Patterns of literacy learning in German primary schools over the summer and the influence of home literacy practices
Abstract
The ‘summer learning effect’ (SLE) is described as a stall or drop in achievement over summer, especially in schools serving poor or ‘minority’ communities. There has been little research in Europe on the effect, and research internationally has primarily focused on the effect in reading, with minimal focus on writing. This paper describes the extent and nature of the SLE in reading comprehension and writing in second grade classrooms in Germany. The SLE was evident in both subject areas with all students experiencing lower progress rates over summer. In reading, students attending the high‐income school progressed significantly more over summer than their low‐income peers, while there was no significant difference in writing progress over summer. Literacy logbooks over summer and interviews with a subset of students provided information on how home literacy practices influenced the effect.
Reading and writing are both essential cultural tools and are vital to the educational success of children (Kerckhoff, Raudenbush, & Glennie, 2001). Major international studies consistently identify issues relating to inequalities in the attainment of literacy skills for certain groups of students (OECD, 2010, 2011). In North America, research has identified the summer break as a major influence on differential attainment of low‐income and high‐income students (Entwisle & Alexander, 1992; Entwisle, Alexander, & Olson, 1997; Heyns, 1978). Such seasonal comparisons of achievement have shown that low‐income students tend to learn at the same rate as their more affluent peers while in school. Over summer, these students typically stagnate or lose ground in their learning, while high‐income students tend to continue to gain. Cooper, Nye, Charlton, Lindsa, and Greathouse (1996) conducted a meta‐analysis of 39 studies that suggested that the summer break created on average an annual achievement gap of about three months between low‐income and high‐income students. Longitudinal data in the United States showed summer losses accumulated and widened the achievement gap over consecutive years (Alexander, Entwisle, & Olson, 2007).
This negative impact of summer vacation is known variously as the ‘summer slide’, ‘summer setback’ or ‘summer learning effect’ (SLE) and has received considerable policy attention in the United States. A growing movement is calling for more summer schools and shorter summer holidays, and a national summer learning day has been declared to call attention to the issue. TIME magazine ran a cover article in 2010 titled ‘The case against summer vacation’ (von Drehle, 2010). However, Slates, Alexander, Entwisle, and Olson (2012) investigated the variability of the effect across social lines and found that not all low‐income students are prone to suffer from the effect. Further research is needed to investigate the mechanisms behind the SLE and extend our knowledge of the effect in regard to different groups of students within different settings and contexts.
Previous research also suggests that the effect differs by subject area. Losses have been found to be typically greater in mathematics than for reading (Cooper et al., 1996). Much of the previous research about the SLE has focused on reading and mathematics, with scant research examining the effect on students' writing progress. The current study also investigates the SLE for writing.
SLE research in Europe
While there have been several North American studies, little research has investigated the SLE in the European context. The effect may differ in different education systems, communities and countries; given for example, the shorter summer break in most European countries (6–10 weeks) in contrast to the longer break in the United States (10–12 weeks). Only three studies could be identified that have investigated summer learning patterns in Europe to date (Becker, Stanat, Baumert, & Lehmann, 2008). They had varying foci and results, with two studies focusing on mathematics achievement and one on reading comprehension. In reading comprehension, a SLE was found in Germany that was unrelated to students' socioeconomic status (SES) (Becker et al., 2008). Two studies focusing on mathematics, one from Sweden (Lindahl, 2001) and one from the French part of Belgium (Verachtert, Van Damme, Onghena, & Ghesquière, 2009), needed to adjust test scores statistically because their test dates included several weeks of schooling within the period analysed as ‘the summer break’; hence, their results could be confounded by what happened in the included school weeks. While the German research that investigated reading comprehension over summer used test dates close to the summer break, the authors did not compare summer achievement to achievement over school periods (Becker et al., 2008). Thus, they could not report on the general nature of the SLE.
When comparing groups from different socioeconomic backgrounds in a European context one needs to consider the number of migrant students who most often speak a different language at home. Becker et al. (2008), who focused on migrant students and SES did not find an additional effect for language spoken at home, while other European studies (Lindahl, 2001; Verachtert et al., 2009) found a positive effect for learning over the school term for students speaking another language at home.
Further research into the SLE in reading and writing and its dependence on students' background in a European context is thus needed. In this article, we explore the SLE in a German context with second grade students from a high‐income and a low‐income school, aiming to compare summer learning and school learning in the area of literacy for these two student groups. At the same time, the present study aims to overcome a methodological problem in early summer learning studies. These studies often included several weeks of instructional time in the summer period, potentially underestimating the SLE (Cooper et al., 1996). More recent studies have reacted to this criticism by statistically estimating test scores as if tests were taken on the last and first school days before and after summer (e.g., Burkam, Ready, Lee, & LoGerfo, 2004; Downey, von Hippel, & Broh, 2004; Lindahl, 2001). Von Hippel (2009), however, argues that setting actual test dates closer to the summer break would be the most valid measure of summer learning. In the study presented here, test dates were set in the school week before and after the summer break.
Variability across subjects
North American research suggests diverse patterns in summer learning across different subject areas. Cooper et al. (1996) documented how losses were more pronounced in domains involving memorisation and procedural skills (mathematics and spelling) than in domains needing conceptual understanding (mathematical concepts, reasoning and reading comprehension). Also, there may be more resources and activities already present in communities and families for reading outside of school than for solving mathematical problems or practicing spelling (Cooper et al., 1996). Thus, arguably, spelling and higher‐order writing skills are likely to be as susceptible as reading to SLE and possibly even more like mathematics in these respects. It may involve more memorisation and procedural knowledge in the beginning stages than reading, because of the extra encoding and decoding requirements, and there may be fewer practices and resources for writing at home. Thus, our hypothesis was that losses would be at least similar to those in reading and perhaps greater than those in mathematics. In previous studies, subject variability of the SLE was further associated with students' socioeconomic background. All students, regardless of their background, had similar declines in mathematical skills, but considerable differences were found in reading. Low‐income students tended to show losses in reading (word recognition), whereas high‐income students tended to gain reading skills over summer. In reading comprehension a decline was seen in both groups, but again, the scores for low‐income students declined even further. Our study expands on the current research by examining writing in addition to reading comprehension over summer of high and low‐income students in a European context.
