From Early Starters to Late Finishers? A Longitudinal Study of Early Foreign Language Learning in School
We would like to thank the four anonymous reviewers for their constructive comments on our manuscript. Furthermore, we would like to thank Dr. Dominik Rumlich for his insightful and helpful feedback. The research reported in this article was collected as part of the Ganz In—All‐Day Schools for a Brighter Future project funded by the Mercator Stiftung. Ganz In is a joint research project of the TU Dortmund University, Ruhr‐University Bochum, and the University of Duisburg‐Essen.
Abstract
Foreign language education has now been implemented at the elementary school level across Europe, and early foreign language education has gained traction following language policies set by the European Commission. The long‐term effects of an early start, however, have not received ample scientific scrutiny. The present study assessed early receptive skills of two cohorts of English language learners in Year 5 (beginning of secondary education in Germany) and two years later in Year 7. The factors distinguishing between these two cohorts were onset of foreign language education and the amount of language exposure. The effects of the earlier start were found in the results for Year 5, when the early cohort outperformed peers with less and later exposure to English. However, in Year 7, the late starters surpassed their early starting peers.
Introduction
Foreign language learning in pre‐ and elementary schools has seen a rapid development in Europe. Parents, educators, and politicians have long been advocating for early foreign language education in mainstream schooling, believing in the credo “the earlier, the better.” This movement's impact can be recognized in Europe's aim of fostering a plurilingual, multicultural society throughout the continent. The “2+1” policy, for example, is an ambitious agenda to promote language learning to the extent that every European is fluent in at least two languages in addition to their mother tongue (Council of the European Union, 2002; European Commission, 1995). In the hope of achieving this bold goal, early foreign language education has been identified as a potentially viable approach (Council of the European Union, 1997). Consequently, across Europe, foreign language education in elementary school has become the rule rather than the exception, with a total enrollment of early language learners surpassing 78% in 2010 (Education, Audiovisual & Culture Executive Agency, 2012). As a result, research into early foreign language education has been growing steadily (e.g., Larson‐Hall, 2008; Muñoz, 2006; Pfenninger, 2014b). This surge of research interest can, at least partially, be attributed to the European Union's (EU) language agenda. Countries throughout Europe have been adapting their language policies following EU encouragement. Primary aims are the promotion of multilingualism and multicultural understanding as well as a response to the growing demand of fluent second language (L2) speakers for the job market.
The principal research objective of the present study was to investigate the sustainability of an early onset of instruction for English as a Foreign Language (EFL) in Year 1 (age 6–7 years) as opposed to Year 3 (age 8–9 years) and its impact on students’ English language proficiency in Years 5 and 7. The study investigated the effects of moving EFL education from Year 3 to Year 1 of elementary school in Germany. Two cohorts of students participated in the study: one cohort of early starters (ES; age 6–7 years) beginning EFL in the second half of the first school year and one cohort of late starters (LS: age 8–9 years) commencing EFL in Year 3. Additionally, individual student characteristics (i.e., gender, parents’ socioeconomic status, cultural capital, home language, and cognitive abilities) were included as background variables in statistical analyses to determine their impact on students’ language proficiency. Two crucial, inextricably linked factors are central to the present study: learners’ age at the time of onset of EFL teaching and their amount of exposure to English. As an earlier start results in more English lessons, it is often impossible to separate the effects either measure has on language proficiency.
Early Foreign Language Learning: Brief Background and Research Perspectives
The assumption that the earlier language learning starts, the better the results will be still prevails in the minds of many parents and policymakers (European Commission, 2011). While the roots of this belief are manifold, a certain disconnect between second language acquisition (SLA) research and early foreign language advocates is evident: Either research does not inform policy making well enough, or it has been misinterpreted. DeKeyser and Larson‐Hall (2005) argue that “[i]n the practical realm, the younger is better argument has been both used and abused, both refuted and misunderstood by advocates of early intervention from the very beginning of formal immersion education to this day” (p. 88).
The Age Issue in the School Context
Research into early foreign language learning, more precisely the age factor, has experienced a significant surge in recent years (de Bot, 2014; Enever, 2011; Muñoz, 2006; Pfenninger & Singleton, 2016). Before, much of what was known about L2 acquisition at an early age was based on research from language acquisition at home or through immersion programs. These findings have been applied to the context of learning a language in preschool or elementary school despite considerable differences, particularly regarding the amount of exposure to the L2, which is often limited in school contexts, compared to immersion environments (Muñoz, 2010). Therefore, in the context of early foreign language research, a crucial distinction needs to be made between age of onset and amount of exposure. When comparing ES with LS with a focus on age of onset, one needs to take into consideration that the amounts of exposure to the L2 for ES and LS will necessarily differ considerably and will thus impact study results and their interpretation. Unfortunately, research in the field of early foreign language education has often neglected this distinction in research designs and analyses, as discussed by Muñoz (2006).
The benefits propagated for moving early L2 learning into elementary schools or preschools are manifold. The increased exposure at this early age is expected to lead to linguistic benefits in terms of higher language proficiency and potentially more nativelike pronunciation (Flege & MacKay, 2011), advantages in academic achievement (Taylor & Lafayette, 2010), and intercultural advantages (Nikolov & Mihaljević Djigunović, 2011), as well as moderate advantages for phonological and morphosyntactic achievement (Larson‐Hall, 2008), which may, however, disappear quickly (Pfenninger, 2014a). Furthermore, younger learners show fewer problems with language anxiety (Johnstone, 2009), higher levels of motivation and positive attitudes toward language learning (Börner, Engel, & Groot‐Wilken, 2013; Graham, Courtney, Tonkyn, & Marinis, 2016; Mihaljević Djigunović & Lopriore, 2011), and potentially better employability in the future. These attributed benefits are, however, not always clear due to differences across studies in the age of students, contextual differences, or a lack of rigorous research.
