Journal of Research in Reading, ISSN 0141-0423 Volume 30, Issue 1, 2007, pp 38–58 DOI: 10.1111/j.1467-9817.2006.00333.x Playing football without a ball: language, reading and academic performance in a high-poverty school Elizabeth J. Pretorius Department of Linguistics, University of South Africa Deborah Maphoko Mampuru Department of African Languages, University of South Africa Second language (L2) reading research suggests that there is a complex interplay between L2 proficiency, first language (L1) reading and L2 reading. However, not much is known about the effect of L1 proficiency on L1 reading, and of L1 reading on L2 reading, or vice versa, in bilingual settings when readers have few opportunities for extensive reading in their L1. The relationships between L1 (Northern Sotho) and L2 (English) proficiency and L1 and L2 reading were examined in Grade 7 learners attending a high-poverty primary school in South Africa, during the course of a year when a reading intervention programme was implemented. The effect that attention to reading and accessibility of books had on the learners’ reading proficiency in both languages was examined, and the factors that predicted academic performance were analysed. When the learners were engaged in more reading, L2 reading contributed more variance to L1 reading than L1 proficiency. Reading in both languages also contributed significantly to academic performance. The study highlights the need for more cross-linguistic reading research in different educational settings. One of the central issues in second language (L2) reading research, first articulated by Alderson (1984), revolves round the question: to what extent is L2 reading a language or a reading problem? The African continent is characterised by linguistic diversity but due to its colonial past, the majority of learners in Sub-Saharan Africa do not do their schooling in their home language but through the medium of a former colonial language. If schooling does occur in the home language, it does so for a few years only, before switching to the former colonial language. Schooling in Africa tends therefore to be characterised by literacy acquisition in an L2, or by some form of bilingual schooling where initial literacy is acquired in the home language and a switch is later made to L2 literacy. Africa thus offers an ideal setting for studying the language/reading questions that preoccupy much of L2 reading research. Yet surprisingly little research has emerged from this rich context. In this article, we examine these language/reading questions from an African perspective, using data from a high-poverty primary school in an urban South African setting. To this r United Kingdom Literacy Association 2007. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA PLAYING FOOTBALL WITHOUT A BALL 39 debate we also add a rather perverse question, namely, to what extent is reading in a home language in the African context a language or a reading issue? Before moving on to the study itself, we first look at some of the issues in the bigger literacy landscape. Literacy assessments and reading research Worldwide, reading is used as an index of how well an education system is delivering on its mandate. There is a long and rich tradition of reading and literacy assessment and research in the developed world. Various linguistic, sociocultural, socioeconomic and developmental factors in home, school and classroom contexts have been identified that impact directly or indirectly on language development and reading achievement. The sheer number of academic journals in the English-speaking world dedicated to reading and writing research attest to this deeply embedded history of literacy research and the research-based knowledge that has mushroomed in the past six or so decades. In contrast, there has been relatively little research on literacy development in Africa, and no academic journal on the continent dedicated to reading. This is starting to change; the last decade has seen an increasing number of individual studies as well as large-scale national assessments being undertaken. Many of these findings point to much the same conclusion: the formal accomplishment of literacy does not happen easily for many learners in Africa. For example, the Southern African Consortium for Monitoring Educational Quality (SACMEQ) study compared literacy and numeracy levels of Grade 6 learners across 14 African countries between 2000 and 2002. The percentage of learners achieving minimal levels of literacy across the 14 countries was 44.1%. South African learners came 8th of the 14 countries, with only 36.7% of learners reaching minimum levels of mastery in reading at Grade 6 level (Mothibeli, 2005). The language factor is regularly cited as a reason for literacy underachievement in developing countries. By Year 6 most children in Sub-Saharan Africa have already switched over to an L2 as the language of learning and teaching (LoLT) if they have not already been using it since Year 1. There is indeed overwhelming research from both developed and developing countries that supports the position that schooling in the home language puts children at an advantage, especially in the early stages of literacy learning (e.g. Cummins, 2000; Thomas & Collier, 2002). Williams (1996), for example, compared the reading abilities of Malawian and Zambian children in the local language and in English at Year 5. The results showed that learning to read in their local language gave the Malawian children an advantage, while going straight for English and 5 years of exposure to literacy in English did not give the Zambian children an English reading edge over their Malawian peers. Such findings are consonant with research findings elsewhere. Cummins (2000) points out that there have been close to 150 empirical studies in the past 30 years that ‘show beneficial effects of additive bilingualism on