Learning and Individual Differences 19 (2009) 209–218 Contents lists available at ScienceDirect Learning and Individual Differences j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / l i n d i f The cognitive contribution to the development of proficiency in a foreign language Benő Csapó a, Marianne Nikolov b,⁎ a b Institute of Education, University of Szeged, 6722 Szeged, Petőfi sgt. 30–34, Hungary Department of English Applied Linguistics, University of Pécs, 7624 Pécs, Ifjúság u. 6., Hungary a r t i c l e i n f o Article history: Received 4 October 2006 Received in revised form 29 November 2008 Accepted 1 January 2009 Keywords: Cognitive skills Language proficiency in English and German Longitudinal study a b s t r a c t The present paper reports results of a longitudinal research project studying the contribution of cognitive skills and other factors to proficiency in a foreign language (L2) in the Hungarian educational context. The larger project aims to describe the levels of L2 proficiency of school-aged populations in order to explore the conditions and factors contributing to processes and outcomes in foreign language education in public schools. For this purpose, paper and pencil tests were administered in English and German as a foreign language to nationally representative student samples. The project also aims to find answers to some theoretical questions; therefore, a questionnaire and other assessment instruments complemented L2 tests to provide insights into how participants' cognitive, affective and first language (L1) variables, as well as their social and school variables interact with one another over time. Students' general thinking and learning abilities were assessed with an inductive reasoning test. In the present paper we focus on the relationship between students' proficiency in English or German and inductive reasoning skills to show how general cognitive abilities interact with levels of L2 proficiency. We use a multivariate context to explore complex relationships between L2 levels in English and German and inductive reasoning skills if influences of other variables are controlled. We present results of multiple regression analyses on L2 listening, reading, and writing tasks in the two target languages. In the present paper we use both cross-sectional and longitudinal data to examine the relationships between students' L2 proficiency in the first phase (2000) and two years later (2002). Thus, a longitudinal research design was implemented by repeating cross-sectional assessment at a two-year interval. © 2009 Elsevier Inc. All rights reserved. 1. Background to study A number of studies have been conducted into the assessment of skills and competencies over the last two decades. The most intensively studied skills and abilities are related to learning skills determining success in learning tasks in the long run. Inductive reasoning is one of these skills often considered the main component of basic learning abilities (Pellegrino & Glaser, 1982; Ropo, 1987) and an indicator of learning potential (Resing, 1993). The role and importance of inductive reasoning in educational contexts has been studied in a number of empirical projects (see e.g., Csapó, 1997; Klauer, 1999), with several training experiments carried out with the assumption that its development contributes to the success of learning in a wide range of domains (e.g., Klauer, 1997; Klauer, Willmes, & Phye, 2002; Klauer & Phye, 2008). A similar trend has characterized research into cognitive variables contributing to second language acquisition (SLA), as the relationship between proficiency in L2 and general thinking skills has been of interest to researchers for a long time. Language learning aptitude has ⁎ Corresponding author. E-mail addresses: [email protected] (B. Csapó), [email protected] (M. Nikolov). 1041-6080/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.lindif.2009.01.002 been hypothesized to be the best predictor of L2 attainment, and recently discussions of its role have re-emerged (e.g., Carroll, 1990, 1993; Ellis, 1994; Gardner & MacIntyre, 1992, 1993; Grigorienko, Sternberg, & Ehrman, 2000; Harley & Hart, 1997; McDonough, 1999; Oxford & Ehrman, 1993; Parry & Stansfield, 1990; Robinson, 2002; Skehan, 1991, 2002; Sparks & Ganschow, 2001; Sternberg, 2002; Spolsky, 2000). As Sparks and Ganschow (2001, p. 100) suggest, aptitude tests should be updated and new norms developed. However, no large-scale study is available on how aptitude contributes to L2 proficiency. Language learning aptitude is traditionally conceptualized as a four-component construct and can be measured by the Modern Language Aptitude Test (MLAT, Carroll, 1990). Formally, the subtests of MLAT tap into four specific abilities: (1) phonetic coding ability; (2) grammatical sensitivity; (3) inductive ability; and (4) rote learning ability. Components (2) and (3) require learners to apply their inductive reasoning skills, whereas (1) and (4) test short-term memory as well as other areas. A different construct of language learning aptitude was proposed by Cummins (1991) and Cummins and Swain (1986), who distinguish between basic interpersonal communication skills (context-embedded and relatively undemanding cognitively) and cognitive academic skills (related to context-reduced, cognitively demanding communication) based on findings in bilingual 210 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 education research. Despite the wide acceptance of this distinction (e.g., Ellis, 1994), this line of reasoning has not resulted in a validated aptitude measure. Recent cognitive approaches emphasize a different distinction between processes interacting in SLA. As Skehan (1998) claims, two systems contribute to language development: a rule-based analytic system and a formulaic system based on examples. This proposed SLA model is in line with a dual declarative/procedural model widely accepted in neurolinguistics and cognitive science (Ullman, 2001). Approaching the issue from a different perspective, Dörnyei and Skehan (2002, p. 619) suggest that the construct of language learning aptitude should relate to stages in information processing. Despite these new approaches, so far no new validated instrument has been designed to tap into these proposed constructs of aptitude. As aptitude tests are always constructed and administered in learners' first language, such tests need to be validated for each language. Two such instruments, both based on the MLAT, have been validated for Hungarian learners in recent years. MENYÉT (Ottó, 1996), the Hungarian version of MLAT, was designed and validated for adults (Ottó & Nikolov, 2003), whereas Kiss and Nikolov (2005) validated one for young learners. The administration of both aptitude tests requires 60 min, thus their use was not feasible in our large-scale project. In summary, all conceptualizations of aptitude include inductive reasoning and this is how we chose to assess aptitude, since an assessment instrument of inductive reasoning validated on representative samples of Hungarian learners was available (Csapó, 1997, 1998, 2001). The test is easy to administer in 30 min and it has proved challenging enough for students between the ages of 10 and 18. Its three subtests (verbal analogies, number series, and number analogies) tap into verbal and numeric contents of inductive processes. Besides cognitive variables, affective factors have also been explored in the Hungarian context. Dörnyei, Csizér, and Németh (2006) inquired into 8th graders' language learning motivation, language choice and learning effort in large-scale surveys in 1993, 1999, and 2004. The most important findings revealed that the majority of Hungarian learners were highly motivated to study modern languages, their attitudes were favorable towards speakers of the language of their choice, and they were willing to exert an effort to study languages. As for their choices, the majority voted for English and German in the first place, but a major shift towards English was observed over the years. In a different large-scale study, Nikolov (2003) analyzed questionnaire data collected from representative samples of Hungarian learners in 6th, 8th, 10th and 12th grade (participants of the present study). The findings are similar to those of previous surveys, but as classroom processes were also inquired into, some data became available on what happens in English and German classes and how classroom activities interact with students' motivation. Some of these results will be referred to in the discussion section. As has been shown, cognitive and affective variables have been examined in the Hungarian context. The next step forward involves an examination of how cognitive, affective, and other variables contribute to L2 proficiency over time. In the present paper we explore two aspects related to the above discussions: how cognitive variables contribute to (1) learners' performances on language tests (crosssectional design) and (2) their development over a period of