In a review of SLA research in the 1960s and 1970s, Krashen, Long, and Scarcella (1979) concluded that research had consistently pointed to advantages for older learners, adults or children, who acquired language at a faster rate, while onset of L2 learning during early childhood led to higher proficiency. Amount of exposure trumped age of onset, although ES were able to catch up relatively quickly. More recent studies have provided converging evidence that older learners (12 years and older), particularly in the beginning stages, consistently learn at a faster rate than younger learners (Mihaljević Djigunović, Nikolov, & Otto, 2008; Muñoz, 2008; Pfenninger, 2014a). In her longitudinal project, Muñoz (2006), for example, comprehensively assessed L2 development with different groups of language learners who received from 200 to 726 hours of language instruction. Groups differed in the age of onset (i.e., 10–12, 12–14, 14–18 years, and adults). The later in life learners started (i.e., as adolescents or adults), the more progress they made initially after 200 hours. It took the youngest group of learners (10–12 years) 726 hours to catch up to the older learners without outperforming them (Muñoz, 2006). Based on these findings, it has been argued that the level of cognitive maturity favors older learners (Cummins, 1981), especially regarding teaching methods based on explicit instruction applied in secondary education (Muñoz, 2006). This advantage of older learners over younger learners has been attributed to their ability to learn explicitly—that is, make use of their metalinguistic knowledge, such as applying rules (Pfenninger & Singleton, 2016)—and their more advanced native language (L1) literacy and oracy skills while benefiting from the cognitive advantages of a more mature brain, which has implications for test‐taking strategies. Once younger learners’ cognitive development catches up to that of older learners, younger learners close the proficiency gap. To achieve similar levels of language proficiency, younger learners would thus require an increased amount of exposure to achieve what older learners accomplish in a shorter time. In sum, under nonimmersive conditions or without increased exposure in school environments, amount of exposure seems to be more important than time of onset (Curtain, 2000; Muñoz, 2006; Steinlen, Håkansson, Housen, & Schelletter, 2010; Unsworth, Persson, Prins, & de Bot, 2015).
An early onset of foreign language education potentially increases both length and overall amount of exposure to the L2. Current curricula, however, do not always reflect these results. The curriculum of North‐Rhine‐Westphalia (NRW), Germany, for example, moved EFL to the second term of Year 1 with two 45‐minute lessons per week, while the number of lessons in grammar (secondary) schools was reduced by one lesson per week to four for Years 5 and 6 (MSW–NRW, 2015b; Verband Bildung und Erziehung Landesverband NRW, 2006). By moving EFL into Year 1, the overall number of hours of EFL across elementary and grammar school was roughly retained. In light of the available research, this adjustment, however, cannot be viewed as a serious sign of interest in promoting EFL education. As described above, research into early foreign language learning has shown that to achieve equal or higher levels of language proficiency, younger learners, compared to learners that start later, require longer and more intense exposure. This means that minimal input of an hour or two per week does not suffice (DeKeyser & Larson‐Hall, 2005; Lightbown & Spada, 2006). Thus, traditional foreign language education in school can hardly provide the required contact time with a L2.
Implementation of EFL
Early language learning is often implemented through methodology focused on implicit learning, which is typically understood as playful acquisition of the L2 through meaningful exposure and communicative activities, contextualized language practice, and an absence of metalanguage (Housen & Pierrard, 2005). Younger learners require more exposure to target language structures “to infer rules without awareness” and “internalize the underlying rule/pattern without their attention being explicitly focused on it” (Ellis, 2009, p. 16; Mihaljević Djigunović et al., 2008; Muñoz, 2008). With the transition to grammar (secondary) school, methodology shifts from a stronger emphasis on implicit learning to explicit learning and teaching, that is, the overt teaching and learning of the L2, such as grammar or vocabulary and the use of metalinguistic skills (Housen & Pierrard, 2005). Metalanguage slowly builds in students’ L1s during elementary school, and thus students cannot yet benefit from a transfer to the L2. Explicit and implicit learning rely on different neural systems in the brain; the latter requires significantly fewer cognitive resources (R. Ellis, 2009; N. C. Ellis, 2011). In addition, although we acquire our L1 implicitly, it likely takes considerably longer to learn to communicate in a L1, compared to a classroom‐based L2 (Birdsong, 2006).
Methodological recommendations in elementary school curricula that are in line with the idea of implicit learning have received criticism within Germany, specifically Baden‐Württemberg. For example, Schmelter has criticized policymakers’ understanding that L2 learning should resemble L1 acquisition (Schmelter, 2010; Standing Conference of the Ministers of Education, 2013). The allocated time of around 90 minutes per week can hardly be compared to the amount of exposure in L1 acquisition. L1 acquisition processes are characterized by consistent exposure, frequent repetition, and supportive one‐to‐one settings in the family, as well as community support and interaction with peers in preschool. These circumstances cannot be compared to L2 acquisition in an institutional setting with limited exposure to the L2 provided by nonnative teachers. Furthermore, while methodology may build on students’ L1 acquisition experience, the L2 does not necessarily have the same instrumental significance for students (Nicholas & Lightbown, 2008). To benefit from an early start, more exposure to the L2 in a meaningful context would be required. The cognition hypothesis argues that increasing the cognitive demands of tasks will foster interaction and “push learners to greater accuracy and complexity of L2 production” (Robinson, 2003, p. 45; Robinson, 2007). Thus, immersion or content‐based instruction could provide more intensive learning contexts, while the increase in cognitive demands might offer the added benefit of processing the L2 more deeply (i.e., through elaboration and rehearsal linking language to content, images, or in‐depth analyses), particularly as these approaches rely on implicit learning (de Graaf & Housen, 2009).
A further point of criticism involves the hasty implementation of foreign language education into the elementary school curriculum without a sufficient number of qualified teachers (Edelenbos, Johnstone, & Kubanek, 2006; Piske, 2013). The few research studies available have highlighted that the more qualified and proficient EFL teachers in elementary schools are, the higher students score (de Bot, 2014; May, 2007; Unsworth et al., 2015). Elementary school teachers are “the most important stakeholders,” but research on their abilities, level of training, methodological knowledge, and language proficiency remains scarce (Nikolov & Djigunović, 2006).
Despite the criticisms, adjustments can help ensure success in language learning during and beyond elementary education. First, the amount of classroom exposure should be increased beyond an hour or two per week (Larson‐Hall, 2008; Muñoz, 2008) and should include enhanced L2 exposure to extramural English outside of school, for example, through listening to music, reading books, or watching movies in the L2 (Mihaljević Djigunović et al., 2008). Second, ensuring teachers’ language proficiency and language‐specific training for younger student populations is imperative (Marinova‐Todd, Marshall, & Snow, 2000; Mihaljević Djigunović et al., 2008). Lastly, facilitating learners’ smooth transition into early foreign language education (e.g., by acknowledging and building on learners’ L1 skills), adjusting methodology to that used in elementary school, and communicating with elementary school teachers from elementary to secondary education will significantly improve early language education (e.g., Kolb & Mayer, 2010; Marinova‐Todd et al., 2000).
Reading and Listening in Early Foreign Language Learning
The present study focused on receptive language skills (i.e., listening and reading). Listening skills are particularly essential, as they constitute the primary source of language input in the communicative language classroom. Elementary school students who learn implicitly need ample input to master this complex skillset that requires neurological, linguistic, semantic, and pragmatic processes to work interdependently to decode spoken language (Rost, 2011). To transform sound to meaning, learners need to rely on both bottom‐up processes, which include segmentation of the speech stream into meaningful units, and top‐down processes, which require the use of established knowledge and understanding of context to infer meaning. With increased levels of language proficiency and a growing body of knowledge, listening becomes progressively more automated (Vandergrift & Goh, 2012). Although L2 reading is only gradually introduced in early foreign language learning (MSW–NRW, 2008), it promotes overall language development, especially phonological and phonemic awareness, thus linking listening to reading and, ultimately, writing skills (Dlugosz, 2000) and orthography. Similar to listening, reading is an interactive process involving various simultaneous top‐down processes, such as integrating knowledge and expectations, and bottom‐up processes, such as decoding letters and words (Birch, 2015; Grabe & Stoller, 2011).