Furthermore, the existing research on students' writing practices at home has predominantly looked at the developmental processes of writing before children enter school (Mayer, 2007; Pinto et al., 2012). Research seems to overlook the importance of home‐writing practices once children start school. In this article we use a summer effect study to examine home‐writing practices of school age children attending to this gap in the literature.
Family literacy practices influencing summer learning
It seems well established that there are differences between school and summer learning. However, it is less clear which mechanisms influence summer learning. A number of theories exist, building on developmental learning theories, educational attainment research, sociological or educational research or evaluations of summer school programs (Borman, Benson, & Overman, 2005; Burkam et al., 2004; Slates et al., 2012). A commonality is that they all tend to draw on the concept of ‘cultural capital’ (Bourdieu 1986). Drawing on this concept, children from higher income backgrounds are seen as immersed in a daily home environment that increasingly advantages them by expanding their cultural repertoire. This cultural repertoire provides these children with skills, knowledge and behaviours that are favoured by society and help them to succeed. Entwisle, Alexander, and Olson (2001) proposed a model that linked the differential learning patterns over the summer break with the cultural resources available to different groups of children, using a ‘resource faucet’ metaphor. The faucet is turned on for all children when school is in session, and thus, all students gain more equally; however, when school is not in session, the school resource faucet is turned off resulting in unequal gains as disadvantaged families cannot make up for the resources the school is otherwise providing. Their children stall or lose in their achievement over summer. Higher income families, who can access various resources, can make up for the absence of the school's resources more effectively, by providing a richer cultural capital environment to their children.
However, the exact nature of the differences in environmental resources that are or are not flowing to the advantage of children's literacy skills over summer is still under‐researched. There is, however, considerable research on the presence and patterns of family literacy activities before school and their significance for early reading development (Fitzpatrick, Grissmer, & Hastedt, 2011; Heath, 1983; Paris & McNaughton, 2010; Pinto, Bigozzi, Gamannossi, & Vezzani, 2012; Sastry, 2010; Skibbe, Grimm, Bowles, & Morrison, 2012; Waldfogel, 2012). There are inconsistencies in the research, however, around how much variability exists. Some studies show that early literacy practices are similar across a range of SES backgrounds (Bus, van IJzendoorn, & Pellegrini, 1995; Hood, Conlon, & Andrews, 2008), while Serpell, Baker, and Sonnenschein (2005) found that practices differed. Also, Stainthorp and Hughes (2000) showed that the amount of parent–child reading did not differentiate early readers from nonreaders, while Bus et al. (1995) concluded that more shared reading practices had a positive impact on children's reading outcomes.
Thus, the exact role cultural capital plays in children's literacy development, especially over summer, needs further exploration (Downey, von Hippel, & Hughes, 2008). Access to reading materials is important, although previous research investigating the effect of differential access to literacy resources at home found that sheer access to books may have little positive effect or, at best, a very small positive effect on reading achievement (Carver & Leibert, 1995; Kim & White, 2008; Mol & Bus, 2011; Myrberg & Rosén, 2009; National Reading Panel, 2000). One factor likely to contribute to summer learning is the time actually spent reading. In Heyns' (1978) classic study every additional hour a day spent reading over summer was worth an extra month of achievement on the standardised test. However, Burkam et al. (2004) only found a marginal relationship between summer literacy activities and summer learning, and Slates et al. (2012) reported that only two of the examined summer activities reached statistical significance in their study. Parents of low SES students, who gained in reading over summer, were significantly more likely to take their children to the library during summer and check out books while there.
While Burkam et al. (2004) and Slates et al. (2012) used established cultural capital frameworks, the current study draws from a model of literacy development that is based on socio‐cultural and ecological concepts, in which the explicit practices assume more significance than the given resources assumed present on the basis of the families' SES (Bronfenbrenner, 1979; Bronfenbrenner, 1986; Bronfenbrenner & Morris, 2006; McNaughton, 1995). Here, literacy development is regarded as a process which is co‐constructed by the learner and a guiding person, through frequently and regularly occurring activities over extended periods of time that children engage in with other persons or with symbolic objects and which comprise valued socialisation practices (Bronfenbrenner & Morris, 2006). These practices take place in schools, neighbourhoods and in students' homes, and hence, literacy development is influenced by practices in different environments as well as by the inter‐relationships between these immediate environments of the child and the inter‐relationships of the practices in those (Bronfenbrenner, 1979; Bronfenbrenner & Morris, 2006). At the most immediate developmental sites (the microsystem level), in the family and classroom context, these practices are characterised by interpersonal relationships and face‐to‐face activities. At a more distant level, the mesosystem, the environment includes connections between settings and external influences in which children do not directly participate but by which they are indirectly affected (i.e., exosystems). These are, for example, neighbourhood poverty, school settings and parents' employment or unemployment (Hart & Risley, 1995; Neumann & Celano, 2001). Such indirect influences may exert a powerful impact on the social processes that take place within immediate settings. Analyses of learning patterns over the school year and summer provide a means of extending not only our understanding of these ecological connections within these systems but also our understanding of how variability in the social and cultural practices in families contribute to sustained learning.