two years (longitudinal design). The present paper focuses on a particular socio-educational context – a member state of the European Union – and as such the study needs to be placed in a wider historical perspective. Recently, the learning of modern foreign languages, and levels of attainment in them, have moved to the fore in Europe (Council of Europe, 2001), as the aim is to encourage students to become plurilingual, culturally aware, and tolerant citizens of a multiethnic Europe. The knowledge of modern foreign languages is especially important in Hungary, where the first language is not widely spoken in other countries. For several social and historical reasons, L2 education has lagged behind other European countries (Fekete, Major, & Nikolov, 1999; Medgyes & Miklósy, 2000). Although Russian was a mandatory foreign language for decades, the majority of the population failed to achieve a useful level in it. Since the change of regime in 1989, foreign languages have played an exceptional role in Hungarian education. Perceived as extremely important, they enjoy a prestigious status: for example, proficiency exam certificate holders get special bonuses at entrance to universities, workplaces pay them special top-ups, and a decisive criterion in parents' choice of school for their offspring is the foreign language program of the institution. Recently, the demand for English as a lingua franca has consistently increased, but schools can hardly fulfill students' requests (Andor, 2000; Vágó, 2007), as teachers are tenured in their jobs and German classes also need to be filled. As a result of these processes, schools stream students in different language groups; thus, differences have emerged between students studying English and German. The cognitive and academic abilities of the latter group are significantly lower than those of the former (Csapó & Nikolov, 2002). Therefore, it is necessary to examine the conditions of L2 learning in the two most widely taught foreign languages. As for the language testing background to this study, the conceptualization and assessment of language proficiency is based on taxonomies of communicative competences and language ability (Bachman, 1990; Canale, 1983; Canale & Swain, 1980; Council of Europe, 2001), in which learners' performances are assessed in their language skills. In the choice of task and text types, piloting and validating tests, and evaluating results, we followed the principles of communicative language testing (Alderson, Clapham, & Wall, 1995; Bachman & Palmer, 1996). 2. The study Hungarian students' level of proficiency was assessed in English and German in 2000 and 2002 (Csapó & Nikolov, 2001, 2002; Nikolov & Csapó, 2002) in line with major Hungarian educational research projects monitoring other school subjects (Csapó, 1998, 2002). The present study looks into the relationships between cognitive factors and language proficiency to answer the following research questions: (1) How do cognitive skills and reading in the first language (L1) correlate with learners' performances in English and German in 6th, 8th, 10th and 12th grade (ages 12, 14, 16, and 18)? (2) What is the relationship between cognitive skills, prior knowledge, affective factors, school and family characteristics, reading in L1 and success in learning English and German over a period of two years in 6th and 8th, and in 10th and 12th grades? 3. Methods 3.1. Participants Representative samples of Hungarian students were chosen from primary and secondary schools in the spring of 2000. The units for sampling and data collection were school classes from all types of schools (mostly 8-year primary and 4-year secondary schools, but representative samples of other institutions were also involved). Since previous small-scale studies indicated considerable regional differences in levels of proficiency, relatively large samples were necessary to allow us to examine variations; thus, approximately 300 schools participated. At the first measurement point in May, 2000 altogether 29,126 students were assessed in grades 6, 8, and 10 (ages of 12, 14, and 16). In the second phase (May, 2002) 41,055 students were involved in grades 6, 8, 10, and 12 (age 18) from the same schools. This sampling allowed a longitudinal design for students assessed in the first phase in their 6th and 10th grade, as they attended the 8th and 12th grade of the same schools in the second phase, two years later. In the 8 plus 4 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 Fig. 1. The design of the entire project (sample sizes in parentheses). system students change schools after the 8th grade, thus, no link could be established between the 8th and 10th graders. Also, students who changed schools between the two phases were not followed. This research design resulted in two sets of cross-sectional data (2000 and 2002) and two sets of longitudinal data (see Fig. 1), as a number of students in the 6th and 10th grades in 2000 were assessed a second time as 8th and 12th graders in 2002. From the participants of the year 2000 samples 4958 6th graders and 4999 10th graders were assessed two years later. As the sub-samples consisting of those participants who were assessed twice (see arrows in Fig. 1) did not result from direct sampling, these samples are not strictly representative for the whole corresponding cohorts. Therefore, in test analyses and other detailed examinations we use data of the largest available (representative) samples, whereas longitudinal analyses are performed on the data of students assessed twice. Some students changed schools or dropped out between the two measurements and were not retested; therefore, the performances of the longitudinal sub-samples may differ slightly from the corresponding whole cohort of the first and the second assessment. The differences between the statistical characteristics of the whole samples and the sub-samples tested twice (both in 2000 and 2002) were carefully examined and they turned out to be so marginal that they do not affect the generalizability of the results for the whole cohort. For example, the difference of results on the inductive reasoning test for the whole 6th grader sample in 2000 and the first measurement of the longitudinal sub-sample was very small (whole 6th grader sample: mean= 34.287, sd = 16.979; longitudinal sub-sample: mean= 34.695, sd= 16.377). In a similar way, we may compare the second measurement of the longitudinal sub-sample with the entire sample of the 2002 8th graders. The differences (longitudinal sub-sample: mean= 50.040, sd= 18.060; whole 8th grader sample: mean= 49.110, sd = 18.280) are similarly small. This fact allows us to exploit the benefits of both crosssectional and longitudinal design. 3.2. Instruments Three types of instruments were used for data collection in 2000: (1) L2 proficiency tests in English and German, (2) a questionnaire on learners' school achievements, attitudes toward L2 learning, their 211 language learning plans and self concept, and their parents' level of education (to indicate socio-economic status), and (3) a standardized inductive reasoning skills test. In addition to these, in 2002 a reading comprehension test was also administered in the first language (L1, Hungarian). All participants' proficiency was assessed in three L2 skills (listening and reading comprehension, and writing) by paper and pencil tests in regular classroom settings at the end of the school year. The English and German tests were based on the prescribed achievement targets of the national core curriculum; they were identical in their construct, structure, type of texts and tasks and length (number of items) for the two target languages. Table 1 illustrates the structure of the booklets, the text and the task types for grade 8. The construct, the task and text types were the same in 2000 and 2002, but the actual texts were different. The task types were familiar to the vast majority of the participants, as they were similar to the ones used in published course materials used in schools. All tasks focused on meaning (and not form) and were in harmony with curricular achievement targets for the four age groups. The texts were authentic, except for the listening tasks in grades 6, 8 and 10, in which scripted materials were used, as listening comprehension (the most neglected skill in schools) was expected to be problematic according to classroom observation studies (Nikolov, 1999). In grade 6 writing tasks integrated reading, as in the pilot phase the majority of the students failed to produce meaningful sentences in a short writing task. Therefore, tasks required learners to choose and copy words into a gapped invitation to a party and to fill in a form based on a short text with personal data. In grade 8 students were asked to write a short letter based on diary entries, whereas 10th and 12th graders wrote a guided composition on whom they would nominate to win a competition and why. Booklets were produced in two versions: the sequence of the tasks was different, but