students’ linguistic, cognitive or academic growth’ (2000, p. 37). However, studies of literacy in African home languages may paint a rather bleak picture. For example, the Eritrean survey of reading achievement in Grades 1, 3 and 5 in all eight languages of the country found that two-thirds of Grade 1 children could hardly read in the local languages used for instructional purposes at the elementary level. By Grade 5, while local language decoding skills were fairly competent, comprehension skills were not well developed (Walter & Davis, 2005). Similarly, in an audit by the Department of Education in 2001 across all nine provinces in South Africa, Grade 3 r United Kingdom Literacy Association 2007 40 PRETORIUS and MAMPURU learners had a mean of 38% on a reading comprehension test in their home language (Education Library Information and Technology Services, 2003). In all these studies, the language factor becomes submerged by a host of other factors. Variables such as poorly resourced schools, inappropriate instructional methods, printpoor environments, overcrowded classrooms, reduced time-on-task and poorly trained teachers are consistently identified as impacting negatively on literacy accomplishment (e.g. Grade 6 Systemic Evaluation Report, 2005; Walter & Davis, 2005). A further issue is the way in which the complex linguistic situation affects literacy development. For many children, the local language may not be their ‘real’ mother tongue. Geographic and social varieties considered non-standard are not used as media of instruction, and so only children familiar with the standard variety may be advantaged. Given the importance of the learning context, discussions of literacy accomplishment in Africa must take into account three important macro variables, namely, the complex linguistic fabric of African states and communities, widespread poverty and generally low adult literacy levels. All these variables impact on the educational context and give it diverse shapes and outcomes, depending on local conditions. Language diversity The African continent has a rich language heritage. Although many of these languages have developed orthographies, their written usage often does not extend to civic and social functions but is restricted to religious and educational texts, especially in the early grades. Invariably, the majority of learners do not have the advantage of being able to do all – or even some of – their schooling in their home language. South Africa, for example, has 11 official languages: Zulu, Xhosa, Swazi and Ndebele (which belong to the Nguni language group), N. Sotho, S. Sotho and Tswana (which belong to the Sotho language group) and the smaller language groups of Tsonga and Venda, as well as English and Afrikaans. Literacy is usually first developed in the home language in Grades 1–3, with English introduced as an additional language in Grade 1 or 2 and becoming the LoLT in Grade 4 in most schools. African languages are not used as LoLTs after Grade 3 or 4, but African children are expected to continue studying their home language as a school subject until Grade 12. Although initial schooling in the home language is encouraged by the Department of Education, school governing bodies have a choice and can opt for a ‘straight for English’ policy. Because English is regarded as the language of status, opportunity and education, the misperception often persists among parents that primary schools that offer ‘straight for English’ will provide a good education for their children. Poverty Africa in general has been experiencing economic problems for decades and has difficulty competing in the global economy. Poverty is widespread. Worldwide, research shows that it is not easy to educate poor children. The variables that co-occur with poverty do not create enabling contexts for literacy development. Formal learning environments in Africa tend to be framed by disadvantage: schools have inadequate physical resources, overcrowded classrooms and inadequate supplies of learning materials and books. Many children come to school hungry, and many schools in South Africa have feeding schemes where they supply one meal a day to children. Taken together, these conditions are not conducive to promoting literacy development in classrooms. r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 41 Literacy levels Although at 85.6% the adult literacy rate in South Africa (age 15 and above) seems relatively healthy, this reflects a basic literacy rate, i.e. being able to read and write with understanding a simple statement in relation to a person’s life (United Nations Report, 2003). Although schooling is now compulsory until Grade 9, only 51% of the South African population has reached high school. Co-occurring with poverty and low literacy levels is a print-poor environment and a vulnerable publishing industry which, as elsewhere on the continent, relies primarily on the education market for its survival. According to Wafawarowa (2000), up to 95% of all books sold in Africa are education books, compared with 35% in Europe. Reading for pleasure is not common. For many children from disadvantaged communities, literacy experiences mainly occur in the schooling context and even that does not generally provide a print-rich environment. In South Africa, only 27% of schools have school libraries. There are also relatively few children’s books printed in the African languages, and many of them are translations from English. There are even fewer non-fiction books. Publishers are reluctant to print books in African languages for which there is not a market. Even when African learners have the advantage of acquiring initial literacy in their home language, they do so in a print-poor environment. To use a rather crude analogy, this is akin to learning to play football without a ball. While studies in developed countries have found that L1 reading is an