The present investigation
The varying results of research on the SLE in European contexts, the lack of research in subject areas beyond reading and mathematics and on the underlying mechanisms of summer learning underscore the need to further investigate the SLE. Based on the theoretical considerations and previous research, this study aimed to obtain a comprehensive picture of the SLE in reading comprehension and writing and the associated literacy practices at home. The present investigation focused on the transitional summer of Grades 2–3. Second graders in German schools are typically 7–8 years old. The chosen summer is of special interest as students in Grade 2 have gained basic literacy skills and start to take more control of their own literacy habits but may be still quite susceptible to influences from the family environment.
The study addressed the following two research questions:
- What is the extent and nature of the SLE in literacy, both reading comprehension and writing, in the communities of two primary schools situated in low and high‐income communities in Germany?
- What kind of family literacy practices can be identified that are associated with continued learning over summer?
Method
Participants
Participants were drawn from the second grade classrooms of two primary schools in a medium‐sized city in Germany. The schools were purposively sampled to have contrasting socioeconomic profiles. Socioeconomic profiles of each school were determined on the basis of statistical district data taking into consideration neighbourhood SES. Furthermore, parental occupation data were collected to compare these profiles against family SES. Achievement data were collected from a total of 82 students from six classrooms. Data from four students, who changed schools during the summer, had to be excluded. Forty‐three students were from school A and 35 from school B. Male students were slightly under‐represented (44%). Based on the sample of questionnaire respondents, 94% of families at school A and only 30% at school B spoke German at home.
Design
A short‐term longitudinal design was used to examine reading comprehension and writing achievement at four time points. The design provided achievement trajectories during school and summer. To investigate family literacy practices associated with continued learning over summer both quantitative and qualitative measures were used. Measures included parents' questionnaires to obtain demographic data, literacy logbooks to record daily summer literacy activities of students and retrospective interviews with students and parents in regard to summer literacy activities.
Measures
Achievement measures
To build the most accurate model of summer learning, careful attention was paid to the testing dates. Achievement measures were collected at four time points (T1, T2, T3 and T4), with T2 and T3 set as close to the summer break as possible, in the week before and after summer. The other two test dates (T1 and T4) were set 7 weeks before and after summer. Therefore, each of the test periods comprised 7 weeks.
Reading comprehension was assessed using ELFE 1–6, a well‐established standardised reading comprehension test for students from Grades 1–6 (Lenhard, 2006). This test assesses reading comprehension through a multiple choice test. It examines not only basal reading comprehension but also syntax understanding and text comprehension. While the test's re‐test reliability is reported to be 0.96 (Lenhard, 2006), the average re‐test reliability for the current study was 0.89.
Writing achievement was assessed using the Hamburger Schreib‐Probe Grade 2, a standardised test assessing students' basic writing skills. It contains 15 single words and three sentences. Words and sentences are read out in normal, fluent speech rather than being dictated; that is, the full sentence is read without emphasised pronunciation of word stems or endings, and it is read in one go, without rereading parts of sentences. Whole sentences are, however, read repeatedly. The test assesses spelling but also students' proficiency in writing strategies: alphabetic, orthographic and morphemic. While the technical manual reports a correlation of scores of r = .89 when re‐testing after a 6‐month interval (May, 2010), the average re‐test reliability for the current study was .91.
To minimise practice effects, test results were only given to teachers for diagnostic purposes, but not to students. Teachers were asked not to discuss test items in between measurements. Because the achievement tests used in this study were standardised at the beginning and middle of the school year, we used the same test materials at all four time points to gather comparable data across the length of the study. Scoring of tests was based on national norms for the end of Grade 2; thus, scores depict actual achievement gains or losses rather than relative ones (for a discussion of relative versus absolute change, see Cooper et al. (1996)). Both tests report achievement in percentile ranks so scores were transformed into z‐scores for statistical analysis to achieve a normal distribution.
Parents' questionnaires
Parents' questionnaires sought information on family demographics: number of children living at home, two or one parent household, language spoken at home, parental education, the number of books and children books at home. Response rates were 75% at school A and 67% at school B.
Occupational data from both parents were coded using the Socio‐Economic Index of Occupational Status (ISEI), and the Highest Socio‐Economic Index of Occupational Status (HISEI) was derived based on this information (Ganzeboom & Treiman, 1996). If there was occupational information missing for either parent, the parent received the ISEI code of the partner. It has to be noted that the ISEI ranking was developed on a sample of only male students in full‐time occupations. Estimations for women have been made using data for men working in characteristically female occupations (Ganzeboom & Treiman, 2003; Rose, 2005). A code for mothers staying at home is missing; consequently, these cases received the ISEI code of their partners. These discrepancies are a problem shared by most socioeconomical coding taxonomies and classification systems (Hauser & Warren, 1997).
Literacy logbooks
Literacy logbooks were distributed to all students in the last week of school, to be completed over the first 2 weeks of summer. The logbook presented eight multiple choice questions per day asking students to record if and what they read and wrote that day, how long they spent reading and writing and with whom. Time was taken in class to help students understand the task and to ensure the information they would give about their literacy activities was as complete and accurate as possible. The researcher advised students to become time conscious and to make mental notes of when they started and stopped activities. A page displaying clocks with different time intervals was added to the logbook as a reference page for the students. A total of 49 students, 31 from school A and 18 from school B, returned the logbook after summer and received an age appropriate book from a popular series. Across schools, respondents were 41% male students.
Students and parents' interviews
Thirty‐four students and parents signalled their availability for an in‐depth interview. Sixteen families were selected, ensuring an equal number of students from each school and an even representation of the different progress rates over summer defined by a summer learning profile. Three summer learning profiles were distinguished on the basis of the change in students' achievement scores over summer. The three groups were (1) gain (defined as gains of >2.33); (2) stall (changes over summer of up to ±2.33); and (3) drop (drops of >−2.33). Parents in the interview sample held a range of occupations with those in school A typically holding much higher ranked professions (HISEI M = 57.75, SD = 15.97) than those in school B (HISEI M = 35.13, SD = 17.32).