the actual tasks were identical. The reliability (Cronbach alpha) of these tests varied between 0.94 and 0.96 for both English and German. Besides the language tests, in both assessment phases a questionnaire was administered to all students on their school marks (in all subjects and in L2), language learning attitudes, motivation, plans and self concept, and family background (to indicate socio-economic status). The questionnaire comprised multiple-choice and Likert-scale items. The inductive reasoning test included verbal analogies, number analogies, and number series tasks in Hungarian to assess the students' general cognitive abilities (Cronbach alpha = 0.94). It had been piloted and validated in previous studies and served as a valid and reliable indicator of Hungarian learners' inductive reasoning skills (Csapó, 1997, 1998, 2001, 2002). The same test was administered both times to all participants. In 2002, a reading comprehension test in L1 was also administered to 6th, 8th and 10th graders. It comprised two tasks; each text was an authentic narrative of about 300 words and each task included 15 items: nine items were closed (true-false, multiple choice, and sequencing), four were open cloze items, and two required short answers. The same test was administered to each cohort. The L1 Table 1 The structure and content of booklets for grade 8. Skills Task Input No of items Reading 1 Reading 2 Reading 3 Reading 4 Reading 5 Listening 1 Listening 2 Writing Match texts with meanings Match questions with answers Match words with definitions Match titles with texts Match ads with missing words/expressions Multiple choice on videos MC on planning a holiday Short informal letter Public notices and their meanings Quiz questions and answers Definitions Titles and blurbs Advertisements Conversation (scripted) Conversation (scripted) Diary entries for seven days 10 8 10 8 9 7 8 marking scheme 212 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 Table 2 Correlations between proficiency in English skills, reading comprehension in Hungarian, and inductive reasoning skills (based on 2002 assessment of whole samples). Year Test 6th grade English writing English listening Hungarian reading Inductive reasoning 8th grade English writing English listening Hungarian reading Inductive reasoning 10th grade English writing English listening Hungarian reading Inductive reasoning 12th grade English writing English listening Inductive reasoning English reading English writing English listening 0.715 0.574 0.487 0.527 0.662 0.569 0.375 0.505 0.609 0.494 0.275 0.430 0.455 0.422 0.335 0.577 0.493 0.519 0.300 0.329 0.505 0.372 0.467 0.303 0.441 0.407 0.276 0.381 0.220 0.310 0.277 0.269 0.149 German, we can examine correlation coefficients of a large number of variables. The results are presented in Table 2 for English and in Table 3 for German. First, we analyze data for the two target languages separately; then, we compare figures for English and German in the four age groups. In our comparisons of correlation coefficients we discuss differences if they are larger than 0.1. (Taking the sample sizes into account, these differences are statistically significant at p b 0.001). 4.2. Relationships between English, inductive reasoning, and L1 reading n N 3000, p b 0.001 for all coefficients. reading comprehension test was not administered to 12th graders because of an expected ceiling effect and time constraints. With the exception of the L2 tests, all instruments were administered in L1. 3.3. Procedures Data collection was implemented in the same way in both phases. The paper and pencil tests were administered locally with the help of trained external assessors in learners' classes in May 2000 and 2002. The reading booklets including five reading tasks (Table 1) were completed in a 45-minute class session, and the listening and writing tasks were done in another session of the same length. A third session was devoted to filling in the inductive reasoning test and the questionnaire. Writing tasks were evaluated centrally by two teams of trained assessors (English and German) after standard-setting sessions; data were entered into files during the summer. Schools got their coded feedback during the fall. In the present paper we use data on questionnaire items, inductive reasoning skills, and L2 proficiency tests from the longitudinal samples. When focusing on predictors of L2 proficiency in some analyses the language proficiency measured in 2002 will be represented by a single variable. For this purpose, the equally weighted reading, writing and listening test results were transformed into a Z-scale and Z-scores of the three skills were totaled (Tables 6–9). As was mentioned in the section on participants' samples, we analyze students' data providing the largest possible representative samples. Tables indicate the samples used for the computations. 4. Results 4.1. Relationships between L2 proficiency and cognitive skills The first research question concerns the relationships between participants' levels of proficiency in L2 (English and German) and cognitive skills, and other variables by analyzing within-age correlations. As besides L2 proficiency, inductive reasoning skills and other data were also collected on a number of student characteristics, these analyses are placed in the context of these relationships. Therefore, in order to increase the validity of the results, some other variables are added to the analyses. Relationships are examined with two variables indicating students' developmental level of cognitive skills: reading comprehension in L1 and general intellectual development (represented by inductive reasoning). The four age groups (grades 6, 8, 10, and 12) and two target languages allow us to study eight separate sets of relationships on the data collected in 2002. As three skills were assessed in English and The correlation coefficients computed from data of the second assessment (2002) are summarized for the three English skills, L1 reading, and inductive reasoning in Table 2. The highest correlations were found between the L2 reading and L2 writing skills in each age group (0.715, 0.662, 0.609 and 0.455). However, in the case of 8th and 12th graders the differences are minimal between these figures and correlations between reading, listening, and writing in grade 8, reading and listening in grade 12. Beyond the similarities of the underlying cognitive and linguistic processes of reading and writing, these high correlations may be attributed to several factors. In grade 6 the strong relationship must be due to the type of tasks used for testing learners' writing skills. For the youngest age group, writing tasks integrated writing with reading: students were expected to choose words from a list to fill in a gapped invitation and a form with information provided in a short text. In other words, writing was actually minimal, it involved copying appropriate words in context, reflecting achievement targets in the curriculum and a low level of interlanguage in L2 writing. In grade 8 reading and writing skills were also interrelated, as students wrote a short personal letter based on short entries in a diary, and they needed to use the given words and expressions in meaningful sentences. The correlations between reading and writing skills in 8th and 10th grade are similar, but lower in grade 12, indicating weaker relationships. This is interesting, as in 10th and 12th grades, writing tasks were guided compositions tapping into learners' productive skills with minimal prompt. Students were required to write about 150 words, and their writings were assessed along four criteria: task achievement, vocabulary, grammar/accuracy, and text cohesion. Correlations between listening comprehension, reading comprehension, and writing in English are fairly high in 6th and 8th grades, indicating that although the booklets did tap into different L2 skills, they are strongly interrelated. In 10th and 12th grades, however, relationships are weaker, especially in the case of listening and writing in the oldest group. Decreasing correlations over the years in the cross-sectional datasets show that proficiency in the three skills Table 3 Correlations between proficiency in German skills, reading comprehension in Hungarian, and inductive reasoning skills (based on 2002 assessment of whole samples). Year Test German reading 6th grade German writing German listening Hungarian reading Inductive reasoning German writing German listening Hungarian reading Inductive reasoning German writing German listening Hungarian reading Inductive reasoning German writing German listening Inductive reasoning 0.635 0.545 0.390 0.453 0.605 0.502 0.278 0.427 0.606 0.547 0.263 0.385 0.566 0.432 0.381 8th grade 10th grade 12th grade n N 3000, p b 0.001 for all coefficients. German writing German listening 0.488 0.459 0.464 0.223 0.291 0.394 0.357 0.422 0.213 0.320 0.444 0.317 0.419 0.186 0.315 0.318 0.282 0.174 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 Table 4 Correlations between proficiency in three skills in English in 6th and 10th grade, and two years later in 8th and 12th grade, respectively. First assessment (2000) Test 6th grade Reading Writing Listening Inductive reasoning Reading Writing Listening Inductive reasoning 10th grade Second assessment (2002; 8th and 12th grade) Reading Writing Listening 0.570 0.528 0.418 0.446 0.623 0.485 0.428 0.390 0.529 0.519 0.388 0.475 0.408 0.449 0.310 0.260 0.407 0.433 0.398 0.359 0.302 0.249 0.238 0.177 n N 2300, p b 0.001 for all coefficients. becomes gradually less dependent on one another, most importantly between 6th and 12th grades. As for the relationship between reading comprehension in L1 and L2, the correlations are highest (0.487) for the youngest age group and decrease gradually (0.375; 0.275) for 8th and 10th graders, respectively. Also, these correlations are lower than correlations among English skills. Correlations between inductive reasoning and the three skills in English indicate similarly strong relationships (0.527; 0.505; 0.430; 0.335) to those found for between-skills. In line with other correlations in Table 2, they decrease over the grades. As for the relationships between inductive reasoning and English skills, they are consistently somewhat stronger for reading comprehension than for the other skills in each grade (see Table 2). 