important predictor of L2 reading, the nature of, and the relationships between, learners’ language and reading abilities in their L1 and L2 become rather muddied in the African educational playing fields. Revisiting the language/reading issue in L2 reading research Two main hypotheses (cf. Bernhardt & Kamil, 1995; Cummins, 2000) underlie the language/reading questions in L2 reading research. The Linguistic Interdependence Hypothesis (or Common Underlying Hypothesis): This position holds that literacy operations and constructs transfer across languages and do not have to be ‘relearned’ in the other language. This type of linguistic transfer relates more commonly to cognitive/academic aspects of L1 and L2 proficiency, although transfer of communicative competence across languages has also been reported for children in Spanish–English, Japanese–English and Turkish–Dutch (cited in Cummins, 2000, p. 184). Cummins (2000) reports extensively on the consistent empirical support for this hypothesis, both in the USA and elsewhere. For example, a Californian study (California State Department of Education, 1985, cited in Cummins, 2000, p. 183), found higher correlations between L2 English and L1 Spanish reading skills than between English reading and English oral language skills. However, it was also found that the strength of L1/L2 reading relationship increased as English communicative skills increased. Although the hypothesis is couched in general terms so as to allow for the possibility of bidirectional transfer, it is generally assumed that the direction of transfer is from the L1 to the L2. Studies examining the reverse situation are rare. Serpell (1989) noted that educated Zambians who were skilled readers in English often had difficulty reading in the local languages that they spoke. However, his comments were based on anecdotal, not empirical, evidence. A longitudinal study by Wagner, Spratt and Ezzaki (1989, cited in r United Kingdom Literacy Association 2007 42 PRETORIUS and MAMPURU Williams and Cooke, 2002) found that L1 Berber students became proficient readers in Arabic L2 after 5 years. The researchers argued that children who are obliged to learn to read in an L2 are not necessarily disadvantaged. However, as Williams and Cooke (2002) point out, the Berber children had the advantage of being heavily exposed to Arabic outside school hours. The Linguistic Threshold Hypothesis (the ‘short-circuit’ hypothesis): According to this position, a threshold level of L2 proficiency is necessary for L1 reading skills and knowledge to transfer to L2 reading. A lack of L2 knowledge will short-circuit the use of L1 reading skills. As Bernhardt and Kamil (1995, p. 17) point out, this hypothesis assumes that ‘language is the key factor in reading activities . . . in order to read a language, one has to ‘know’ the language’. Several studies have examined the effects of L1 reading and L2 proficiency on L2 reading. For example, Bossers (1991) found that Dutch L2 proficiency accounted for 54% of the variance in L2 reading among L1 Turkish students, with L1 reading contributing only 19%. However, reliance on L2 proficiency decreased as L2 reading ability increased; for the skilled L2 readers, L1 reading was the only significant predictor. Using L1 English students learning French as L2, Brisbois (1995) found that L1 reading knowledge can transfer, but only after learners attain a threshold of L2 knowledge. She found that L2 vocabulary contributed more of the variance in L2 reading than did L2 grammar. Like Bossers, she also found support for the Interdependence Hypothesis, in that the upper-level students used their L1 reading skills for L2 reading comprehension, In their study of 187 L1 English students reading in L2 Spanish, Bernhardt and Kamil (1995) concluded that while L2 language proficiency accounted for a greater proportion of the variance (around 30–38%), L1 reading also made a contribution (about 16%). They rephrase the questions around this topic as: how literate does an L2 reader need to be to make the L2 work, and how much L2 language knowledge does an L2 reader need to make L2 reading work? (Bernhardt & Kamil, 1995, p. 32). From the above brief sketch of the L2 reading research domain, it could be argued that research into differences between reading in a first and second language does not always throw light on literacy (under)achievement in developing countries, for three main reasons: 1. Much of the research assumes reading ability in the first language and examines transfer to the second language. As we have seen, many readers in Africa may never acquire literacy in their own language. Even when they do, there may be few opportunities for extensive reading in their own language. For many learners, reading skills may become more strongly developed in the L2, not in the L1. Although Cummins’ interdependence hypothesis allows for this in theory, the possible transfer effect from L2 reading to L1 reading is an under-researched domain. 2. In most of the research, the notion of reading in the first language is treated as unproblematic. However, as Williams (1996, p. 184) points out, ‘the question of the extent to which L1 reading predicts L2 reading can only be validly posed . . . where the putative L1 is the L1 of the readers’. In Africa, even though many children are initially taught to read in an African language, they themselves may speak another language, another dialect or a non-standard variety. Furthermore, in most studies on L1/L2 reading transfer, the L2 proficiency of the subjects is measured but not their L1 proficiency. However, if there are educational contexts in which the notion of L1 proficiency is not unproblematic, then L1 proficiency levels also need to be factored into the equation. r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 43 3. It is by now axiomatic that the context in which literacy is acquired is vitally