The profile distribution of students for the families that were interviewed is shown in Table 1 in the succeeding texts. Some students exhibited different progress for reading than for writing. Two students gained in reading and writing, while three dropped in both areas. The rest of the student mostly stalled in one area while gaining or dropping in the other. Ten of the 16 families spoke German at home, while four families spoke Turkish or Polish at home, all of which were in school B, and two families raised their children bilingual, both of them were in school A.
| School | Reading | Writing | ||||
|---|---|---|---|---|---|---|
| Gain | Stall | Drop | Gain | Stall | Drop | |
| School A | 3 | 2 | 3 | 2 | 1 | 5 |
| School B | 2 | 4 | 2 | 4 | 1 | 3 |
The interviews addressed literacy practices and aspects of students' home literacy environment: students and parents' literacy practices (own reading and writing preferences and practices), shared literacy practices (parents and child), access to reading materials (library use and books at home) and parental guidance and support. Questions focused mostly on practices over summer, but included some questions on general literacy practices.
Analyses
Quantitative analysis
Initially, all quantitative variables were analysed for patterns of missingness to determine whether imputation was an appropriate response. Imputation was inappropriate in this context because data were either available for all aspects of the measure, or none. For example, those that completed the questionnaire responded to all items, while those with missing responses did not complete any aspect of the questionnaire. There were no significant differences between achievement and SLE of students whose parents completed the questionnaire compared with those whose parents did not. Similarly, there were no significant differences between achievement, SLE and HISEI (where available) of students who completed the literacy logbooks compared with those who did not.
Quantitative data from parents' questionnaires and students' literacy logbooks were either coded as continuous variables (e.g., HISEI) or categorical variables (binary or polytomous). Multivariate hierarchical linear growth models (HLMs; Raudenbush & Bryk, 2002) were estimated using MLwiN 2.26 (Rasbash, Charlton, Browne, Healy, & Cameron, 2012). These models tested predictors of progress over the four time points, with a particular focus on the change during summer. HLM uses a multilevel regression framework, in which the hierarchical model incorporates information about the underlying structure in its estimates. This is advantageous, because educational data typically violate the statistical assumption of independence required in traditional methods of analyses (Osborne, 2000). For example, there is likely to be a degree of dependence among students attending a particular school that is not shared across different schools (e.g., differences in school climate and learning experiences). Students in the same school can be conceptualised as nested within schools, reflecting two levels of hierarchy, with students as Level 1 and schools as Level 2. In the current case, however, there are only two schools, so little would be gained by including a school level in the hierarchy. Instead, the use of HLM is advantageous here because it allows individual student achievement growth to be modelled as occasions, that is time‐varying repeated measures (Level 1), nested within students (Level 2) (Rasbash, Charlton, Browne, Healy & Cameron, 2012). That is, individual differences in progress are estimated for each student at Level 1, while the overall effects of the student‐level predictors on these individual differences in progress are estimated at Level 2. Adding classrooms as a level did not explain additional variance in the analysis. Since the primary aim was to compare progress during school with progress during summer, and to determine whether predictors of progress differed during these periods, a piece‐wise growth model was used. A piece‐wise model estimates linear growth from baseline to the third and fourth measurements, allowing for the shifts over summer and during each school period to be calculated. Therefore, the general growth model was specified as follows:
-
- Level 1
-
- StudentAchievementi = π0i + π1i(Time) + eti
-
- Level 2 Intercept
-
- π0i = β00 + β0nXn + r0i
-
- Level 2 Growth
-
- π1i = β10 + β1nXn + r1i
Qualitative analysis
Qualitative data came from interview recordings from student interviews and notes from parent interviews. As a first step, the researchers ‘immersed’ themselves in the data by reading interview notes and listening to recordings repeatedly (Mostyn, 1985). As summer learning profiles of students often differed in reading and writing, the data were analysed with a focus on either reading or writing. The second step of qualitative data processing lay in reducing the data (Miles & Huberman, 1994). Data were coded with codes corresponding to the interview questions (descriptive codes). Descriptive codes were, for example, shared reading practices, appreciation of reading and reading materials, among others. NVivo was used as a coding tool and assisted in structuring, saving and copying raw materials. A random sample of 100 text units from the interviews was taken, representing 20% of the given data, to measure coding reliability. A second coder, who had not been involved in the project, was asked to code the units with the given codes. Inter‐rater reliability was calculated using ReCal2 (Freelon, 2010) and was assessed as 83% agreement or Krippendorff's alpha = .797, which is a considered as an acceptable to high agreement (Banerjee, Capozzoli, McSweeney, & Sinha, 1999; Krippendorff, 2004; Taylor & Watkinson, 2007). Data were then displayed in a matrix in which the rows were formed by the descriptive codes and columns depicted families. Short citations or descriptions were given in each cell for the student and/or the parent. The matrix was further divided into profile groups (gain, stall and drop) and allowed for comparisons between students and parents' data and between and within profile groups.
Results
The summer learning effect
Our analyses showed that students typically experienced a change in achievement growth in both reading comprehension and writing. Overall, achievement over summer stalled in reading comprehension (.03, SE = .06) and declined significantly in writing (−.09, SE = .04) (Table 2, Model 2). Of particular interest was whether there were factors that were associated with differential progress during summer. Differences were identified based on the SES of the school that students were attending (35 students attended the low‐income school and 42 the high‐income school), as well as the language spoken at home (56 students spoke German at home, while 21 spoke a language other than German). These results are also shown in Table 2 and discussed in greater detail in the succeeding texts.