213 tasks in these two grades. (3) L1 and L2 reading comprehension scores are less strongly related than the three L2 skills. (4) The relationships between inductive reasoning skills and L2 skills are consistently stronger for reading and in some cases writing than for listening comprehension in 6th and 12th grades. (5) Correlations between inductive reasoning and L2 skills in both languages are generally larger than the correlations between L1 and L2 skills and in many cases these differences tend to be greater than 0.1. Systematic differences are also found in the two sets of data: (1) Some correlations in 6th and 8th grades are higher for learners of English indicating somewhat stronger relationships in their case; (2) the picture is more mixed in the datasets for 10th and 12th grades: relationships are either fairly similar or even a bit stronger for German than in the case of English. 4.5. Relationships between proficiency levels in L2 and inductive reasoning skills over time In this section we seek an answer to the second research question involving the same participants tested in 2000 and 2002. We analyze the relationship between cognitive skills and L2 proficiency in grades 6 and 10 and two years later in the levels of performances in English and German in grades 8 and 12. On the basis of the relationships between the data collected in 2000 and 2002, we can estimate the factors influencing the success of language learning. For these analyses, first we explore inter-age relationships between language skills and inductive reasoning; then, we apply multiple regression models to involve more variables and determine the main influencing factors. 4.6. Relationships for English 4.3. Relationships between German, inductive reasoning, and L1 reading Let us now analyze data on learners of German presented in Table 3. Similarly to English, the highest correlations were found between the L2 reading and L2 writing skills in the four age groups (0.635, 0.605, 0.606 and 0.566), although these correlations are very similar to those characterizing relationships between listening and reading in 8th and 10th grades. The weakest relationships are found between German listening and writing in 8th, 10th, and 12th grades. Correlations weaken over the years, especially between 6th and 12th grades. As to the relationships between reading comprehension in German and L1, correlations are highest (0.390) in grade 6, and a slight decrease characterizes results in 8th and 10th grade, though figures are similar in the older groups (0.278; 0.263). Correlations between German learners' inductive reasoning skills and their proficiency in L2 reading comprehension indicate moderately strong relationships (0.453; 0.427; 0.385; 0.381), similarly to correlations with L2 writing and L2 listening; they are, however, similar in all four years. As for the relationships between inductive reasoning and the three skills in German, they are stronger for reading comprehension and writing, and weaker for listening in the two younger grades, whereas the relationships are less clear in the older years. A slight decrease characterizes relationships over time (see Table 3). 4.4. Comparing learners of English and German A comparison of data for learners of English (Table 2) and German (Table 3) indicates similarities. (1) Overall, relationships between variables weaken over time, especially between 6th and 12th grades. (2) In all four years the highest correlations are found between reading comprehension and writing in both English and German, although the other figures across skills are fairly similar in the younger groups. Relationships are almost identical in 8th and 10th grades, despite the major difference in the cognitive and linguistic demand in the writing Inter-age correlations for the three skills in English presented in Table 4 show how the earlier (6th and 10th grade) proficiency level in a skill relates to the level of the same skill two years later (in 8th and 12th grade). Thus, the correlations indicate the stability of skill development. The between-skill correlations show how the level of proficiency in an L2 skill is related to the development of another L2 skill. These data allow us to analyze the relationships between skills in prior L2 knowledge and L2 proficiency two years later in the same learners. As for the inter-age relationships for learners of English (Table 4), reading proved to be the most stable skill with correlations of 0.570 between 6th and 8th grade, and 0.623 between 10th and 12th grade. Correlations between reading and writing and writing two years later are similarly high. English writing performances in grades 6 and 10 are the second best predictors of success in all three skills, but the correlations are lower than for reading in the older age group. The correlations for listening show the weakest relationships with all skills, including listening skills two years later, especially in the transition between 10th and 12th year. Correlations between Table 5 Correlations between proficiency in three skills in German in 6th and 10th grade, and two years later in 8th and 12th grade, respectively. First assessment (2000) Test 6th grade Reading Writing Listening Inductive reasoning Reading Writing Listening Inductive reasoning 10th grade n N 2300, p b 0.001 for all coefficients. Second assessment (2002; 8th and 12th grade) Reading Writing Listening 0.469 0.334 0.180 0.345 0.590 0.553 0.356 0.370 0.436 0.369 0.160 0.391 0.475 0.603 0.279 0.269 0.355 0.246 0.135 0.252 0.352 0.329 0.259 0.118 214 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 Table 6 The percentage of variance of proficiency in English in grade 8 explained by variables assessed in grade 6. Independent variables: measured in first assessment (6th grade) Dependent variables assessed in grade 8 Reading Listening Writing English total Grade point average Language mark Attitude toward learning language The intended level of education Language learning plans Parents' education Language self concept Inductive reasoning English reading English writing English listening Total variance explained (R2) 0.8 6.3 0.1 1.6 1.7 5.1 2.7 1.6 17.5 2.4 5.2 44.9 0.6 1.9 1.5 3.1 1.6 3.2 1.0 1.7 4.3 6.4 4.6 29.8 3.3 3.4 0.7 1.0 4.0 8.8 3.4 3.3 11.5 0.2 6.3 45.9 2.4 6.0 0.9 2.6 3.2 3.3 3.2 7.7 14.6 7.5 4.2 55.7 Figures significant at least at p b 0.01 are in italics. inductive reasoning and the English language skills are quite high in the transition from 6th to 8th grade, but somewhat lower in 10th to 12th grade data. The overall picture indicates a systematic relationship between learners' proficiency in English in 6th and 10th grade and their performances two years later. 