Martini & Sénéchal, 2012). One possible reason for the apparent discrepancy is that parents consider talk about letter names to convey information about letter sounds and so report that they teach their children about letter sounds. Although the names of many letters do contain their sounds, children who have learned for example that L is called /ɛl/ cannot necessarily extract /l/ from /ɛl/ and determine that the letter symbolizes just this part of its name. The differences between the observational and questionnaire studies with respect to letter sounds underscore the importance of observing the home literacy environment rather than relying in an uncritical way on parents’ reports about it.

Our results show that children’s talk about letters becomes increasingly complex across the toddler and preschool years. Specifically, references to production, sound, and spelling and utterances that mentioned multiple aspects of alphabet knowledge became more likely as children got older. These results demonstrate that everyday conversations can be a helpful indicator of children’s increasing knowledge about the alphabet. Traditional tests of early spelling knowledge, such as those that ask children to provide words beginning with a specific letter, are quite difficult for children below the age of 4 (e.g., Worden & Boettcher, 1990). Our results suggest that young children have some knowledge about how words are composed of letters that does not reveal itself in traditional tests.

Although parents’ references to some aspects of alphabet knowledge did not change as children got older, parents increased the complexity of their talk about letters over time in that they made more references to multiple aspects of letters within the same utterance and made more references to sounds. When parents referenced spelling, moreover, they were more likely to provide more than one of the letters in a word when children were older than when they were younger. In some ways, therefore, parents provide the kind of scaffolding discussed by Vygotsky (1978), tailoring their talk about the alphabet to the age and knowledge of the child.

Given the idea that the child’s name plays an important role in early literacy development (e.g., Levin & Aram, 2005), and given that many North American parents in questionnaire studies report teaching children to write their names (e.g., Martini & Sénéchal, 2012; Puranik et al., 2018), we examined the extent to which parents’ and children’s discussions of spelling focused on personal names versus other words. Although the parents in our study did discuss the spellings of names, over half of the words that parents spelled were content words other than names, such as nut. Such utterances give children a chance to learn that spelling is not a system that is restricted to names. However, the parents did not give children much opportunity to learn about the full breadth of the system, in that they, like the parents in the study of Robins and Treiman (2009), rarely discussed function words such as of as words that have spellings. Although the parents spelled a variety of words, children seemed to be particularly interested in spelling names, with over half of their references to spelling involving personal names.

Many of our findings held across families varying in SES. One difference, however, was that higher-SES families were more likely than lower-SES families to spell words. This reflected, in part, a greater tendency to spell children’s names on the part of higher-SES parents and children. A previous analysis of conversations from the Child Language Data Exchange System found more focus on the spelling of the child’s name in lower-SES families than higher-SES ones (Robins, Ghosh, et al., 2014). However, that study did not analyze utterances in which the word being spelled was not explicitly referenced, such as “J O E”, and it did not include cases in which the letters that made up the name were on separate lines of a transcript. Additionally, first names were not provided for all children in the database. Further research is needed to resolve this discrepancy regarding the spelling of children’s names.

Materials Used in Discussing Letters

Our study examined not only the aspects of alphabet knowledge that parents and children talked about but also the materials that they used to do so. The results of Martini and Sénéchal’s (2012) questionnaire study suggest that, when discussing letters with their young children, parents frequently use materials that are designed for other purposes, including storybooks and labels on household items. In our study, however, materials that were designed to teach about letters, such as activity sheets and alphabet blocks, were used in over three quarters of the letter name utterances that referenced visible letters. Thus, letter teaching was not as informal as expected given parents’ past reports of teaching materials. Interestingly, the lower-SES parents appeared to use materials that were designed for the purpose of letter teaching as much as the higher-SES parents. In the study of Fitzgerald et al. (1991), which asked parents about the literacy-related materials that are important to have at home, parents with lower literacy skills were more likely than parents with higher literacy skills to mention such things as flashcards and alphabet blocks. The lower-SES parents in the present study did not use materials that were specifically designed for alphabet teaching more often than the higher-SES parents, but it is notable that the lower-SES parents acquired these materials, even though they may be expensive. The lower-SES parents were not more likely than the higher-SES parents to use materials that were already in the home, such as the labels on food packages, a result that fails to support the hypothesis put forward by Chen et al. (2012). Our results suggest that U.S. parents of a variety of backgrounds consider it important for children to learn about letters before kindergarten begins and that they acquire and use materials that are specifically designed for this purpose.