Learner Characteristics and Their Relationship to L2 Proficiency in the School Context
The second focus of this study was on the investigation of individual learner characteristics, such as students’ gender, their cognitive abilities, home language, parents’ socioeconomic status (SES), and cultural capital, as well as their contribution to L2 proficiency in the context of early EFL education. For example, in large‐scale assessments, such as those carried out within the Program for International Student Assessment (PISA), individual differences in students’ gender or their parents’ SES have often been shown to have significant repercussions for students’ academic achievement (Ehmke & Jude, 2010; Organisation for Economic Co‐operation and Development, 2015; Sälzer, Reiss, Schiepe‐Tiska, Prenzel, & Heinze, 2013). Disentangling the age factor from other individual difference variables, such as motivation, attitudes, or environmental variables (e.g., teaching quality or teacher proficiency), poses a considerable challenge (see Pfenninger & Singleton, 2016). Despite their general importance for learning across different subjects, research on foreign language acquisition has often failed to incorporate individual difference variables (Csapó & Nikolov, 2009; Nikolov & Djigunović, 2006). This might lead to increased margins of error and overestimation of effects attributable to the variables under scrutiny.
Our understanding of gender differences in SLA is often based on “common knowledge” (van der Slik, van Hout, & Schepens, 2015, p. 1) and the “widespread belief… that females tend to be better L2 learners than males” (Saville‐Troike, 2006, p. 90). Although focusing on adult learners, van der Slik et al.’s investigation of gender differences in SLA among immigrants from 88 countries who speak 49 mother tongues in the Netherlands (N = 27,119) has contributed valuable insight to the discussion of gender differences in L2 learning. Women's speaking and writing skills consistently surpassed the scores achieved by men, who showed slight advantages in reading scores, with no differences reported for listening comprehension. Studies on EFL attainment in either elementary or secondary education in Germany generally tend to confirm the widespread belief of a gender advantage, even for younger learners. For instance, the large‐scale study KESS 4 (Kompetenzen und Einstellungen von Schülerinnen und Schülern [Competences and Attitudes of Students]) focused on English proficiency just before students’ transition to secondary education after Year 4 in Hamburg, Germany. For listening comprehension, scores were slightly higher for girls than boys (May, 2007). In the follow‐up studies, KESS 8 and 10/11 in Years 8, 10, and 11 in secondary school, girls maintained their lead over boys in listening comprehension and outperformed boys in general language proficiency assessed through literacy‐based C‐Tests (Nikolova, 2011). The DESI (Deutsch‐Englisch‐Schülerleistungen‐International [Assessment of Student Achievements in German and English as a Foreign Language]) study (N = 10,543) comprehensively evaluated Year 9 students’ English skills across secondary school types in Germany, with boys performing almost as well as girls in listening comprehension and girls performing significantly better than boys in reading and particularly in writing (Hartig & Jude, 2008). These findings of girls’ superior language skills, especially concerning literacy‐based tests, are corroborated by other large‐scale studies (Courtney, Graham, Tonkyn, & Marinis, 2015; Hartig & Jude, 2008; Jaekel, 2015; Nikolova & Ivanov, 2010; Rumlich, 2016). For speaking skills, however, slight advantages have been reported for boys regarding pronunciation, fluency, and sentence structure (Nold & Rossa, 2008).
Home language constitutes another key variable in SLA. Teachers today need to cope with cultural and linguistic heterogeneity. Research targeting the impact of nonmajority L1 (e.g., Turkish, Russian, or Slovak in the context of Germany) on the learning of a foreign language in school, which would be a third language (L3) for learners studying in L2‐medium contexts, is growing slowly. However, the ability to speak two or more languages alone does not seem to positively impact L3 acquisition, whereas biliteracy in L1 and L2 does (Rauch, 2014; Rauch, Naumann, & Jude, 2012). Following the threshold hypothesis (Cummins, 1976), the (very) early start of foreign language learning may be an obstacle in L2 development for learners with limited L1 proficiency. For example, for English listening comprehension in Year 4 in Hamburg, no significant differences were found between majority and nonmajority students who have at least one parent that was born in Germany, whereas students with two foreign‐born parents scored significantly lower (May, 2007). These results were also confirmed for receptive English proficiency in the Netherlands for learners aged 4 versus 8–9 (de Bot, 2014). In Switzerland, Pfenninger (2014a) drew attention to the potential struggle learners have to face if their L1 is neither German nor French, which means that, for most students, English acquired in school will be their fourth language. Other studies have found non‐L1 students in Year 8 (ages 13–14) to achieve significantly lower general language proficiency scores (measured with literacy‐based C‐Tests), compared to the majority of L1 students (Nikolova & Ivanov, 2010).1 Although Jaekel (2015) found no significant differences between majority and nonmajority students in terms of their general English language proficiency measured through C‐Tests, non‐German L1 students reported lower grades than their German L1 peers. As a consequence, for the German state of Baden‐Württemberg, Baumert et al. (2011) have argued for moving early EFL teaching from Year 1 to Year 3 to allow more time for German language development for students whose L1 is not German.
SES has been consistently identified as a key predictor of academic success. Studies have shown that lower‐SES children enter school with poorer school readiness skills, which in turn results in overall lower academic achievement and lower cognitive functioning scores (Blair & Raver, 2012). Along the same lines, the PISA studies have highlighted the influence of social variables, such as economic and cultural capital, on student attainment across the participating countries (Ehmke & Jude, 2010). Similarly, the impact of SES and other background variables, like the number of books at home, has been repeatedly reported in the Trends in International Mathematics and Science Study (TIMSS) and Progress in International Reading Literacy Study (PIRLS) (Gustafsson, Hansen, & Rosén, 2013). SES has been shown to have a strong impact on L2 achievement in Germany (May, 2007; Nikolova & Ivanov, 2010), a result that mirrors effects of SES on general academic achievement (Klieme, 2010; Muñoz, 2007). These findings are understood to be due to early streaming of students for secondary school after Year 4 of elementary school and to the tiered structure of secondary education in Germany, described below (Muñoz, 2007).