4.7. Comparing relationships for German and English As figures in Table 5 illustrate, correlations for learners of German are lower than those for learners of English in the younger groups, whereas the figures show a different trend in the transition from 10th to 12th grade. As for the relationships between performances on the inductive reasoning and English and German proficiency tests in 6th and 10th grade and two years later, figures are smaller in the transition between 6th and 8th grade for learners of German, and they are the same in the case of the older learners. The pattern for the relationship between writing in English in 6th grade and the relatively strong relationship between performances in reading and writing two years later makes sense, as in grade 6 reading and writing skills were integrated in the writing tasks. However, a much weaker relationship is reflected by the low correlations between German writing in 6th grade and reading and writing two years later. This finding may be related to the generally lower performances in German (in the present study, see Csapó & Nikolov, 2001 for further detail, and in a follow-up enquiry, see Nikolov & Józsa, 2006) and it also means that proficiency in German is less stable than in English. The level of proficiency in L2 skills and general intellectual abilities (represented by inductive reasoning) show weaker relationships with L2 proficiency two years later in the case of German than in the case of English in the transition between 6th and 8th grade, whereas the correlations are almost identical in the older groups. It is likely that other variables such as learning opportunities, attitudes, motivation, and factors related to school, family and peers influence the pace of development in the German language differently than in the case of English. This phenomenon is in line with the observations worded in the previous sections. To put it differently, as less enthusiastic and more disadvantaged students study German in the primary schools, their performances tend to depend more on unpredictable and incidental influences than those of their peers studying English. In the case of the older learners, students responsible for low relationships may not be in secondary schools, as they do not continue their education. Correlations between German listening comprehension in 6th grade and the three skills in 8th grade indicate very weak relationships. In contrast, data for the transition between 10th and 12th grades show a different picture. The correlations between reading and writing indicate that performances in 10th grade are indeed good predictors of L2 proficiency two years later; in fact, the strongest relationship is found between writing in the two grades (0.603), stronger than for English (0.449). Thus, less variation is found in students' performances in their transition from 10th to 12th grade than in the case of younger students. Comparing the correlations for inductive reasoning skills and the three skills in English and German in Tables 4 and 5, some trends are consistent: (1) figures are higher for English than for German in the younger groups and (2) also for relationships between the two younger grades than for the older age groups. In addition, (3) relationships tend to be stronger between inductive reasoning and reading and writing in both L2s than for listening. 4.8. Multiple regression models In order to estimate how certain variables predict L2 proficiency two years later, multiple regression analyses were performed. In regression analysis the impact of one (independent) variable on another (dependent) one is estimated by the proportion of variance explained by the examined independent variable. As the independent variables may also correlate with each other, some of them may indicate mediated effect. As regression models take explained variance into account only once, mediated impacts can be filtered out. Language proficiency, the dependent variable in the models, is represented both by data on the three L2 skills and a single L2 variable composed by combining the three skills for the purpose of multiple regression analysis. As for the independent variables (predictors), we used variables showing high correlations with one or more L2 skills in previous analyses. These variables represent different areas affecting language learning. Separate analyses were performed for the two target languages (English or German) and for the two age groups in them (see Tables 6–9). The same set of independent variables was used in each analysis, thus the effect of these variables on the target variables can be compared directly. Tables 6–9 include only the percentages of the variance explained by the independent variables (beta multiplied by the correlation coefficient and by 100). As a first step, the regression models for English are examined. Figures in Tables 6 and 7 indicate the percentage of the variance of the dependent variables (students' proficiency in three skills in English at the end of grade 8 and grade 12) explained by the eleven independent variables used in the models. As expected on the basis of the correlations presented in the previous sections, the best predictor of overall proficiency in English in grade 8 is the level of L2 reading comprehension measured in grade 6 (14.6% variance explained in Table 6). The ratio of explained variance of the level of reading comprehension in grade 8 is even higher: 17.5% and it is also high in the case of writing: 11.5%. Other components of Table 7 The percentage of variance of proficiency in English in grade 12 explained by variables assessed in grade 10. Independent variables: measured in first assessment (10th grade) Dependent variables assessed in grade 12 Reading Listening Writing English total Grade point average Language mark Attitude toward learning language The intended level of education Language learning plans Parents' education Language self concept Inductive reasoning English reading English writing English listening Total variance explained (R2) 3.7 2.2 − 0.8 0.2 1.5 2.9 3.4 0.6 24.3 4.1 7.7 49.7 0.7 0.4 − 0.3 0.4 − 0.1 0.8 1.3 0.5 3.9 2.9 2.5 13.1 5.5 0.8 3.3 2.1 2.7 0.1 − 0.3 0.1 5.6 2.3 9.0 31.4 5.0 1.4 0.6 1.2 1.9 2.2 0.6 2.1 5.7 11.0 5.6 47.2 Figures significant at least at p b 0.01 are in italics. B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 215 Table 8 The percentage of variance of German language knowledge in grade 8 explained by variables assessed in grade 6. Table 10 Summary of the results of regression analyses: percentage of variances explained by the independent variables. Independent variables: measured in first assessment (6th grade) Dependent variables assessed in grade 8 Reading Listening Writing German total Independent variables: measured in first assessment (grade 6 or 10) Grade point average Language mark Attitude toward learning language The intended level of education Language learning plans Parents' education Language self concept Inductive reasoning German reading German writing German listening Total variance explained (R2) 2.9 1.7 1.8 0.8 0.8 4.0 − 0.2 3.3 15.0 0.4 0.5 31.1 3.3 0.9 0.5 − 0.1 − 0.2 1.3 0.2 1.9 10.2 0.2 − 0.3 17.9 9.7 5.5 2.0 − 0.7 1.6 4.7 0.3 3.4 9.4 − 0.3 1.6 37.3 7.6 3.8 2.0 0.0 0.9 0.0 4.3 4.6 17.6 0.3 0.0 41.1 Figures significant at least at p b 0.01 are in italics. prior knowledge, the two other language skills have smaller contributions to the explained variance (7.5%; 4.2%), whereas the second best predictor of L2 proficiency is the score on the inductive reasoning test (7.7%). Language mark in 6th grade explains 6% of variance; language learning plans, parents' education, language self concept, and grade point average also explain a significant ratio of variance. The three language skills in grade 6 explain over 26% of the variance in grade 8; the total variance explained by the independent variables amounts to 55.7%. The total variances in the three skills vary: they are high for reading and writing (44.9% and 45.9%, respectively) and lower for listening comprehension (29.8%). The figures show a different picture for English in the 12th grade (Table 7). The three language skills combined explain over 22% of variance, including 11% on English writing. Four other variables contribute significantly to the explained variance: grade point average 5.0%, language learning plans, self concept, and inductive reasoning about 2% each. The total variance explained in the model is 47.2%. The ratio of explained variance is different for the older learners in the three L2 skills: 24.3% of variance in English reading is explained by proficiency in the same skill two years earlier (in grade 10), whereas the ratio for writing is lower (5.6%). The reason must be that these two skills were more clearly separated in the writing tasks than in the case of younger learners. Figures in Tables 8 and 9 summarize the results of regression analyses for learners of the German language. As for the relationships between 6th and 8th graders (Table 8), reading comprehension in German explains the highest ratio of variance (17.6%), followed by 7.6% for grade point average, 4.6% for inductive reasoning, 4.3% for self concept, 3.8% for language mark and 2.0% for attitudes towards Table 9 The percentage of variance of proficiency in German in grade 12 explained by variables assessed in grade 10. Independent variables: measured in first assessment (10th grade) Dependent variables assessed in grade 12 Reading Listening Writing German Total Grade point average Language mark Attitude toward learning language The intended level of education Language learning plans Parents' education Language self concept Inductive reasoning German reading German writing German listening Total variance explained (R2) 1.4 1.2 0.1 0.0 0.5 4.9 0.6 3.6 19.7 3.0 13.9 48.9 − 1.2 0.0 1.8 − 0.6 1.3 − 0.4 0.0 − 0.1 8.9 3.1 6.2 19.1 4.4 − 0.2 4.6 1.5 0.8 0.3 0.0 2.4 6.7 0.3 22.5 43.4 0.0 0.5 3.6 0.2 2.0 0.3 2.9 1.3 17.3 21.7 3.8 53.2 Figures significant at least at p b 0.01 are in italics. Grade point