| Parameter | Model 1 | Model 2 | Model 3 | Model 4 | Model 5 |
|---|---|---|---|---|---|
| Reading effects | |||||
| Intercept | .048 (.13) | −.622* (.14) | −1.17* (.19) | −.384* (.14) | −.978* (.26) |
| T1–T2 | .656* (.06) | .695* (.08) | .659* (.07) | .753* (.12) | |
| T2–T3 (SLE) | .033 (.06) | −.09 (.08) | .058 (.07) | .143 (.12) | |
| T3–T4 | .645* (.06) | .768* (.08) | .555* (.07) | .604* (.12) | |
| High SES | .998* (.26) | −.799* (.31) | |||
| T1–T2 * SES | −.072 (.11) | −.128 (.14) | |||
| T2–T3 (SLE) * SES | .226* (.11) | .275* (.14) | |||
| T3–T4 * SES | −.225* (.11) | −.068 (.14) | |||
| Lang | −.827* (.28) | −.305 (.33) | |||
| T1–T2 * Lang | −.014 (.13) | −.1 (.16) | |||
| T2–T3 (SLE) * Lang | −.094 (.13) | .093 (.16) | |||
| T3–T4 * Lang | .332* (.13) | .285 (.16) | |||
| Writing effects | |||||
| Intercept | −.048 (.11) | −.28* (.11) | −.789* (.15) | −.082* (.11) | −.706* (.2) |
| T1–T2 | .252* (.04) | .296* (.06) | .202* (.05) | .193* (.09) | |
| T2–T3 (SLE) | −.092* (.04) | −.135* (.06) | −.051 (.05) | −.051 (.09) | |
| T3–T4 | .35* (.04) | .333* (.06) | .327* (.05) | .241* (.09) | |
| High SES | .929* (.2) | .842* (.24) | |||
| T1–T2 * SES | −.079 (.09) | .013 (.1) | |||
| T2–T3 (SLE) * SES | .075 (.09) | −.001 (.11) | |||
| T3–T4 * SES | .032 (.09) | .116 (.11) | |||
| Lang | −.68* (.22) | −.119 (.25) | |||
| T1–T2 * Lang | .173* (.1) | .181 (.12) | |||
| T2–T3 (SLE) * Lang | −.144 (.09) | −.145 (.12) | |||
| T3–T4 * Lang | .083 (.1) | .158 (.12) | |||
| Deviance (MCMC) | 718.42 | 285.19 | 284.12 | 278.02 | 277.31 |
- Lang, language other than German; SES, high SES school (school A).
- * p < .05.
Language and school‐level SES were found to have an effect on progress rates during summer, with both variables providing significant improvements in model fit compared with the baseline model (which included only changes over time). The best model fit was provided by inclusion of the binary dummy variable for language spoken at home (0 = German, 1 = Other; Model 4), while comparison between the low and high SES school added little once language was accounted for (Model 5).
It is interesting to note, however, that progress among students attending the high‐income school was higher during summer, but lower during schooling, relative to students attending the low‐income school, despite significantly higher baseline achievement (Model 3). Overall, students attending the high‐income school made marginal (nonsignificant) gains in reading during summer (0.14 ± 0.11), while those attending the low‐income school stalled (−0.09 ± 0.08). The difference in summer progress was statistically significant (0.23 ± 0.11) between the low‐income and high‐income schools for reading. In writing, students in the low‐income school had a significant achievement drop (−0.14 ± 0.06), while students attending the high‐income school stalled (summer drop of −0.14 + 0.08 = −0.06); these differences were marginal and not statistically significant.
Model 4 indicates that students who spoke a language other than German at home were likely to make larger gains (0.87 ± 0.13) during the school term after summer in reading than students who spoke German at home (0.56 ± 0.07). Conversely, in writing, these students made modest gains during the two periods of schooling (T1 to T2 = 0.38 ± 0.1 and T3 to T4 = 0.41 ± 0.1), relative to students speaking German at home. Over summer, students speaking another language at home appeared to fall further behind in reading and writing achievement compared with students speaking German at home, but this difference was not statistically significant (−0.09 ± 0.13 in reading and −0.14 ± 0.09 in writing).
Other variables tested are shown in Table 3 and included categorical and polytomous variables. The data were collected using different measures; hence, the sample size differs depending on the availability of questionnaire and logbook data.
| Categorical predictors | Na | Characteristic | n | % |
|---|---|---|---|---|
| Gender | 77 | Male | 35 | 45.5 |
| Female | 42 | 54.5 | ||
| Father's education | 47 | No school qualification | 2 | 4.3 |
| Secondary school – lowest tier | 8 | 17.0 | ||
| Secondary school – medium or highest tier | 6 | 12.8 | ||
| Vocational training | 14 | 29.8 | ||
| Polytechnic degree | 4 | 8.5 | ||
| University degree | 13 | 27.7 | ||
| Mother's education | 49 | No school qualification | 3 | 6.1 |
| Secondary school – lowest tier | 9 | 18.4 | ||
| Secondary school – medium or highest tier | 9 | 18.4 | ||
| Vocational training | 12 | 24.5 | ||
| Polytechnic degree | 7 | 14.3 | ||
| University degree | 9 | 18.4 | ||
| One‐parent or two‐parent household | 51 | One‐parent household | 11 | 21.6 |
| Two‐parent household | 40 | 78.4 | ||
| Number of books at home | 51 | Up to 200 | 29 | 56.9 |
| More than 200 | 22 | 43.1 | ||
| Number of children's books at home | 51 | Up to 50 | 17 | 33.3 |
| 50 to 100 | 21 | 41.2 | ||
| More than 100 | 13 | 25.5 | ||
| Continuous predictors | Characteristic | M | SD | |
| HISEI | 51 | HISEI scale ranging from 16 to 88 | 48.4 | 18.5 |
| Time read per dayb | 47 | Average time read per day in minutes | 22.5 | 24.7 |
| Time wrote per dayc | 47 | Average time wrote per day in minutes | 10.0 | 10.2 |
- a Sample size differs depending on availability of achievement, questionnaire and logbook data.
- b Maximum average time for reading per day was 120 minutes.