The cultural capital an individual or family has at their disposal, for instance, in terms of the number of books and the level of reading enjoyment, as well as museum and concert visits, has been shown to benefit students’ academic achievement (Bourdieu & Passeron, 1990; Gräsel, Göbel, & Stark, 2007). Indeed, research on the effects of cultural capital on academic success has, to a large extent, confirmed their positive relationship (e.g., see Jæger, 2011, for a detailed discussion). Although students’ L1 has been shown to benefit from students’ cultural capital in the form of books, reading habits, or enjoyment of reading (e.g., Georg, 2004) and despite the fact that large‐scale studies that include L1 literacy measures regularly target different measures of cultural capital, SLA research rarely considers cultural capital as a variable.
Last but not least, learners’ cognitive abilities have been associated with their academic achievement in general and have also been identified as a key variable in SLA (e.g., Dallinger, 2015; Genesee, 1976; Jaekel, 2015; Skehan, 1998; Sparks, Patton, & Ganschow, 2012). In this context, it is important to distinguish between aptitude and intelligence. Both dimensions are closely related; however, the former is a domain‐specific measure of ability, whereas the latter is general in nature and is independent of experience or knowledge (Biedroń, 2011). Beyond language learning, intelligence appears to be highly relevant for test taking (Muñoz, 2008).
The Present Study
The present study compared the EFL proficiency of LS, age 8–9 years (beginning of Year 3 at elementary school), with ES, age 6–7 years (second half of Year 1), who had 1.5 years of additional language exposure. The study specifically focused on L2 English reading and listening comprehension scores for both groups at the beginning of Year 5 (Time 1), which corresponded to 2 years (or 140 hours) of English language instruction for the LS group and to 3.5 years (or 245 hours) of instruction for the ES group, and at the beginning of Year 7 (Time 2), which coincided with 4 years (444 hours) of instruction for the LS group and with 5.5 years (549 hours) of instruction for the ES group. Because (as discussed previously) individual differences in learners’ gender, SES, cultural capital, cognitive abilities, and home language may significantly affect foreign language proficiency, the current study also targeted potential effects of these variables on the performance of the LS and ES groups in reading and listening comprehension. The following research questions guided this study:
-
Do ES EFL students in early English language education outperform LS EFL students in language proficiency over time?
-
What impact do individual differences (gender, SES, cultural capital, IQ, and home language) have on students’ language proficiency?
Method
Research Context
Germany recently introduced early foreign language education into elementary schools. In most states in Germany, after 4 years of elementary school, students are streamed into secondary schools based on teacher ratings (Ditton & Krüsken, 2006). The three‐tiered secondary school system consists of Haupt‐ (lower secondary) and Realschule (middle secondary), which offer a 6‐year middle school degree, and Gymnasium (grammar school/upper secondary) or Gesamtschule (comprehensive school), which offer 8‐ and 9‐year high school degrees, providing access to tertiary education. Most states introduce foreign language learning in Year 3, while six states have opted for a start in Year 1 (Standing Conference of the Ministers of Education, 2013). NRW, for example, moved the start of EFL into Year 1 in 2008 after its initial inclusion into the elementary school curriculum in 2003 (MSW–NRW, 2008). Despite the differences, the majority of states in Germany aim for an A1 proficiency at the end of elementary school in Year 4 for reading, writing, listening, speaking, and mediation (Standing Conference of the Ministers of Education, 2013), according to the Common European Framework of Reference for Languages (Council of Europe, 2001).
Teaching Methodology and Transition to Secondary School
Teaching methodology at this early age (6–7 years) significantly differs from that used in secondary education and needs to account for students’ levels of cognitive development as well as other learner characteristics, such as L1 proficiency and level of motivation. Oracy (i.e., listening and speaking) is emphasized especially in the earliest stages of learning while literacy (i.e., reading and writing) is targeted, for example, in word recognition through regular exposure (MSW–NRW, 2008). In general, early foreign language education relies heavily on communicative language learning, which aims to provide ample exposure and opportunities to use the L2. In the process of early foreign language learning, the ultimate success, which is defined as sustaining high levels of motivation and continuous development of language proficiency, hinges upon a successful transition from elementary to secondary education. However, this transition has been described as involving an abrupt shift (a) from implicit learning to explicit instruction that requires metalinguistic knowledge and (b) from an oracy‐focused curriculum to one that builds on literacy and steep grammatical progression (BIG‐Kreis, 2009). Furthermore, a lack of communication between elementary and secondary schools has been identified as a crucial problem (Thürmann, 2009). This is particularly important in harmonizing methodology and content in secondary EFL education.
Sample
The sample consisted of two cohorts of students (N = 5,130) from 31 grammar schools in NRW, Germany. Participating schools were distributed somewhat evenly across NRW so that rural and urban areas were considered to similar degrees. All 31 schools participated voluntarily in the longitudinal Ganz In—All‐Day Schools for a Brighter Future study endorsed by the Ministry of Education (MSW–NRW, 2015a). At elementary schools in NRW, English was introduced in Year 3 in 2003. In 2008, it was moved forward into the second half of Year 1. The LS cohort (N = 2,632), who started elementary school in 2006 and grammar school in 2010, had received 2 years (140 hours) of EFL teaching before they started grammar school. In contrast, the ES cohort (N = 2,498) had received 3.5 years (245 hours). This group started school in 2008 and moved on to grammar school in 2012, which corresponds to a lag of 2 years between the two sample groups and makes them the first cohort with additional EFL instruction in elementary school (105 additional hours spread over 1.5 years at the age of 6–7 in Years 1 and 2 of elementary school).2 Only students that took part in the proficiency tests in Year 5 and Year 7 were included in the analyses. Of the initial sample (N = 6,652, NLS = 3,312, NES = 3,340), 77.1% were retained and assessed longitudinally at the end of Years 5 and 7.
Table 1 displays a range of statistics summarizing differences between the two cohorts regarding individual learner characteristics at the beginning of Year 5 (with Cohen's d values calculated as indicators of effect sizes). As shown in this table, ES and LS did not differ significantly with regard to gender (p = .739), age (p = .182), cognitive ability (p = .821), and English grade at the end of Year 4 (p = .796). In Year 5, students were, on average, 10 years old, and classes consisted of slightly more boys than girls. Significant differences between both cohorts were revealed based on their reported mother tongue (p = .013), cultural capital (p = .018), and parental income (p = .034). In the LS group, fewer students reported having a non‐German L1 background than in the ES group. ES parents had a slightly higher income (MLS = 5.43 vs. MES = 5.58, p = .022), while LS households were characterized by a slightly higher cultural capital (books) (MLS = 5.51 vs. MES = 5.44, p = .042). However, the effect sizes for these differences were (very) small, indicating negligible effects. According to Wolf (1986), the threshold for Cohen's d magnitudes of academic relevance in educational research is .25. In the subsequent statistical analyses, these background variables were controlled for. In sum, the ES group differed from the LS group in that parents reported a higher income (economic capital), had fewer books (cultural capital), and had more students with a non‐German home language.