Limitations

The present study examined utterances that included the name of a specific letter, but it did not examine utterances that referenced letters indirectly. Thus, it did not include utterances such as “Which one is this?” in the identification category, “How do you spell it?” in the spelling category, “You need to make another line on that” in the production category, and “It says sss” while pointing to a letter in a book in the sound category. Coding would have been more difficult if we had defined letter-related utterances more broadly. For example, it could be hard to determine whether “that” in “You need to make another line on that” referred to a letter a child was writing or a picture she was drawing. If such issues could be addressed, it would be important to determine whether the distribution of utterances that reference different aspects of alphabet knowledge is similar when indirect references to letters are included.

Another limitation of the present study is that it considered letter-related conversations only between children and their parents. Children also discuss these topics with siblings (Segal, Howe, Persram, Martin-Chang, & Ross, 2018), and inclusion of such discussions would provide a fuller picture of the home literacy environment. Also, the present study was limited to typically developing monolingual speakers of English, the majority of whom were White. Future studies should include children from other backgrounds and children who have disabilities, expanding on questionnaire-based studies with such populations (e.g., Breit-Smith, Cabell, & Justice, 2010).

Implications for Instruction

Although the present study has some limitations, it offers implications for instruction. Perhaps the most important implication relates to the fact that knowledge of letter sounds is critical for decoding written words and for constructing spellings of words whose spellings have not been stored in memory (e.g. Caravolas et al., 2001; Ehri, 2015). Based on the results of questionnaire studies in which most North American parents report that they teach letter sounds to their young children (Haney & Hill, 2004; Martini & Sénéchal, 2012), it would be natural to think that children have many opportunities to learn about this aspect of alphabet knowledge at home before kindergarten begins. Our results, together with those of other observational studies (Bindman et al., 2014; Robins, Treiman, et al., 2014; Robins et al., 2012; Skibbe et al., 2013), suggest that this is not the case. Rather, most of the letter talk that young children hear from their parents focuses on letters as labels for visible shapes. Parents sometimes spell words for their children, but they do not usually refer to letters’ sounds to explain why words are spelled as they are, potentially encouraging children to think of spelling as a memorization activity. The low rates of letter sound talk in our study are particularly striking given the frequent use of materials that are designed for letter teaching, such as worksheets and magnetic letters. The results suggest that teachers should not overestimate the knowledge of letter sounds that children bring with them to school. Teachers should provide systematic instruction about letter sounds and how to use them to read and spell words, following research-based guidelines (e.g., Jones, Clark, & Reutzel, 2013; Piasta, 2016; Stahl, 2015).

Studies show that alphabet instruction is more successful when it is provided by teachers than by parents (Piasta & Wagner, 2010). This difference may reflect the importance of content knowledge and pedagogical knowledge in literacy instruction (e.g., Moats, 2014). Teaching children about the links between letters and sounds and how to use them in reading and spelling words probably requires more specialized knowledge than teaching children about some other aspects of the alphabet. Therefore, it may be unrealistic to expect that parents can or should do much teaching about the sound aspect of the alphabet at home.

Another gap between what parents provide at home and what children need to learn in order to learn to read involves conceptual knowledge about words. The understanding that speech can be divided into words and that each word is represented in writing is important in learning to read and write (NELP, 2008), but dividing speech into words can be difficult for children. Dividing function words such as the and to from adjacent content words is especially difficult (e.g., Chaney, 1989; Holden & MacGinitie, 1972). Hearing about items such as the and to as words that have spellings could potentially promote word awareness in children, but our results suggest that parents rarely provide such input. This is another gap that teachers can fill.

Conclusions

The present study provides a more detailed picture of one aspect of U.S. children’s home literacy environment—learning about the alphabet—than is available from past studies. This picture differs in some ways from the one painted by questionnaire-based research. Notably, the results show that children have less exposure to letter sounds at home than expected on the basis of parental reports and more exposure to materials designed for letter teaching. By better characterizing parental practices—not just asking parents about what they do but observing them as they talk and interact with their children—we can better understand the home literacy environment. A better understanding of the aspects of alphabet knowledge to which children are exposed at home can, in turn, help in designing literacy instruction at school. In particular, the finding that U.S. parents do not provide their children with much information about letter sounds at home suggests that this information should be taught systematically at school.

Highlights.

• Conversations about letters at home help prepare children for literacy learning

• We analyzed letter-related conversations between parents and one- to four-year-olds

• Letter-related talk primarily focused on identifying letters

• Letter writing and spelling were also discussed, but sounds very little

• Materials designed for letter teaching (e.g., activity sheets) were often used