- c Maximum average time for writing per day was 65 minutes.
The demographic variables including gender, HISEI, father and mother's education level, two or one parent household and variables around student access to reading materials (number of books and children's books at home) were not reliably associated with progress rates during summer. Furthermore, literacy activities of students as reported in the literacy logbook (time read per day, time wrote per day) could also not reliably be associated with progress rates during summer. Missing data for the parents' questionnaires and literacy logbooks halved the effective sample size, making it difficult to pinpoint any effects on summer learning.
Table 4 presents the relative progress rates associated with these variables; however, while some variables may appear to have an impact on progress rates over summer, none was statistically significant, and their addition to the models did not improve model fit. Therefore, the results presented in Table 4 should not be used to draw implications, but rather to point to interesting areas for further research. When mapping individual gain scores over summer, the scores reveal high variability of learning patterns in reading comprehension and writing at both schools. These results point, for example, to the importance of the quality of shared literacy activities and prompt further research into the relationship between writing activities and reading achievement and vice versa.
| Nonsignificant predictorsa | Nc | T1–T2 | T2–T3 (Summer) | T3–T4 |
|---|---|---|---|---|
| Reading | ||||
| Gender (female)*time | 77 | .031 (.11) | .025 (.11) | .039(.11) |
| HISEI*time | 51 | .000(.004) | .006(.004) | .002(.004) |
| Father's educationb*time | 51 | −.033(.05) | .083(.05) | .007(.05) |
| Mother's educationb*time | 51 | −.04(.05) | .142(.05) | −.038(.05) |
| Two‐parent or one‐parent household*time | 51 | .159(.16) | −.354(.16) | −.148(.16) |
| Number of books at homeb*time | 51 | −.072(.14) | .194(.14) | −.026(.14) |
| Number of children's books at homeb*time | 51 | .021(.09) | .092(.09) | −.065(.09) |
| Time read per dayb*time | 47 | −.12(.09) | .144(.09) | −.104(.09) |
| Time wrote per dayb*time | 47 | −.032(.09) | .111(.09) | −.013(.09) |
| Writing | ||||
| Gender (female)*time | 78 | −.059(.08) | .107 (.08) | .055(.08) |
| HISEI*time | 52 | .001(.003) | −.001(.003) | .002(.003) |
| Father's educationb*time | 52 | .007(.04) | −.012(.04) | −.015(.04) |
| Mother's educationb*time | 52 | .012(.04) | .002(.04) | .022(.04) |
| Two‐patent or one‐parent household*time | 52 | .132(.13) | .087(.13) | −.169(.13) |
| Number of books at homeb*time | 52 | .065(.11) | .074(.11) | −.11(.11) |
| Number of children's books at homeb*time | 52 | −.059(.07) | .044(.07) | −.054(.07) |
| Time read per dayb*time | 48 | −.02(.08) | −.02(.08) | .092(.08) |
| Time wrote per dayb*time | 48 | −.04(.07) | −.062(.07) | .113(.07) |
- a Factors are on different scales or are categorical. Refer to method section and in‐text descriptions of variables.
- b Variables have been recoded into equal categories (e.g., low, medium, high or low and high).
- c Sample size differs depending on availability of achievement, questionnaire and logbook data.
Families' literacy practices
Based on the variability of the SLE, the qualitative analysis sought to examine families' literacy practices in relation to achievement profiles over summer (gain, stall and drop).
Reading comprehension
Indicative patterns emerged in families' reading practices in the profile groups; see Table 5 for an overview of results. There tended to be differences in students' appreciation of reading, which were associated with the frequency with which students engaged in reading practices, the way students accessed reading materials and here especially the families' use of the library, the appropriateness of the materials and shared reading experiences.
| Profile group | |||
|---|---|---|---|
| Theme | Gain | Stall | Drop |
| Appreciation of reading | Liked reading; agreement between parents and students | Liked reading; agreement between parents and students | Mixed appreciation of reading, prone to change, dependent on subject or book; mixed agreement between parents and students |
| Reading mileage | Daily, reading seen as ‘normal’ | Daily or almost daily | two to three times a week, irregular reading times |
| Appropriateness of reading materials | Materials with appropriate reading levels | Mostly materials with appropriate reading levels | Materials not appropriate to students’ reading levels |
| Parental guidance | Knew what their children were reading, generally involved in choosing materials, had effective strategies to identify appropriate materials | Unsure what their children were reading, less involved in choosing materials, had effective strategies to identify appropriate materials | Sometimes knew what their children were reading, less involved in choosing materials, had no effective strategies to identify appropriate materials |
| Access to reading material | Books as gifts, owned books, inherited books (from siblings and wider family), borrowed from the library | Owned books, inherited books (from siblings and wider family), borrowed from friends | Owned books, inherited books (from siblings and wider family), students bought books or comics themselves |
| Use of the library | Regular visits, also in the holidays | Irregular use of the library | No use of the library |
| Shared reading practices | Students read mostly by themselves, parents regularly read to their children | Students read mostly by themselves, parents regularly read to their children | Students read mostly by themselves, different accounts on someone reading to the child |
| Parental help and encouragement | ‘No need’ for encouragement | Helped, encouraged and reminded students to read | Did not encourage, helped if children asked for help |
| Parental reading practices | Mixed reports, mostly daily reading activities for entertainment and to be informed | Mostly liked reading, read two to three times a week for entertainment and to be informed | Liked reading, read daily for entertainment, some referred to work related reading tasks |
| Parental perception of importance of reading | Rated as important, to develop thinking, imagination and language | Rated as important, entertainment, to develop ‘mind and soul’, language | Rated as important, means to dive into another world |
Students who gained during summer typically showed a general appreciation of reading. These students typically engaged regularly in reading activities on their own over summer and reading was described as a part of daily life. Parents were usually informed about what their children were reading and involved in choosing reading materials. They tended to have explicit strategies to identify appropriate materials. As a result, students in this group were more likely to read books appropriate to their reading levels. Parents and students identified the classroom and public library as a means to access books. All students had visited the library in the holidays, mostly more than once, and visited regularly during the school term. Students were often reading by themselves, but parents in this group also tended to engage in shared reading practices with their children, with some doing so daily.