| Variable | LS cohort (n = 2,632) | ES cohort (n = 2,498) | Test statistic | p | Cohen's d |
|---|---|---|---|---|---|
| Gender (females) | 46.73% | 47.20% | χ2(1) = 0.11 | .739 | −0.009 |
| Mean age (years) | 10.37 (5.6) | 10.35 (5.3) | t(5,127) = –132.68 | .182 | 0.037 |
| Mother tongue (German only) | 75.87% | 72.80% | χ2(1) = 6.05 | .015 | 0.097 |
| Mean cognitive ability score (max = 25) | 17.75 (6.20) | 17.79 (6.00) | t(4,815) = –0.23 | .821 | −0.007 |
| Mean value for books (1–5)a | 3.51 (1.13) | 3.44 (1.12) | t(5,065) = 2.36 | .042 | 0.062 |
| Mean income value (1–8)b | 5.43 (2.17) | 5.58 (2.22) | t(3,673) = –2.12 | .022 | −0.068 |
| Mean English grade in Year 4 (1–6)c | 1.74 (0.60) | 1.73 (0.62) | t(4,963) = 0.80 | .796 | 0.016 |
- Notes. a1 = 0–10 books, 2 = 11–25 books, 3 = 26–100 books, 4 = 101–200 books, 5 = more than 200 books. b1 = less than €10,000, 2 = €10,000–19,999, 3 = €20,000–29,999, 4 = €30,000–39,999, 5 = €40,000–49,999, 6 = €50,000–59.999, 7 = €60,000–69,999€, and 8 = €70,000 or more. c1 = very good, 6 = insufficient.
Instruments
As part of this study, receptive English language skills (i.e., listening and reading comprehension) were assessed both at the beginning of Year 5 (Time 1) and Year 7 (Time 2). Standardized reading and listening tests were employed that had previously been validated in other large‐scale studies in Germany (see below). Year 5 receptive language proficiency was assessed based on the previously validated scales from the Evening study (Evaluation Englisch in der Grundschule [Evaluation of English as a Foreign Language in Primary School]), conducted by Engel and Ehlers (2013). For listening, students answered 28 multiple‐choice questions targeting picture recognition (17) and sentence completion (11) in German. For reading, 20 multiple‐choice and 4 open‐answer items assessed text understanding (α = .71). In Year 7, the selected scales were based on a statewide assessment scheme in Brandenburg (Vergleichsarbeiten der Jahrgangsstufe 6 – VERA 6 [Comparative Tests in Year 6]), developed by the Institut zur Qualitätssicherung im Bildungswesen (IQB [Institute for Educational Quality Improvement], 2014). For listening, 17 sentence‐completion (11 from the Year 5 assessment) and 11 multiple‐choice items were used to assess English listening (α = .79). For reading, students answered 11 multiple‐choice and 15 open‐answer items (α = .89), based on the materials from the IQB. Responses were coded dichotomously (correct vs. incorrect). The tests were identical for both cohorts in Year 5 and in Year 7. To obtain proficiency scores, a simple one‐dimensional logistic item response model (Rasch, 1980) was calculated, using weighted likelihood estimators (Warm, 1989). In this process, the item parameters were estimated before the person parameters. Items were checked for conformity by assessing item characteristic curves, discrimination parameters, mean squared errors (MNSQ), and their respective t values. The thresholds to determine acceptable item fit were chosen according to common guidelines used in large‐scale assessment studies (Adams & Wu, 2002; Wright & Linacre, 1994). Items that did not conform to the model were excluded before the main analyses. To obtain better estimates for the differences within cohorts, both Year 5 and Year 7 subsamples were scaled simultaneously. The final scores were standardized to a mean of 500 points with a standard deviation of 100.
Cognitive abilities were assessed with the Figural Analogy subtest of the Kognitiver Fähigkeitstest (Cognitive Abilities Test) by Heller and Perleth (2000), as it allows for an estimate of students’ general cognitive abilities independent of their L1 proficiency. In the present sample, the test reached good reliability in line with the norm‐sample values (αnorm = .94, αLS = .92, αES = .90). Demographic variables, including cultural capital and home language, were based on students’ responses in questionnaires. Regarding cultural capital, the students were asked how many books were present at their household. Five categories were offered: 0–10, 11–25, 26–100, 101–200, or more than 200. Additionally, five pictograms of bookshelves with an according number of books were depicted in the questionnaires. For home language, participants were asked which language they regarded as their mother tongue. Where data were not available from students because of missing responses, parental responses were used instead (when available). Income was based on parents’ responses. Parents were asked to classify the gross household income per year into the following eight categories: below €10,000, €10,000–19,999, €20,000–29,999, €30,000–39,999, €40,000–49,999, €50,000–59.999, €60,000–69,999€, and €70,000 or more. All scales originated from the international TIMSS and PIRLS assessments and were adapted for use in Germany (Bos, Bonsen, Gröhlich, Guill, & Scharenberg, 2009). There was no information on the children's ages for 0.9% of the students, L1s for 0.1% of the students, cognitive abilities for 6.1% of the students, books for 1.2% of the students, and the household income for 28.2% of the students.
Procedure
Student participation was voluntary; written consent was obtained from parents before data collection commenced. The tests were administered between weeks five and nine of the new school year. Data for the LS cohort were collected in the summers of 2010 and 2012 for Years 5 and 7, respectively, while the ES cohort's data were collected in 2012 and 2014. Data collection was conducted during regular school lessons. Data were analyzed through SPSS 23 (IBM Corp, 2015b), AMOS 23 (IBM Corp, 2015a), and ConQuest 3.0.1 (Adams, Wu, & Wilson, 2012).
Results
Do ES Outperform LS?
In Year 5 (see Figure 1), the ES participants significantly outperformed their LS peers in both reading (MES = 514.24, MLS = 486.49) and listening comprehension (MES = 517.24, MLS = 483.64). As shown in Table 2, the results of the independent‐samples t tests suggested the existence of statistically significant differences between the two cohorts’ language proficiency at the beginning of Year 5 for reading (p ≤ .001) and listening comprehension (p ≤ .001). The 1.5 years of additional EFL (or 105 extra hours) translated into a significant advantage of 27.5 points on average in reading and 33.6 points in listening comprehension, equaling about a third of a standard deviation. The resulting differences amounted to small‐to‐medium effect sizes (see Table 2).
| Language score | LS cohort (n = 2,632) | ES cohort (n = 2,498) | Test statistic | p | Cohen's d |
|---|---|---|---|---|---|
| Reading Year 5 | 486.49 (99.45) | 514.24 (98.61) | t(5,128) = –10.03 | < .001 | –0.280 |
| Reading Year 7 | 516.55 (97.59) | 482.57 (99.56) | t(5,128) = 12.34 | < .001 | 0.345 |
| Listening Year 5 | 483.64 (95.82) | 517.24 (101.43) | t(5,068) = –12.18 | < .001 | –0.341 |
| Listening Year 7 | 508.45 (96.86) | 491.1 (10.47) | t(5,068) = 6.23 | < .001 | 0.174 |
In Year 7 (see Figure 1), the LS cohort outscored the ES cohort in reading comprehension 516.55 to 482.57 and in listening comprehension 508.45 to 491.10. On average, LS participants scored almost 34 points higher in reading and 17.35 points higher in listening comprehension than their ES peers. As shown in Table 2, independent‐samples t tests confirmed the significant differences between the two groups for reading (p ≤ .001) and listening comprehension (p ≤ .001). In the 2 years of grammar school following the initial test, the LS were able to close the gap, and they outperformed their early‐starting peers.