average Language mark Attitude toward learning language The intended level of education Language learning plans Parents' education Language self concept Inductive reasoning L2 reading L2 writing L2 listening Total variance explained (R2) Dependent variables: total scores measured in second assessment (grade 8 or 12) English English German German 6–8 10–12 6–8 10–12 2.4 6.0 ns 2.6 3.2 3.3 3.2 7.7 14.6 7.5 4.2 55.7 5.0 ns ns ns 1.9 2.2 ns 2.1 5.7 11.0 5.6 47.2 7.6 3.8 2.0 ns ns ns 4.3 4.6 17.6 ns ns 41.1 ns ns 3.6 ns 2.0 ns 2.9 ns 17.3 21.7 3.8 53.2 Figures are significant at least at p b 0.01. The three highest figures are in bold in each column. language learning. Interestingly, the contribution of the other language skills is not significant (data for writing and listening are extremely low). In other words, it does not seem to matter how good students' listening and writing skills are in 6th grade, what matters is their reading skills. In this age group the lowest ratio of variance is explained by the independent variables: 41.1%. A different picture emerges from analyzing the data of 10th and 12th graders (Table 9). German writing results in 10th grade explain 21.7% of variance two years later and German reading explains 17.3% of variance in grade 12. The contribution of listening skills is also significant (3.8%), similarly to attitudes towards learning German (3.6%) and language self concept (2.9%). The total variance explained by these variables is 53.2%. If we compare the role of prior L2 proficiency in the four models large differences can be found in the case of the two languages: in the case of English, the ratios of explained variance are about 26% in the younger grades, and 22% in the case of older learners. As for German, the percentages are 18 and 43, respectively. In summary, different patterns unfold as we compare the two age groups (grades 6–8 and 10–12) and the two target languages (Tables 6–9). Table 10 allows us to overview our data. (1) The eleven independent variables explain about 50% of variance in the four models. (2) In both languages and both age transitions previous language proficiency in the three L2 skills explains the highest ratio of variance, although the actual percentages vary, and reading and writing are mostly responsible for this. (3) Inductive reasoning skills explain higher ratios of variance for the younger age groups, as well as for learners of English. (4) Students' parental background does not seem to play a significant role in the case of German, unlike in the case of English. (5) Grade point average and language mark explain a different amount of variance in the four models: 2.4% and 5.0% in the case of English, 7.6% and 0 in the case of German learners, and 6.0% and 1.4% for English learners, 3.8% and 0.5% for German learners, respectively. (6) Attitudes towards language learning have significant effects in the groups of German learners. In contrast, parents' educational level seems to exert significant effects only in the groups learning English. (7) The total variance explained is higher in the case of the younger cohorts learning English (55.7%) and lower for the older learners (47.2%). The pattern for learners of German is reverse: 41.1% for the younger groups and 53.2% for the more mature students. (8) Variances for English were significant for more variables included in the model than for German. (9) It is remarkable that the six significant variables explain such a high proportion of variance in the case of the older cohort of German learners. (10) Prior knowledge predicts L2 proficiency to a different degree in the two target 216 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 languages: whereas in the younger groups learning German only reading proficiency is what really matters two years later, proficiency in reading and writing in 10th grade is a very good predictor of how successful learners are two years later. The picture is more balanced in the case of English. The analyses indicate that although there are some clear tendencies, the patterns identified in the factors that determine outcomes in L2 learning vary. The two target languages represent different cases and the relationships between factors change with age. pointed out, while taking a reading test it is possible to look back to previous parts of the text, whereas in listening a higher demand is placed on memory, thus, listening can be considered more cognitively demanding than reading. Another point raised by our reviewer relates to the fact that the inductive reasoning test required reading, which could be another reason that correlations and regression coefficients are higher with reading and writing than for listening (we are grateful to our reviewer for these comments). 5.4. Relationships between L1 and L2 literacy skills 5. Discussion 5.1. The cognitive contribution to proficiency in L2 In the present study a validated inductive reasoning test was used to examine the cognitive contribution to L2 proficiency in four age groups and over time. Several findings have transpired. (1) The cognitive contribution to L2 proficiency is significant but not substantial, with various degrees of ratio in explained variance in the two target languages and in the two age groups. (2) The relationships between inductive reasoning skills and the three L2 skills consistently weaken between 6th and 12th grade. (3) Also, participants' developmental level of inductive reasoning skills explains a higher ratio of variance in the transition between 6th and 8th grades in both L2s than in the older groups (10th and 12th grades), but the ratio is higher for English, (4) the relationships are consistently stronger for reading and writing than for listening comprehension, (5) the relationships between the three L2 skills weaken over the years, and (6) a similar tendency is found between L1 reading comprehension and L2 skills, (7) all relationships are somewhat stronger for English than for German. Finally, (8) the best predictor of proficiency over two years is the level of proficiency in L2 assessed two years earlier. Table 10 summarizes the data for the two languages and the two transitions. These findings will be discussed in what follows. 5.2. The role of aptitude In comparison to findings in another study on Hungarian students' aptitude, the amount of variance explained by inductive reasoning is relatively low (between 1.3 and 7.7), but the results make sense. The second study involved a convenience sample of 419 sixth graders studying English (Kiss & Nikolov, 2005). Although learners of German did not participate, the findings are relevant for the youngest age group. Language learning aptitude explained 22% of the variance in English language performances, whereas motivation (an index was applied on a number of motivational items) also had a significant contribution (8%). Besides these two variables, marks in English (17%), and the length of exposure (8.5%) explained a total of 55.5% of the variation. Correlations on the level of individual learners showed a strong relationship between aptitude and English scores (0.627). These findings are in line with the outcomes of our analyses; the MLAT-based aptitude test obviously taps into a wider range of cognitive and linguistic skills than the inductive reasoning test; hence, both the correlation and the explained variance were higher. 5.3. Relationships between L2 skills Our database shows consistently stronger relationships between L2 reading and L2 writing than between these and L2 listening comprehension. Reading and writing are obviously interrelated literacy skills, whereas listening comprehension is not and it is often seen as the least academic of these skills: reading and writing are cognitively more demanding. This is in line with Cummins' (1991) distinction between cognitive academic language proficiency and basic interpersonal communication skills: listening comprehension tests tap into the latter. However, as one of our anonymous reviewers The relationship between literacy skills in L1 and L2 also deserve special attention. As was shown in Tables 2–5, it is not only correlations between L2 skills that tend to decrease with age, indicating that skills become less interdependent as they develop over time. A similar tendency was found between reading comprehension in L1 and all three skills in L2. These findings show that learners at lower levels of proficiency rely more on their general skills; as they gain proficiency in the L2, their language skills become gradually more independent from both cognitive and L1 skills. This finding is in harmony with results of other educational research on Hungarian students' school achievements (Csapó, 1998, 2002), as well as with findings in bilingual education (Cummins, 1991) and in studies on the assessment of reading comprehension in L1 and L2. Obviously, the linguistic and literacy knowledge and skills students have learned in their L1 bear on the learning of L2 involving, for example, knowledge of how to approach a text and background knowledge of the world. The problem can be worded slightly differently: Is reading in L2 a language problem or a reading problem (Alderson, 1984)? It is clear from the literature that L2 knowledge and reading knowledge interact, and in L2 reading comprehension L2 proficiency is a more important factor than L1 reading ability (Alderson, 2000, p. 23). Our findings are in line with this: as learners' L2 proficiency develops over years, L2 reading comprehension gradually becomes less dependent on L1 reading, and shows stronger relationships with other L2 skills. 