Students who stalled during summer usually showed a general appreciation of reading and had typically engaged in reading activities on their own regularly over summer. Parents in this group tended to be somewhat less knowledgeable about what their children were reading and seemed less involved in choosing materials. Still, students usually read books appropriate to their reading levels. Parents and students did not regularly use the library as a way to access books as students from the gain‐group did. Only two students had visited the library in summer. Students mostly read alone and not many shared reading practices were identified.
Students who lost during summer typically showed mixed reactions to reading and had not regularly engaged in reading activities over summer. Parents usually were somewhat knowledgeable about what their children were reading, but did not seem involved in choosing materials. Some parents and students named strategies to identify reading materials; however, these were mostly not effective (e.g., looking for thickness of the book or size of the letters). Students tended to read books not appropriate to their reading levels. Parents only identified two ways to access reading materials: buying books or inheriting them from siblings or the wider family. The library seemed not to be used regularly and only one student had visited the library in summer. Students in this group had mostly read alone, and mixed accounts were given when asked if someone read to the child.
Writing
Indicative patterns emerged in families' writing practices for the profile groups. Differences usually lay in students' appreciation of writing which were associated with the frequency with which students engaged in writing activities, the kind of writing activities students engaged in, parental guidance and parental writing. (Table 6)
| Profile groups | |||
|---|---|---|---|
| Theme | Gain | Stall | Drop |
| Appreciation of writing | Liked writing, agreement between parents and students | No appreciation of writing, agreement between parents and students | Mixed appreciation of writing, prone to change or writing activities restricted to homework, mixed agreement between parents and students |
| Writing mileage | Two to three times a week | Every few days or once a week | Seldom or sometimes |
| Writing activities | Communicative writing (letters, postcards and emails) or activities that offered personal meaning and enjoyment (diary, superscribing drawings, jokes and stories) | Communicative writing (postcard) or activities that offered personal meaning and enjoyment (diary and writing stories) | Mostly school‐related writing task (used school materials) |
| Shared writing practices | Students wrote mostly by themselves | Students wrote mostly by themselves | Students wrote mostly by themselves |
| Parental guidance | Students asked their parents to correct their work or asked questions, no encouragement | No encouragement, parents corrected students' writing | Parents encouraged students and corrected their writing, some parents gave ideas for writing |
| Parental writing practices | Mixed reports about their writing, mostly wrote two or three times a week, mostly emails | Disliked writing, only wrote seldom, mostly emails | Mixed reports about their writing, parents, who wrote daily, referred to work related writing activities |
| Parental perception of importance of writing | Important for the language development, developing an understanding and perception of the world | Important for later work context, to be able to express themselves and for language development | An important skill in regard to later employment |
Students who gained during summer tended to show a general appreciation of writing and usually engaged in writing activities frequently over summer. They mainly engaged in communicative writing (letters and postcards) or in writing activities that offered personal meaning or enjoyment (stories, jokes and diary). Parents hardly provided personal encouragement for their children to write; however, students seemed self‐determined about writing asking their parents questions about writing or to correct their work. Parents partly liked writing, but would not engage in writing activities daily.
Students who stalled during summer did not usually show an appreciation of writing. They seemed to engage in writing activities only every few days (copying text, writing postcards and stories). Students tended to write by themselves and parents seemed not to encourage or help them. Parents in this group disliked writing and only seldom engaged in writing activities.
Students who lost during summer tended to show mostly negative reactions to writing. In several cases students and parents' answers to the question if students liked writing contradicted each other. Students in this group had only sometimes or seldom engaged in writing activities over summer. Often writing activities were school related and involved school materials. Most parents corrected their children's writing and some encouraged them to write. Parents in this group tended to report regular writing activities, which were work related.
Discussion
This study adds to the existing body of research by investigating the effects of the school break over summer on children's literacy learning in Germany, which is – with only 6 weeks – one of the shortest summer breaks in Europe. Our results suggest that a SLE is evident in German primary schools for both reading comprehension and writing. Students generally progressed considerably during school periods and stalled or lost in achievement over summer. The overall effects are thus similar to those reported for literacy in North America (Downey et al., 2004; Entwisle, Alexander, & Olson, 1997; Heyns, 1978; Skibbe et al., 2012) and Belgium (Verachtert et al., 2009).
Summer effects were overall more pronounced in writing than for reading, with a statistically significant drop in writing achievement. Not only did students at both schools drop in writing over summer, gains over school periods were also less rapid than in reading. This indicates a somewhat more general problem about the sustainability of writing development. Students seemed less able to retain writing skills over summer than to retain reading comprehension skills and typically struggled more to recover from the drop experienced over summer.
The findings point to differences in family practices which increase the susceptibility of writing to summer effects. Students in this study did engage in less writing activities overall over summer and parents seemed less focused on motivating and scaffolding writing activities over summer for their children. Parents' perception of writing being a ‘school task’ and not an enjoyable pastime might have influenced their approach to writing. As Pajares and Valiante (1997) point out, children's own motivation and self‐efficacy in regard to writing are significant factors in the engagement and in the prediction of writing performance in general, and as shown here, also over summer. In contrast, students and families engaged more regularly in reading activities over summer and parents typically engaged in active scaffolding of children's reading experiences through helping to choose materials and by taking part in the literacy activity themselves.