important. Many of the reading studies cited in the research literature involve educational contexts in developed countries where educational resources, literacy levels, instructional methods and availability of L1 texts are not really ‘issues’. Does the transfer of L1 reading to L2 still occur when the educational context is less than ideal? In sum, there is a large body of research that shows that reading skills do transfer from L1 to L2, but the extent to which this happens depends on the L2 proficiency of the readers. There is also overwhelming evidence of the importance of developing literacy skills in the L1 before changing over to an L2 as the LoLT. However, there is a need for more cross-linguistic reading research in different educational settings. In this study we examine this view of L1 and L2 literacy with a very different, and younger, set of learners in a very different literacy context. We compare, over an 8-month period, the language proficiency and reading abilities of Grade 7 learners, in the local school language, Northern Sotho (henceforth N. Sotho), and in English. We examine the relationships between language proficiency and reading skill in both these languages as well as their relationship to the learners’ academic performance. It is hoped that the findings from this research can add to the growing body of research into reading in developing countries where the educational, sociopolitical and sociolinguistic contexts differ so strikingly from those in developed countries. Rationale for current study Research over a 3-year period (2002–2004) in high-poverty high schools in a South African township revealed that the majority of Grade 8 learners were coming into high school with very poor reading skills, regardless of whether they were reading in their home language or in English (Matjila & Pretorius, 2004). The findings suggested, inter alia, that inadequate attention was being given to the development of reading in the primary schools, in both the African languages and in English. On hearing about the high school reading study, the principal of a neighbouring primary school contacted the first author of this article and asked if a reading project could be initiated at her school. It was decided to pilot a 3-year reading project in this primary school, to examine more closely the nature of reading problems in the school, to make books available and to see how reading challenges could be addressed. Aspects of this project during its first year of implementation are described and the broader school context in which reading is developed is sketched out. Research questions The research reported here is exploratory. The specific research questions that inform this paper are: 1. What are the differences in language and reading test scores in the local school language, N. Sotho (L1), and in English (L2)? 2. Does attention to reading and greater access to books improve reading test scores among the Grade 7 learners in N. Sotho and in English? 3. What is the relationship between language and reading ability in N. Sotho and in English, and does it change during the course of the year? r United Kingdom Literacy Association 2007 44 PRETORIUS and MAMPURU 4. What is the relationship between L1 and L2 language, L1 and L2 reading ability and school performance (as measured by the final examination results)? Method Design and aims of project The reading project was launched in 2005 at Batho Pele Primary School.1 The basic aims of the project are to create a culture of reading at the school, to raise reading levels in the local school language and in English, and thereby ultimately improve school performance. A multi-level approach has been adopted that involves the participation of the learners, teachers and parents. The project includes building resources as well as building capacity. The resource-building component entails making books readily accessible to learners and teachers by building up the school library and establishing collections of books in classrooms. The capacity-building component includes building capacity among teachers (by raising awareness of the importance of reading, and introducing them to various reading methods), learners (by encouraging them to read and instilling in them an enjoyment of reading) and parents (by encouraging them to take an interest in their children’s literacy development). Broader school context In the high-density urbanised townships in Gauteng Province there are mixed and complex linguistic communities. African language teachers often complain that their learners do not speak their home languages ‘properly’ but speak a mixture of African languages. In the township where this study is being conducted there are 26 primary schools, of which 10 are predominantly N. Sotho speaking, 9 Tswana, 3 Zulu, 2 Tsonga, 1 Venda and 1 South Sotho. School context The school has over 600 learners and a staff of 16 teachers plus the principal. The school serves a socioeconomically disadvantaged community and school fees are R120 (about $20) per annum. Only about 40% of the parents pay full school fees. The school has a feeding scheme, where 400 children ranging from Grades R to 7 are fed once a day. There are two classes at each grade level. In the Foundation Phase, there are about 35 children per class. This increases to about 54 per class in Grade 7. N. Sotho is the initial language of learning and teaching from Grade R to Grade 3, after which English becomes the LoLT. N. Sotho then becomes a subject of instruction from Grades 4 to Grade 7. About 70% of the learners at this school come from primarily N. Sotho-speaking homes. Reading context before intervention Initial observations of the classrooms revealed that very few books were visible. There were no bookshelves or storybooks in any of the classrooms. Besides school textbooks and exercise books, additional access to print material of an