5.5. The role of L1 As for the relationship between L1 and L2, it is reasonable to assume that reading in a L1 requires less conscious knowledge, whereas in a foreign language learners rely more on their conscious knowledge and monitor their performance. On the other hand, a higher level of general intellectual skills may facilitate constructing meaning of a text in an L2. The processes of induction may be similar to those of meaning making. Another issue with implications for language teaching methodology also deserves discussion. As findings concerning the relationship between L1 and L2 show, performances in L1 and L2 reading comprehension are less strongly related than those between foreign language skills. This outcome shows that L1 and L2 develop quite independently from one another despite extensive uses of translation and contrastive analysis in Hungarian classrooms (also documented in questionnaire data in the present study, see Nikolov, 2003). Finally, a linguistic argument needs to be mentioned in connection with the cognitive variable and the L1. Many language teachers as well as lay people assume that English is easier to master than German and a lower level of fluency is more easily available in English than in German, as the latter is seen as cognitively more demanding because of its more complex morphology. The Hungarian language is not related to either of the two target languages and its morphology is complex indeed. The present study shows that a higher ratio of variance is explained by the cognitive variable for learners of English than for German, and the ratio in the older groups is lower. These two findings could be interpreted as counter-evidence to the above assumptions, but obviously further research is necessary. B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 5.6. Younger versus older learners Results concerning relationships between younger and older groups in the two target languages present an exciting point of departure, as higher ratios of variance are explained by inductive reasoning skills for the younger age groups (Table 10). The literature on the age factor (e.g., Nikolov & Mihaljevic-Djigunovic, 2006; Singleton & Ryan, 2004) and studies comparing early and later start foreign language programs (e.g., García Mayo & García Lecumberri, 2003; Muñoz, 2006; Nikolov, 2000) unanimously claim that younger learners are characterized by implicit language learning, memory skills and formulaic chunks, whereas older learners apply a rule-based analytic system. This may be the explanation why a larger ratio of variance is explained in the transition between 6th and 8th graders' L2 proficiency than in the case of the older cohorts. A higher level of cognitive skills at an earlier age is a better predictor of performances two years later than in the high-school cohorts in which less variation characterizes participants in the study. As to why a higher ratio of variance is explained in the case of English than German, the most probable reason is that, as is shown in Table 10, other data characterizing younger learners of English are consistently higher than those of their peers learning German. Their intended level of proficiency and language learning plans, both reflecting awareness and reasoning skills, in English explain 2.6 and 3.2%, respectively, whereas the data are not significant for learners of German. On the other hand, attitudes, an affective variable, explain 2.0% of variance in the case of younger learners of German, but no significant relationship characterizes learners of English. It is also remarkable that two out of the three L2 skills show very different patterns for younger learners of English than for their peers studying German. No significant data were found for writing and listening skills for the latter, whereas 7.5 and 4.2% of variance is explained by the level of these skills in 6th grade in English. In addition to these, parents' education explains quite a large portion of variance in both age transitions of English learners (3.3; 2.2), but the data are not significant for either groups of German learners. These trends show more random variation in data for learners of German. 5.7. Other factors contributing to L2 proficiency As for the differences in the cognitive contribution between learners of English and German, it is important to point out that in all grades learners of English had slightly but significantly better inductive reasoning skills, and their performances were significantly higher than those of their peers learning German (Csapó & Nikolov, 2002). This latter finding is in harmony with a nation-wide follow-up study conducted a year later (Nikolov & Józsa, 2006). The first point may partly explain the second one, but learners' inductive reasoning skills interact with other factors. Language learning attitudes and motivation, as well as parents' level of education, were found to be more favorable in the case of learners of English than for students studying German (Nikolov, 2003). Also, they tended to achieve better grade point averages and marks in the L2 (Csapó & Nikolov, 2002). Thus, more favorable school feedback may increase and help maintain their language learning motivation to a greater extent. The perceived status of English as a more prestigious lingua franca (Dörnyei et al., 2006) may boost learners' attitudes and motivation, whereas learners of German may feel that their language is less in demand. 6. Conclusion The present study set out to examine the cognitive contribution to L2 proficiency in four different age groups and over a period of two years. As has been shown, over time learners' characteristics mediate family and contextual variables and educational practices in complex ways. In the regression analyses eleven variables explain about 50% of 217 variance in the models, but fewer variables exert a significant influence on outcomes in the case of German than in the case of English. In the younger age groups inductive reasoning skills explain larger portions of the variance, and the relationships between the cognitive variable, L1 reading, and other factors are stronger than in the case of older learners. As an effect of maturation and language development, the relationships between L2 skills weaken. Reading comprehension and writing skills tend to be more strongly related to one another, as well as to cognitive skills and L1 reading, than listening comprehension indicating that L2 literacy skills are cognitively more demanding than aural skills. The best predictor of L2 proficiency after two years of study is prior level of proficiency, more specifically reading comprehension and writing in L2. The issue of the extent to which cognitive skills and prior knowledge predict success in learning English and German over time has important implications for teaching. Language teachers often complain that students' level of proficiency when they start teaching them is below their expected level. One wonders how teachers in upper grades rely on what students bring with them and how efficiently they scaffold their learning based on what they can do at the time of transition. These questions address important issues beyond our study, as according to a recent survey, two thirds of students in secondary schools start their L2 study either in a new language or as false beginners (Vágó, 2007). An equally important issue concerns how teachers adjust their methodology to learners' abilities. These questions need further enquiries. Acknowledgements The language assessment projects were supported by the Hungarian Ministry of Education and carried out by the Research Group on the Development of Competencies, Hungarian Academy of Sciences (MTA-SZTE Képességkutató Csoport). The authors are grateful to two anonymous reviewers for the insightful and helpful comments and suggestions. References Alderson, J. C. (1984). Reading in a foreign language: A reading problem or a language problem? In J. C. Alderson, & U. H. Urquhart (Eds.), Reading in a foreign language (pp. 122–135). New York: Longman. Alderson, J. C. (2000). Assessing reading. Cambridge: Cambridge University Press. Alderson, J. C., Clapham, C., & Wall, D. (1995). Language test construction and evaluation. Cambridge: Cambridge University Press. Andor, M. (2000). A nyelvtudás szociális háttere [Social background to language knowledge]. Educatio, 9(4), 717–728. Bachman, L. (1990). Fundamental considerations in language testing. Oxford: Oxford University