Factors that could be associated with differential progress over summer were the language spoken at home and SES. Interestingly, the language spoken at home had stronger effects on progress rates during school periods than SES. However, in regard to summer effects, the language spoken at home made no significant difference to summer learning, while SES made a significant difference to summer learning in reading, but not in writing. On a positive note, students not speaking German at home tended to make considerable gains in reading and modest gains in writing during the second period of schooling compared with German speaking students. In other words, they tended to catch up to the German speaking students, who had a significantly higher baseline achievement. The same phenomenon was observed for low‐income and high‐income students for reading, but not for writing.
These results suggest, using a socio‐cultural and ecological lens, that the microsystems provided by the school can be effective in closing the achievement gap between certain groups of students, but that the summer can mitigate this success. Thus, examining practices in a microsystem that play a greater role over the summer – in students' homes – gives an important insight in what mechanisms support or hinder learning over summer. Differences between the subject areas of reading and writing were apparent. But what is indicated is that the differences often attributed to SES are to be found in quite discrete family practices and these vary both within and between SES groups. For example, a prevalence of certain literacy practices in families' homes with different SES was not found – a finding congruent with earlier research in the emergent literacy field (Adams, 1999; Purcell‐Gates, 2000). But some home literacy practices of students were associated with different summer learning patterns, suggesting that certain practices had a moderating impact on summer learning. These were family practices that emphasised the enjoyment of reading and writing activities and integrated them in (daily) routines, for example: regular visits to the library, reading to each other, writing shopping lists together and writing postcards and notes. Furthermore, parents knowing effective strategies to judge their child's reading and writing ability and hence choosing appropriate materials was important for children to enjoy those activities and gain from them academically. Access to reading materials was usually sought through regular visits to the library, explaining why the widely used proxy variable of ‘books at home’ was not a strong predictor for summer learning in the present study (see also Park, 2008).
The findings of the present study suggest that certain literacy practices tend to support summer learning. These practices or proximal processes are not dependent on a family's socioeconomic standing but on a family's approach to literacy practices, although they may on average be associated with SES. Important are positive, frequent practices that become increasingly more complex and are reciprocal; which are also characterised by some degree of engagement from the child (Bronfenbrenner & Morris, 2006). The same principles apply for school literacy practices. Schools can play an important role in fostering a love for literacy activities that are not deemed to be just ‘school tasks’ but relate to a child's interest. Furthermore, literacy practices that take place at home and in school, the two most immediate microsystems of children, need to be connected, especially in the primary school age group. In this way, practices and processes are affirmed and reinforced. Strong home and school relationships can help build more coherent literacy practices across both environments resulting in more frequent engagement by students.
These findings have implications for future interventions aimed at summer learning. A socio‐cultural and ecological framework with a focus on family literacy practices – their enjoyment, their frequency and their reciprocity – may be better suited to comprehend and support effective literacy practices at home, as it goes beyond SES labels to identify salient education contexts. It further prompts researchers to consider simultaneous influences from multiple sources, or settings of a child's development, and their inter‐relationship (Bronfenbrenner, 1979; Bronfenbrenner & Morris, 2006). Future research could take into consideration the dispositions (e.g., attitude, motivation and efficacy) and resources of the people involved in a system. While dispositions can activate proximal processes in particular developmental domains, different resources or level of resources (e.g., ability, experience, skill and knowledge) might be required at different developmental stages (Bronfenbrenner & Morris, 2006).
The present study has limitations because of its size and time span. The study focused on a relatively small sample from one city. Thus, the findings should be interpreted cautiously as there is limited scope for generalisability. Additionally, this study only investigated a single instance of summer; it is possible that this summer could be unique and not entirely comparable to other summers. Certainly, replicating the results with a larger, nationally representative sample over multiple years would yield greater analytic capabilities and generalisability. Further research is also needed in different subject areas to depict an even more complete picture of summer learning patterns. Furthermore, it would be of interest to compare the use of different testing dates to depict the SLE. Another factor that may be important, but which could not be fully addressed in this study, is the quality of the parent–child literacy interactions as the interviews and literacy logbooks could only capture the views of parents and students, and of a small sample. However, the study attempted to collect rich data and triangulate those data by interviewing parents and students separately. Furthermore, data from literacy logbooks did not yield any significant results in the analysis possibly because of the sample size. Literacy logbooks can be used to record daily literacy activities and remove the bias prevalent in retrospective measures; thus, it would be desirable to use them as a measure with a greater sample in future research.
Acknowledgements
We are especially grateful to the teachers, parents and students for their willingness to participate in this research. The research was supported by funding from The University of Auckland International Doctoral Scholarship. The completion of this paper was assisted by a writing award from the School of Curriculum and Pedagogy, University of Auckland.
Biographies
Frauke Meyer is a postdoctoral research fellow with the Leadership Research Group at the Faculty of Education, The University of Auckland. Her research interests lie in the fields of literacy, educational leadership and teacher education. The paper presents the findings from her PhD research.
Kane Meissel is a Lecturer in the School of Learning, Development and Professional Practice, within the Faculty of Education at the University of Auckland, specialising in quantitative research methodology. He is also interested in professional development, schooling improvement and evaluation, as well as literacy and second language acquisition.
Stuart McNaughton is the Director of the Woolf Fisher Research Centre at the University of Auckland. His research interests are in literacy and language development, the design of effective educational programmes for culturally and linguistically diverse communities and cultural processes in development. Publications include research on development in family and school settings, instructional design and intervention models in large‐scale interventions.
Number of times cited: 3
- Jo Fletcher, Supporting and encouraging young adolescents in New Zealand to be effective readers, Educational Review, (1), (2017).
- Jackie Shinwell and Margaret Anne Defeyter, Investigation of Summer Learning Loss in the UK—Implications for Holiday Club Provision, Frontiers in Public Health, 5, (2017).
- Rachel Williamson and Rebecca Jesson, Log on and blog, English Teaching: Practice & Critique, 16, 2, (222), (2017).