educational or recreational nature was almost non-existent. Although a small room was officially designated the library, it was used as a storeroom and as an office by a senior teacher. It contained rows of dusty shelves stacked with old textbooks and some children’s books, mainly in r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 45 English, many outdated and many age inappropriate. There was no cataloguing or lending system. The school had appointed a library committee the previous year and although these teachers had tried to ‘sort out’ the library, they felt ‘lost’ as they had never before been involved in such a task. Before the start of the project the teachers were asked to fill in a questionnaire that investigated their perceptions of and attitudes to reading and their home and school reading practices. Of the 17 members of staff, 15 had a post-Grade 12 2-year primary school teacher’s certificate qualification, and 3 had a postgraduate honours degree (the principal, vice-principal and one of the N. Sotho teachers). Several teachers were studying to upgrade their qualifications. None of the teachers was a member of a community library and 11 of the teachers indicated that they had 10 or fewer books in the home. In sum, there was a feeling at the school that learners at all grade levels struggled with reading but the teachers did not know how to address this problem. Building up resources Because the learners came from a disadvantaged community, priority was given to establishing a ‘proper’ school library. A computer and library software programme were purchased. An unemployed young man from the community with basic computer skills was appointed as school librarian and trained to catalogue and enter all the books on the computer system. New books were purchased in N. Sotho and English with project funding. Learners were encouraged to become members of the local community library in the township and teachers were encouraged to build up collections of reading material in their classrooms. After 9 months, all the learners and teachers were issued with school library cards. There is at present a collection of over 2,500 books in the library. Of these, 168 books are in N. Sotho (109 titles), and the rest are in English. All the non-fiction and reference books are in English (with the exception of N. Sotho dictionaries). There are no non-fiction books available in N. Sotho, and most of the 109 N. Sotho titles are storybooks for children ranging from preschool to about 10 years. There are very few storybooks in N. Sotho for teenagers. Building teacher capacity Regular workshops were held one afternoon a fortnight for all members of staff, to raise awareness of the importance of reading in the school context, to acquaint teachers with a broader frame of reference with regard to reading accomplishment and the factors that impact on its development, to acquaint them with sound instructional practices and to familiarise them with library practices and information literacy. Involving parents: the family literacy component Many parents in high-poverty areas have low literacy levels, are not familiar with the formal school context and often feel diffident about talking to teachers about their children’s progress at school. A series of Family Literacy workshops were held for parents in order to draw parents’ attention to the importance of reading and to encourage them to read to their children and/or to listen to them reading. The principal and teachers also attended the workshops. The head librarian at the local community library also addressed the parents at the workshop; she explained how the library works and how families can become members. Refreshments were served after each workshop to give parents an opportunity to talk to teachers in an informal context. r United Kingdom Literacy Association 2007 46 PRETORIUS and MAMPURU Monitoring and assessment In order to assess the efficacy of the programme it was important to determine baseline entry and exit reading levels during the first year of the project. Assessments were thus done at Grade 1 (N. Sotho) and Grade 7 levels (N. Sotho and English), soon after the start of the project in May (the pre-tests), and again in the last quarter in November (the posttests). The results of the literacy assessments were discussed with the principal and teaching staff after each round of assessment. Owing to space limitations, only the results of the Grade 7 assessments are reported here. Test design All the Grade 7 learners (n 5 104) were assessed. To facilitate comparisons across languages, the same tests were given in English and in N. Sotho. The tests were first developed in English from Grade 7 English books, then translated into N. Sotho, the translations being agreed by a team consisting of five specialists in language and education. Because the tests were the same in both languages, different sets of pre- and post-tests were designed. The N. Sotho tests were administered 4 weeks after the English tests. The time lag between testing in both languages in the pre- and post-tests was deemed long enough to minimise memory effects. Language test N. Sotho and English language proficiency were operationally defined as language proficiency obtained in a dictation test in the respective languages. There are at present no standardised language tests in any official languages in South Africa for different age groups. It was felt that a dictation test would tap into ‘school’ language proficiency in both languages without requiring the learners to read a text. Even though learners may speak a mixture of African languages on the playgrounds and streets, they are taught standard N. Sotho in their language classes. To ensure that the level of difficulty of the dictation test was in line with the age group, the dictation passages were taken from approved textbooks currently available for Grade 7 N. Sotho and