What Impact Do Individual Differences Have on Language Proficiency?
To assess the complex relationships among various individual difference variables and their impact on language proficiency, structural equation modeling (SEM) was employed. SEM provides an analytic framework to observe single or multiple latent variables, their structures, and the relationships of those variables while considering measurement errors. A maximum likelihood estimator was used. To handle missing values, the full information maximum likelihood approach was applied (Lüdtke, Robitzsch, Trautwein, & Köller, 2007). The SEM included two latent variables consisting of receptive language performance in Years 5 and 7, respectively. A latent regression between the two latent variables defines the core of the structural model. Extraneous variables were added to assess their relative impact. Covariance between extraneous variables was allowed when appropriate. In addition to the dichotomous cohort variable, five background variables were chosen: gender, income (SES), books (cultural capital), home language (L1), and cognitive ability score. The use of cohort membership as a control variable allowed for the inspection of the relative importance of the latent variables. By including all covariates simultaneously, the familiar confounding structures between them could be controlled.
To assess model fit, we first considered the χ2 value of 234.37, which was highly significant (p ≤ .001; df = 25). This was to be expected considering the large sample size. Due to the significant χ2 value, other fit indices had to be considered (Byrne, 2010; Kim & Bentler, 2006). The comparative fit index (CFI) of .964 indicated that the model is fitting, as it lies above the .95 threshold (Hu & Bentler, 1999). A second fit index, the root mean square error of approximation (RMSEA), together with the PCLOSE, which tests the null hypothesis that RMSEA is less than .05, was also employed. A RMSEA of .04 and a PCLOSE of 1.0 corroborated a good fit of the SEM. The complete model is shown in Figure 2.
Gender, cognitive abilities, cultural and economic capital, and cohort contributed to explaining the variance of receptive language performance in Year 5 (see Figure 2). Cohort membership and cognitive abilities showed the largest effects on receptive language proficiency, such that belonging to the ES cohort and scoring highly on the cognitive abilities test had significant positive effects. Being female and having higher cultural and economic capital also significantly predicted higher test scores. The SEM accounted for 14% of the variance in language proficiency in Year 5. Home language (L1) did not have a significant effect on the receptive language proficiency in Year 5.
Receptive language proficiency in Year 7 received the largest contribution to explaining its variance from language proficiency in Year 5. Apart from Year 5 proficiency, cohort, cognitive abilities, gender, and home language also significantly predicted the receptive language proficiency in Year 7. Of these effects, the second strongest effect on receptive language proficiency in Year 7 came from the cohort variable. In agreement with the descriptive analyses discussed previously, the influence from Year 5 was negative, such that the LS cohort scored significantly higher in Year 7 compared to the ES cohort. The remaining variables contributed little to explain Year 7 receptive language proficiency. Girls and students who scored higher on the cognitive function test still performed slightly better than their peers. While home language had no impact in Year 5, a marginal effect could be observed in Year 7. SES measured in Year 5 (i.e., both cultural and economic capital) did not retain its predictive power with respect to language proficiency in Year 7. Overall, the SEM explained 49% of the variance in receptive language proficiency in Year 7. More variance could be explained for Year 7 than for Year 5 due to the Year 5 data serving as a predictor.
Discussion
The current results are in line with a growing body of research that confirms older learners to be at an advantage (in the long run) in learning a foreign language over students in early foreign language education with minimal input (Krashen et al., 1979; Larson‐Hall, 2008; Muñoz, 2006; Pfenninger, 2014b). Our findings suggested that the combination of an earlier onset and thus more exposure to the foreign language led to a positive short‐term effect for receptive language proficiency in L2 English, as demonstrated for the Year 5 data (see also Wilden, Porsch, & Ritter, 2013, for cross‐sectional analyses of Year 5 data). However, as demonstrated by the Year 7 data, LS not only closed the proficiency gap in both reading and listening comprehension, but also surpassed ES significantly, outperforming them in both reading and listening comprehension. Compared to the ES group, the LS group thus appeared to benefit more from the 2 years of learning English in grammar school. Many explanations offered in the research literature to account for these and similar findings may also apply to this study. Of these potential explanations, contextual factors—such as the amount of L2 instruction and students’ transition from elementary to secondary school, which includes a shift in teaching methodology—need to be considered in conjunction with students’ individual characteristics to account for these results.
Impact of Age on Language Proficiency
First, despite the common belief that younger learners are better language learners, research has consistently shown that older learners make faster progress in classroom language learning, potentially due to higher levels of cognitive maturity and their ability to learn languages through explicit instruction (Krashen et al., 1979; Muñoz, 2006). The LS cohort in the present study quickly caught up to the ES cohort and built a lead of its own. Muñoz argues that older learners, compared to younger ones, particularly benefit from the rule‐based and grammar‐reliant language teaching in the early phases of secondary education in EFL classroom environments. Our results thus align well with both recent studies (Muñoz, 2006; Pfenninger, 2014a, 2014b) and older analyses (Krashen et al., 1979). While other studies have included much more pronounced age gaps between their cohorts (e.g., Larson‐Hall, 2008; Muñoz, 2006; Pfenninger, 2014a), in the present study, this gap was only 2 years. As a result, differences between cohorts, for example, in terms of cognitive maturity, would be less conspicuous here than in prior studies.
Second, both cohorts received only minimal English language input of 90 minutes (usually two 45‐minute lessons) or less per week over the course of 3.5 years for the ES and 2 years for the LS groups. SLA research has shown that younger students require more L2 exposure to retain the language and benefit from an early start in the long term (DeKeyser & Larson‐Hall, 2005; Muñoz, 2008). While early language teaching methodology may draw on the playful character of the acquisition process, the minimal amount of instruction may have prevented students from benefiting fully from an early start to language teaching. Although exposure to English was ultimately doubled, transitioning from Year 4 to 5, the additional exposure to language, as spoken by the same body of EFL teachers, still did not allow the ES cohort to sustain their L2 proficiency lead. By contrast, for the LS cohort, their lead was sufficient enough to surpass their ES peers. The reduction of instructional time by one 45‐minute lesson per week (for a total of four 45‐minute lessons per week) at grammar school may be a crucial mistake—cognitive maturity, explicit instruction, and growing L1 literacy make L2 learning more effective at this age (Cummins, 1979; Muñoz, 2008). The additional exposure would also suggest that the ES cohort could have a head start that they could build on; otherwise, for these learners, exposure to the L2 in the early years (without an increase in exposure later on) may have been too sparse to have a longer‐lasting effect.