Press. Bachman, L., & Palmer, A. (1996). Language testing in practice. Oxford: Oxford University Press. Canale, M. (1983). From communicative competence to language pedagogy. In J. Richards, & J. Schmidt (Eds.), Language and communication (pp. 21–43). London: Longman. Canale, M., & Swain, M. (1980). Theoretical bases of communicative approaches to second language teaching and testing. Applied Linguistics, 1, 1–47. Carroll, J. (1990). Cognitive abilities in foreign language aptitude: Then and now. In T. Parry, & C. Stansfield (Eds.), Language aptitude reconsidered (pp. 1–29). Englewood Cliffs, NJ: Prentice Hall. Carroll, J. B. (1993). Human cognitive abilities. Cambridge: Cambridge University Press. Council of Europe (2001). Common European framework of reference for languages: Learning, teaching, assessment. Cambridge: Cambridge University Press. Csapó, B. (1997). The development of inductive reasoning: Cross-sectional assessments in an educational context. International Journal of Behavioral Development, 20(4), 609–626. Csapó, B. (Ed.). (1998). Az iskolai tudás [School knowledge] Budapest: Osiris. Csapó, B. (2001). Az induktív gondolkodás fejlődésének elemzése országos reprezentatív felmérés alapján [An analysis of the development of inductive reasoning skills based on a study involving a nation-wide representative sample]. Magyar Pedagógia, 3, sz. 373–391. Csapó, B. (Ed.). (2002). Az iskolai műveltség [School literacy] Budapest: Osiris. Csapó, B., & Nikolov, M. (2001, July). Hungarian students' performances on English and German tests. Paper presented at the European Language Testing in a Global Context Conference. Universitat Pompeu Fabra, Barcelona, Spain. Csapó, B., & Nikolov, M., (2002, April). The relationship between students' foreign language achievement and general thinking skills. Paper presented at the American Educational Research Association Annual Meeting, New Orleans, U. S. A. 218 B. Csapó, M. Nikolov / Learning and Individual Differences 19 (2009) 209–218 Cummins, J. (1991). Interdependence of first- and second-language proficiency in bilingual children. In E. Bialystok (Ed.), Language processing in bilingual children (pp. 70–89). Cambridge: Cambridge University Press. Cummins, J., & Swain, M. (1986). Bilingualism in education. New York: Longman. Dörnyei, Z., Csizér, K., & Németh, N. (2006). Motivation, language attitudes and globalization: A Hungarian perspective. Clevedon: Multilingual Matters. Dörnyei, Z., & Skehan, P. (2002). Individual differences in second language learning. In C. J. Doughty, & M. H. Long (Eds.), Handbook of second language acquisition (pp. 589–630). Oxford: Blackwell. Ellis, R. (1994). The study of second language acquisition. Oxford: Oxford University Press. Fekete, H., Major, É., & Nikolov, M. (Eds.). (1999). English language education in Hungary: A baseline study Budapest: British Council. García Mayo, M. P., & García Lecumberri, M. L. (2003). Age and the acquisition of English as a foreign language Clevedon, Avon: Multilingual Matters. Gardner, R. C., & MacIntyre, P. D. (1992). A student's contributions to second language learning. Part I: Cognitive variables. Language Teaching, 25, 211–220. Gardner, R. C., & MacIntyre, P. D. (1993). A student's contributions to second-language learning. Part 2: Affective variables. Language Teaching, 26, 1–11. Grigorienko, E. L., Sternberg, J. R., & Ehrman, M. E. (2000). A theory-based approach to the measurement of foreign language learning ability: The Canal-F theory and test. The Modern Language Journal, 84, 390–405. Harley, B., & Hart, D. (1997). Language aptitude and second language proficiency in classroom learners of different starting ages. Studies in Second Language Acquisition, 19(4), 379–400. Kiss, Cs., & Nikolov, M. (2005). Preparing, piloting and validating an instrument to measure young learners' aptitude. Language Learning, 55, 99–150. Klauer, K.J., (1997, August). Enhancing learning and cognitive ability through a training of inductive reasoning. Paper presented at the 7th European Conference on Learning and Instruction, Athens, Greece. Klauer, K. J. (1999). On the impact of inductive reasoning on the acquisition of abstract generic knowledge with elementary school and with learning disabled children. Psychologie in Erziehung und Unterricht, 46(1), 7–28. Klauer, K. J., & Phye, G. D. (2008). Inductive reasoning: A training approach. Review of Educational Research, 78(1), 85–123. Klauer, K. J., Willmes, K., & Phye, G. D. (2002). Inducing inductive reasoning: Does it transfer to fluid intelligence? Contemporary Educational Psychology, 27(1), 1–25. McDonough, S. H. (1999). Learner strategies. Language Teaching, 32, 1–18. Medgyes, P., & Miklósy, K. (2000). The language situation in Hungary. Current Issues in Language Planning, 1(2), 148–242. Muñoz, C. (2006). The effects of age on foreign language learning: the BAF project. In: Age and the rate of foreign language learning Clevedon, Avon: Multilingual Matters. Nikolov, M. (1999). Classroom observation project. In H. Fekete, É. Major, & M. Nikolov (Eds.), English language education in Hungary: A baseline study (pp. 221–246). Budapest: British Council. Nikolov, M. (2000). Research into early second language acquisition. In J. Moon, & M. Nikolov (Eds.), Research into teaching English to young learners: International perspectives (pp. 7–43). Pécs: University Press Pécs. Nikolov, M. (2003). Angolul és németül tanuló diákok nyelvtanulási attitűdje és motivációja [Attitudes and motivation of learners of English and German]. Iskolakultúra, 13(8), 61–73. Nikolov, M., & Csapó, B., (2002, April). Twelve-year-olds' attitudes towards classroom activities and their performances on tests of English and German as a foreign language. Paper presented at the American Association of Applied Linguists Annual Conference, Salt Lake City, U. S. A. Nikolov, M., & Józsa, K. (2006). Relationships between language achievements in English and German and classroom-related variables. In M. Nikolov, & J. Horváth (Eds.), UPRT 2006: Empirical studies in English applied linguistics (pp. 197–224). Pécs: Lingua Franca Csoport. Nikolov, M., & Mihaljevic-Djigunovic, J. (2006). Recent research on age, second language acquisition, and early foreign language learning. Annual Review of Applied Linguistics, 26, 234–260. Ottó, I. (1996). Language aptitude testing: Unveiling the mystery. Novelty, 3(3), 6–20. Ottó, I., & Nikolov, M. (2003). Magyar felsőoktatási intézmények elsőéves hallgatóinak nyelvérzéke [Language learning aptitude of Hungarian first-year university students]. Iskolakultúra, 13(6–7), 34–44. Oxford, R. L., & Ehrman, M. (1993). Second language research on individual differences. Annual Review of Applied Linguistics, 13, 188–205. Parry, T., & Stansfield, C. (Eds.). (1990). Language aptitude reconsidered Englewood Cliffs, N.J.: Prentice Hall. Pellegrino, J. W., & Glaser, R. (1982). Analyzing aptitudes for learning: Inductive reasoning. In R. Glaser (Ed.), Advances in instructional psychology, Vol. 2. (pp. 269–345). Hillsdale: Erlbaum. Resing, W. C. M. (1993). Measuring inductive reasoning skills: The construction of a learning potential test. In J. H. M. Hammers, K. Sijstma, & A. J. J. Ruijssenaars (Eds.), Learning potential assessment: Theoretical, methodological and practical issues (pp. 219–242). Amsterdam: Swets and Zeitlinger. Robinson, P. (2002). Learning conditions, aptitude complexes and SLA: A framework for research and pedagogy. In P. Robinson (Ed.), Individual differences and instructed language learning (pp. 113–136). Amsterdam: John Benjamins. Ropo, E. (1987). Skills for learning: A review of studies on inductive reasoning. Scandinavian Journal of Educational Research, 31, 31–39. Singleton, D., & Ryan, L. (2004). Language acquisition: The age factor, (2nd edition) Clevedon, Avon: Multilingual Matters. Skehan, P. (1991). Individual differences in second language learning. Studies in Second Language Acquisition, 13, 275–298. Skehan, P. (1998). A cognitive approach to language learning. Oxford: Oxford University Press. Skehan, P. (2002). Theorising and updating aptitude. In P. Robinson (Ed.), Individual differences and instructed language learning (pp. 69–94). Amsterdam: John Benjamins. Sparks, R., & Ganschow, L. (2001). Aptitude for learning a foreign language. Annual Review of Applied Linguistics, 21, 90–111. Spolsky, B. (2000). Language motivation revisited. Applied Linguistics, 21(2), 157–169. Sternberg, R. (2002). The theory of successful intelligence and its implications for language aptitude testing. In P. Robinson (Ed.), Individual differences and instructed language learning (pp. 13–44). Amsterdam: John Benjamins. Ullman, M. (2001). The neural basis of lexicon and grammar in first and second language: The declarative/procedural model. Bilingualism: Language and Cognition, 4, 105–122. Vágó, I. (2007). Nyelvtanulási utak Magyarországon [Language learning paths in Hungary]. In I. Vágó (Ed.), Fókuszban a nyelvtanulás (pp. 137–174). Budapest: Oktatáskutató és Fejlesztő Intézet.
© Copyright 2026 Paperzz