English classes. A set of criteria was drawn up by the English and N. Sotho team members for marking the dictation passages. These were based on a cross-section of good, average and weak responses that were examined and discussed in a joint workshop. The learners were told beforehand that they needed to first listen to a passage read to them, and then write it down in their ‘best’ English/N. Sotho, paying attention also to spelling and punctuation. The passage was first read to the class at normal conversational pace while they simply listened. The second time it was read at conversational pace, but chunked into sections of about five to six words, which the learners wrote down. The chunks were not repeated. This kind of task meets the two naturalness criteria for pragmatic natural processing language tasks, viz. it requires the processing of temporally constrained sequences of linguistic material and, in order to divide the stream of speech into identifiable chunks for writing down, it also requires, to a large degree, understanding the meaning of what was heard (Oller, 1979, p. 39). Oller further argues that despite criticism of the use of dictation as a language test, ‘it does correlate at surprisingly high levels with a vast array of other language tests’ (Oller, 1979, p. 58). r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 47 Reading tests N. Sotho and English reading proficiency were operationally defined as proficiency obtained in a reading comprehension test. Two different tests were used for the pre- and post-tests. To try to maintain similar levels of difficulty between the pre-and post-tests, the text passages in both tests were taken from Grade 7 textbooks currently used in South Africa. The test passages were taken from the English Grade 7 language textbooks written for L2 students in South Africa. The tests were originally designed in English and then the passages and the questions were translated into N. Sotho by specialists in the Department of African Languages at the University of South Africa (including the second author of the article, who is also an experienced N. Sotho translator). The question formats were also identical. Finally, to allow for some kind of benchmark comparison, a 12-mark section of a test used to assess the reading levels of Grade 8 learners entering high school (Matjila & Pretorius, 2004), was also used for the Grade 7 post-test. All question formats were explained to the learners before they started the reading test and examples were given on the chalkboard, in the language of the comprehension test. The question formats were as follows: Inference questions. The ability to answer text-based inference questions rather than literal questions is a reliable indicator of how well a reader understands a text (e.g. Oakhill & Cain, 1998). There were no literal questions (i.e. direct matching of questions with information in the text), interpretive or creative comprehension questions. Cloze items. In each case, a modified cloze task was set, with approximately every ninth word deleted, if it was appropriate and could be inferred from the text. Although the same passages were used in English and N. Sotho for the cloze activities and the same number of deletions was designed, the same linguistic items were not deleted. Deletions were guided by the textual clues provided by the morphosyntactic and semantic features of each language text. Anaphoric resolution. The ability to resolve anaphoric references in a text is an integral part of reading skill (e.g. Webber, 1980). Specific anaphoric items were identified and the learners were required to underline the referents to which they referred. Reading rate With each expository passage, an informal measure of the learners’ reading rate was taken to determine, more or less, at what pace they were reading. After the test preliminaries, the learners were all told to start reading at the same time, and after a minute, they were stopped and asked to circle the word they had been reading when they were stopped. Thereafter, they continued reading the passage and answered the questions that followed. The number of words read within the minute gave the researchers a rough indication of the learners’ range of reading rate, as a group. Because it is difficult to accurately assess reading rate in large groups, the scores are treated with caution. Administrative procedures The English tests were administered first. The pre-tests were administered in May 2005, while the post-tests were written in November 2005. Both sets of tests were written r United Kingdom Literacy Association 2007 48 PRETORIUS and MAMPURU during two periods allocated during school hours and administered by the project researchers. No specific time limits were set for completion of the tests. Results The test data were entered on SPSS. Using the Cronbach alpha model, the reliability scores for the English pre- and post-tests were .77 and .74 respectively, while the alpha scores for the N. Sotho pre- and post-tests were .73 and .75, respectively. These alpha scores are regarded as acceptable (George & Mallory, 2003, p. 231). The mean age of the Grade 7 learners was 13.1 years (the youngest was 11 and the oldest 16). The majority had N. Sotho as their home language, with a minority of learners (18%) with different home languages (9% spoke a Sotho-related language, 6% spoke a Nguni-related language and 3% spoke the minority Venda and Tsonga languages). Research question 1: What are the differences in language and reading test scores in N. Sotho and English? Table 1 shows mean differences in language, reading and reading rate scores in N. Sotho and in English in the pre- and post-tests. Four interesting trends emerge from these descriptive results. Firstly, mean reading comprehension scores on the pre-test results were particularly low, and they were very similar in both N. Sotho and English, irrespective