Third, a central issue in explaining the outcomes of this study and a crucial factor in ensuring successful early foreign language learning is the transition from elementary to secondary education, which has been identified as an Achilles’ heel of early EFL teaching (BIG‐Kreis, 2009). As grammar schools feed from different elementary schools, communication may not always be ideal to support students’ transition, which encompasses several critical issues. An abrupt shift from student‐centered, playful methodology to more teacher‐directed, academically oriented, and faster‐paced lesson rhythms may have impacted the ES cohort more as they experienced it for longer in elementary school. In addition, a potential mismatch of student–teacher expectations regarding methodology used in class may have caused a decrease in motivation (Courtney et al., 2015; Graham et al., 2016), particularly if students did not receive input that addressed their L2 proficiency level (e.g., challenging advanced students, helping others catch up), which may easily cause frustration. Just as students vary, so does the competence of elementary school language teachers (see Nikolov & Djigunović, 2006). Thus, teaching outcomes may differ from one elementary school to another. Additionally, EFL teachers at the elementary school level have, to a large extent, not been trained extensively as these programs are relatively new to tertiary education in Germany (Edelenbos et al., 2006). Last but not least, as grammar schools welcome students from several elementary schools, teachers have to be particularly thorough in assessing L2 skills that students have already attained. Teachers might have been unsure of what to expect from the new, more experienced student cohort or may not yet have adapted their lessons to the overall higher level of language proficiency, in an attempt to level the playing field by teaching to the middle. The latter has obvious repercussions for teaching in general and for student motivation more specifically. A significant advantage for the LS cohort, compared to ES students, may have been due to the fact that the LS students could apply their metacognitive knowledge and thus benefit from explicit language learning earlier in their language learning career. This might have given them a boost and possibly even added to their motivation, as this would have led to their faster progression through learning, feeding into their sense of self‐efficacy. The question that remains unanswered is whether teachers were ready to support and develop students’ actual language proficiency or if teachers’ business‐as‐usual approach to teaching clashed with their students’ expectations and hurt their motivation (e.g., see Courtney et al., 2015).
Finally, at both elementary and secondary school levels, teaching methodology has a crucial role in accommodating the special needs of early‐starting students, maximizing the outcome of minimal input in early language learning and allowing for a smooth transition between elementary and secondary schooling. This includes the availability of adequate textbooks, which should account for students’ increased language proficiency due to the early start. This is particularly important as EFL teaching across secondary schools in Germany relies heavily on textbooks (MSW–NRW, 2012). While no specific data regarding EFL methodology or textbooks were available as part of this project, we noted a mismatch in both Years 5 and 6 between the EFL methodology employed by teachers and the expectations of students, particularly those in the ES cohort, which may have discouraged learners, leading to a lack of motivation (Courtney et al., 2015). On the contrary, the LS cohort in this study exhibited what is now considered to be predicted patterns of advantages for older classroom language learners in school contexts, in line with the findings reported by Muñoz (2008) in Spain.
Learner Characteristics and Their Impact on Language Proficiency
In addition to the cohort variable, which had a significant impact on receptive language proficiency in SEM, students’ background variables also contributed significantly. The present study supported prior findings by showing that girls outscored boys in both English listening and reading comprehension in Years 5 and 7. The present analyses did not provide further insights, but explanations for better performance by girls have been demonstrated based on, for example, advantages in literacy (Courtney et al., 2015), higher interest in language learning (Helmke, Schrader, Wagner, Nold, & Schröder, 2008; Rumlich, 2016), higher levels of self‐efficacy (Jaekel, 2015; Mills, Pajares, & Herron, 2007), or higher motivation (Ruyffelaert & Hadermann, 2012). Ultimately, societal stereotypes of girls being better at language learning may weigh on boys’ self‐perceptions as L2 learners (Kissau, 2006).
Cultural and economic capital has been shown to be a consistent variable predicting success in school (e.g., Klieme, 2010). The two variables covering SES and cultural and economic capital each showed significant results for Year 5, but their impact faded over time. Grammar schools in Germany preselect their students based on academic achievement; the data are thus biased for SES, as there are significant differences with regard to SES and choice of school type in Germany (Ehmke & Jude, 2010). Nevertheless, for Year 5, significant differences for SES within classes affected student performance. The availability of more books at home can particularly be an indicator of larger L1 vocabulary and higher levels of L1 literacy, which would provide students with an advantage learning a L2 (Cummins, 1979). Among the central ideas of all‐day schooling in Germany was the mitigation of SES effects through additional teacher support; the present results may be an indicator of its success, as most upper secondary schools currently offer such a program in Germany (Holtappels, 2006). Another means of explaining the differences could be that students from lower SES backgrounds who have not been able to succeed in the cognitively taxing grammar school environment left the school and were thus excluded from our analyses. In the German school context, particularly in grammar schools, grade retention is still fairly common and may have affected the Year 7 data.
Cognitive abilities measured in Year 5 were the most significant predictor of L2 receptive language proficiency in Year 5 and remained significant for Year 7, albeit at a third of its initial impact. Despite using a nonlinguistic cognitive test instead of a test measuring verbal cognitive abilities, these findings support results of prior L2 research (Dallinger, 2015; Jaekel, 2015), namely, that cognitive abilities are a highly relevant variable in language learning. The quality of L2 teaching, clarity, student‐centered teaching, and overall levels of cognitive abilities in a particular class have been shown to moderate the importance of cognitive abilities in successful language learning (e.g., Helmke et al., 2008). Home language had no significant impact on L2 receptive skills in Year 5, but it gained significance in Year 7, in contrast to findings for secondary (Hesse, Göbel, & Hartig, 2008) and elementary (Steinlen, 2016) schools in Germany. Comparable results have been reported in the KESS 10/11 study, where the English proficiency divide between L1 German and non‐German students expanded over time (Nikolova, 2011).
Limitations
This study is part of a large‐scale research project with two cohorts of students encompassing 5,130 participants across 31 schools involving German, English, mathematics, and the natural sciences, all of which had limited time allotted for assessment. With these time constraints come certain drawbacks. For English and German, for example, it would have been more advantageous to assess verbal cognitive abilities instead of, or alongside, the nonlinguistic cognitive abilities test used here. Time constraints and the feasibility in school to test a large number of students in speaking skills are also a central reason for choosing receptive language proficiency. Additional individual difference variables, such as motivation, attitudes toward language learning, teacher proficiency, quality of teaching, and information regarding the transition from elementary to secondary school could have provided vital insights into the effect of age of onset, but were either not included in the study or would have gone beyond the scope of this article. Furthermore, the project's main scope was the investigation of students’ development at grammar schools, as these schools select or stream their students based on their academic achievement in elementary school; outcomes at other school tiers might vary significantly (e.g., see Klieme, 2010). Additionally, due to the setup of the study and the circumstances in both elementary and grammar schools, it is not possible to attribute the results definitively to either age of onset (Year 1 vs. Year 3) or amount of exposure (245 vs. 140 hours of EFL instruction), as both variables differed between the two cohorts.
The present study involved the first cohort of elementary school students who started English in Year 1. With the first and only chance to conduct this research at such a scale, but without any adjustment period for teachers, this naturalistic setting was unique, and several confounding variables could not be taken into consideration. The teachers and schools, for example, had little to no experience teaching (a) foreign languages to younger students in Year 1 and (b) more experienced EFL students in Year 5 at grammar schools. We would encourage validation of the outcomes of this study through another round of data collection in Years 5 and 7 with another cohort of students.
Recent studies targeting students’ language proficiency have regularly taken into account the variance that is explained by the clustered or hierarchical data structure, because potentially confounding effects at the levels of the classroom or the school might violate the assumption of stochastical independence and might lead not only to misspecification of standard errors but also to the amplification of the observed effects at the individual level (e.g., Lüdtke, Robitzsch, Trautwein, & Kunter, 2009). That the clustered structure was not controlled for must be seen as a drawback. The item response modeling did not take this into account because the main research question of the project focused on mean scores rather than individual students. For the SEM, the clustered structure of the data was not considered because the students were often reassigned at Year 7 so that the classes differed between Year 5 and Year 7. To justify these changes, schools reported, for example, the mandatory introduction of a second foreign language in Year 6, the beginning of Content and Language Integrated Learning (CLIL) streams in Year 7, or newly created focus classes. Here, multiple‐membership models would have been necessary, but these were deemed overly complex for the current research questions that addressed differences in language proficiency according to students’ age of EFL onset.
Conclusion
The present study adds to existing research that cautions not to blindly believe the myth that the earlier in life language learning commences, the better the outcomes of such learning will be. In Germany, the current practice of beginning EFL in Year 1 with 90 minutes or less of instructional time does not yield the expected results that policymakers may have hoped for, at least not in the long run. Extending EFL into elementary school did not seem to produce the anticipated linguistic outcomes. Admittedly, our study was conducted immediately after the introduction of EFL in Year 1 and some teething troubles (e.g., the lack of qualified English teachers in elementary school or outdated school books in secondary school) may have been resolved by now. However, the amount of language exposure in elementary or secondary school has not changed, which contradicts established research findings (Krashen et al., 1979; Larson‐Hall, 2008; Muñoz, 2008).
The field of early foreign language learning remains in dire need of empirical research particularly targeting teacher education, the transition from elementary to secondary education, and the use of textbooks. For example, the impact that elementary school teachers’ language proficiency and overall foreign language pedagogy skills have on students’ learning has received little attention despite its significance (see de Bot, 2014). Understanding the effect of students’ transitioning from elementary to secondary school on their L2 learning requires thorough longitudinal studies to optimize language proficiency outcomes in the long term. The current results provide at least some reasons for more caution, call for further investigation, and suggest, potentially, that early foreign language learning should be adjusted to take available research into consideration. We see two potential options for a possible change in early foreign language education:
-
increase the amount of exposure from Year 1 onward, possibly following immersive or content‐based approaches, or
-
move foreign language education into Year 3 or even Year 5 (i.e., secondary schooling), with more lessons and thus an overall more intensive approach.
If a more focused, intensive approach to early foreign language instruction yields better linguistic results, moving L2 instruction back into Year 3 with 4 instead of 2 45‐minute lessons might be an option worth pursuing, especially as the level of motivation for L2 learning in elementary school has been shown to be high and attitudes have remained positive. The available research should caution us not to assert that an earlier start of L2 education in elementary school will inevitably produce more proficient L2 users. Ultimately, all involved stakeholders have to ask themselves what they expect from foreign language teaching at the elementary and preschool levels and what they can realistically expect from minimal foreign language input of 1 or 2 45‐minute lessons per week.
Notes
- 1 In contrast, the opposite findings were reported in a large‐scale DESI study in Germany, which showed slight advantages for students with a nonmajority L1 background, compared to L1 German students, in terms of students’ English proficiency in Year 9 (Hesse et al., 2008).
- 2 This figure is based on 35 weeks of teaching out of 40 of a school year, following guidelines by the Department for Education in NRW to account for project weeks, school trips, and other missed or canceled lessons (MSW–NRW, 2008).
Number of times cited: 8
- 红 李, The Effects of Age of Onset and Language Input on the Phonological Skills of English Learners in an Under-Developed Rural Area of China, Modern Linguistics, 10.12677/ML.2018.62033, 06, 02, (265-272), (2018).
- Christine Dimroth, Beyond Statistical Learning: Communication Principles and Language Internal Factors Shape Grammar in Child and Adult Beginners Learning Polish Through Controlled Exposure, Language Learning, 68, 4, (863-905), (2018).
- Carmen Muñoz, Teresa Cadierno and Isabel Casas, Different Starting Points for English Language Learning: A Comparative Study of Danish and Spanish Young Learners, Language Learning, 68, 4, (1076-1109), (2018).
- Suzanne Graham, Louise Courtney, Theodoros Marinis and Alan Tonkyn, Early Language Learning: The Impact of Teaching and Teacher Factors, Language Learning, 67, 4, (922-958), (2017).
- Kazuya Saito, To What Extent Does Long‐Term Foreign Language Education Help Improve Spoken Second Language Lexical Proficiency?, TESOL Quarterly, , (2018).
- Robert DeKeyser, Input is not a panacea, International Journal of Bilingualism, 10.1177/1367006918768371, (136700691876837), (2018).
- Vera Busse, Jana Jungclaus, Ingo Roden, Frank A. Russo and Gunter Kreutz, Combining Song—And Speech-Based Language Teaching: An Intervention With Recently Migrated Children, Frontiers in Psychology, 10.3389/fpsyg.2018.02386, 9, (2018).
- Erdem Aksoy, Derya Bozdoğan, Ufuk Akbaş and Gölge Seferoğlu, Old Wine in a New Bottle: Implementation of Intensive Language Program in the 5th Grade in Turkey, Eurasian Journal of Applied Linguistics, 10.32601/ejal.464187, 4, 2, (301-324), (2018).