of the language proficiency scores in both languages. Secondly, reading comprehension scores were higher in L2 than in L1 at the end of the year. Thirdly, there was a large gap between language scores and reading scores in both languages; the gaps narrowed somewhat in the post-tests. For example, even though learners scored quite high for L1 language, their reading scores in L1 were low; the gap between L1 language and L1 reading reduced from about 40% to 30% in the post-test, even though this was a fairly homogeneous group of learners, linguistically, who had acquired literacy in the L1 from Grade 1. Fourthly, although reading rates were slow, the learners were reading slower in N. Sotho than they were in English at the end of the year. This is consistent with Table 1. Mean scores (SD in parentheses) for language, reading comprehension and reading rate in English and N. Sotho. 2005 Language (%) (SD) Minimum Maximum Reading comprehension (%) (SD) Minimum Maximum Reading rate (wpm) Minimum Maximum 2005 English pre-tests English post-tests Gains N Sotho pre-tests N Sotho post-tests 60.2 (22.07) 2.9 92.6 29.5 (13.7) 3.8 62 98 49 167 62.7 (24.6) 0 97 42 (18.7) 4.7 88.1 106 46 166 2.5 70.8 (16.7) 1.8 94.5 30 (12.6) 0 71.4 – – – 68 (12.6) 1 95 37.2 (17.3) 3.2 84.7 93 21 177 12.5 Gains 2.8 7.2 r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 49 Table 2. Differences between pre- and post-test scores. Batho Pele Pre/post Pre/post Pre/post Pre/post differences differences differences differences in in in in L1 L1 L2 L2 language reading language reading T df Significance (two-tailed) 0.98 6.26 2.99 10.61 98 98 96 96 .327 .000 .004 .000 results from the high school study, where Grade 8 reading rates in the African languages were consistently lower than English reading rates (Matjila & Pretorius 2004). Research question 2: Will attention to reading and greater access to books improve reading test scores in both languages? As shown in Table 1, reading comprehension scores increased from pre- to post-tests in both languages, while language scores did not. Paired sample t-tests showed significant differences between pre- and post-test L2 proficiency scores (a small but significant gain of 2.5%) and reading scores (a significant gain of 12.5%). While L1 reading scores also showed a significant difference between pre- and post-test times (a 7.2% gain), L1 proficiency dropped slightly ( 2.8%). The paired t-test results are shown in Table 2. The fact that there was only a small gain in English language proficiency and no significant change in N. Sotho language proficiency during the same time period suggests that the changes in reading performance cannot simply be ascribed to maturational changes over time, as the language measures were not similarly affected by maturation. What was striking about the pre-test results was the large number of very low scores achieved by Grade 7 students (several of whom were over-age for their grade). For example, in the pre-tests there were 31 learners scoring under 20% for reading comprehension. (According to McCormick [1995], readers who achieve 60% or less for comprehension on standardised reading tests are reading at ‘frustration level’.) By November the number of very weak readers had dropped to six, and whereas only one student scored above 60% in L2 reading in the pre-test, by November there were 18 students who scored above 60% for L2 reading. Research question 3: What is the relationship between language and reading ability in N. Sotho and in English, and does it change during the course of the year? To explore the relationship between language and reading, Pearson Product Moment correlations between language proficiency and reading ability in N. Sotho and in English in pre- and post-tests were performed. These correlations are shown in Table 3. Although all the correlations were significant, it is interesting to note that L1 language did not show a strong link with L1 reading at the pre-test stage (r 5 .38); this increased to a moderate correlation by November (r 5 .54). In contrast, L2 language and L2 reading correlated strongly at both the pre- and post-test stages (r 5 .62 and r 5 .70, respectively). With regard to reading relationships, while pre- and post-test L1 reading scores correlated strongly (r 5 .65), this correlation was much stronger with pre- and post-test L2 reading scores (r 5 .76). Initially L1 reading showed a moderately low correlation with L2 reading (r 5 .41), but by the end of the year this relationship had strengthened considerably (r 5 .79). r United Kingdom Literacy Association 2007 50 PRETORIUS and MAMPURU Table 3. Correlation matrix showing relationships between language and reading comprehension in N. Sotho and English. 1. 2. 3. 4. 5. 6. 7. 8. Pre-L1 language Pre-L1 reading Pre-L2 language Pre-L2 reading Post-L1 language Post-L1 reading Post-L2 language Post-L2 reading Note: 2 3 4 5 6 7 8 .380** – .546** .466** – .287** .415** .616** – .636** .478** .388** .388** – .473** .648** .598** .598** .544** – .597** .502** .888** .594** .678** .623** – .597** .582** .681** .759** .497** .792** .696** – ** Correlation is significant at the .01 level two-tailed. Table 4. Simultaneous multiple regression predicting L2 reading. Pre-test b Significance .581 .182 .0005 .047 Predictor variables L2 language L1 reading 96 5 21.532, po.0005 Adjusted R2 5 .384 F3, Post-test F3, b Significance .606 .350 .0005 .0005 Predictor variables L1 reading L2 language 91 5 68.813, po.0005 Adjusted R2 5 .684 Table 5. Simultaneous multiple regression predicting L1 reading. Pre-test Predictor variables L1 language L2 reading F2, 5 15.934, p o .0005 Adjusted R2 5 .232 96 b Significance .286 .333 .002 .000 Post-test Predictor variables L1 language L2 reading F2, 5 93.000, po.0005 Adjusted R2 5 .644 98 b Significance .192 .696 .006 .000 Multiple regression analyses were used to tease apart these relationships more clearly to see which variables predicted L2 reading at both the pre- and the post-test stages. The dependent variable was L2 reading, and the predictor variables were L1 language, L1 reading and L2 language. Because the study was exploratory, the enter (simultaneous) method was used. As shown in Tables 4 and 5, significant models emerged for both the pre-test and the post-test stages, but the strength of the models and the contributions of the variables changed over time. The first regression explored the predictors of L2 reading at pre- and post-test (see Table 4). This analysis revealed that while L2 language accounted for 58% of the L2 reading variance at the start of the reading project (thus providing support for the Linguistic Threshold Hypothesis), by the end of the year the situation had flipped, and L1 reading now contributed 60% of the variance, with language only contributing 35% (thereby providing support for the Interdependence Hypothesis). However, given that L1 reading lagged behind L2 reading, and given too that the learners had far greater exposure to L2 texts than to L1 texts, it was decided to explore the data further by using a regression analysis to determine which variables predicted L1 r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 51 Table 6. Simultaneous multiple regression predicting academic performance. Pre-test Predictor variables L2 language L2 reading L1 reading F4, 5 39.869, p o .0005 Adjusted R2 5 .613 94 b Significance .408 .381 .177 .000 .000 .018 Post-test F4, Predictor variables L2 language L2 reading L1 reading 5 40.483, po.0005 Adjusted R2 5 .629 89 b Significance .401 .279 .235 .000 .016 .030 reading, at both the pre- and post-test stages. These analyses are shown in Table 5. Using the enter method, significant models emerged for both the pre-test and the post-test stages, but again the model only accounted for a small amount of the variance initially, and the contributions of the variables changed over time. This analysis revealed that at both pre- and post-test stages, L2 reading contributed more to the variance in L1 reading than did L1 language. While the L1 contributed 28% of the variance initially, this dropped to 19% at the post-test stage, when L2 reading became a very strong predictor, accounting for 69% of the variance by the end of the year. By this stage, the learners’ reading skills had improved in both languages, but they had had more exposure to a variety of L2 texts than to L1 texts. These results indicate that the direction of influence can go from L2 to L1. Research question 4: What is the relationship between L1 and L2 language, L1 and L2 reading ability and school performance (as measured in the final examination results)? The final research question explores the contributions made by L1 and L2 language, and L1 and L2 reading ability, to school performance, as measured in the final examination results. The examination results reflect the total scores derived from eight different Grade 7 subjects, namely English, N. Sotho, Afrikaans, Maths, Natural Science, Social Science, Life Orientation and Technology. A multiple regression analysis was performed, with L1 language, L1 reading, L2 language and L2 reading included in the model for both pre- and post-test stages, using the enter method. The results are shown in Table 6. In these analyses, L1 language did not account for a significant amount of the variance. While L2 language proficiency accounted for 40% of the variance at both pre- and posttest stages, it is interesting to note how L2 reading accounted for more of the variance at the pre-test (38%) than the post-test. At post-test, there was an increased contribution (23%) from L1 reading, with L2 reading now accounting for 28% of the variance. To examine reading distribution patterns over time, the learners’ performance on the language and reading tests were matched against their academic performance at school. The total mean scores for each learner, as reflected in the final November examinations, placed the learners into one of four achievement categories used by the Department of Education, namely Not Achieved (0–39%), Partly Achieved (40–49%), Achieved (50–69%) and Outstanding (70–100%). Although there is some dissatisfaction with these groupings, they are de facto categories used by all schools. The learners’ academic or ‘examination’ profile across the language and reading variables is shown in Table 7 below. The ‘gap’ column refers to the gap between performance in language and reading in both languages, while the ‘gains’ column reflects the gains in language and reading from pre- to post-test. r United Kingdom Literacy Association 2007 52 PRETORIUS and MAMPURU Table 7. Comparison of L1 and L2 language and reading proficiency across the four academic achievement categories. Academic category Language and reading assessment Not Achieved (n 5 11) L1 L1 L2 L2 L1 L1 L2 L2 L1 L1 L2 L2 L1 L1 L2 L2 Partly Achieved (n 5 15) Achieved (n 5 51) Outstanding (n 5 17) language reading language reading language reading language reading language reading language reading language reading language reading Pre-test mean 52.3 17.8 17.6 14.1 71.3 23.4 49.5 20.4 70.2 30.3 63.8 28.9 78.4 38.3 77.1 44.3 Pre-test gap 34.5 3.5 47.9 29.1 39.9 34.9 40.1 32.8 Post-test mean 39.3 15.3 22.9 18.7 67.7 23.9 52.2 31.7 71.7 39.7 66.2 42.1 77.2 54.9 86.4 67.6 Post-test gap 24 4.2 43.8 20.5 32 24.1 22.3 18.8 Gains 13 2.5 5.3 4.6 3.6 .05 2.7 11.3 1.5 9.4 2.4 13.2 1.2 16.6 9.3 23.3 It is interesting to note, in the pre-tests, the large gaps between achievement in L1 language and L1 reading, as also the large gaps, but to a slightly lesser extent, in L2 language and L2 reading. As reading skills improved, the gap between language and reading performance slowly started closing. This did not happen equally for all the learners. The learners in the Not Achieved and Partly Achieved groups actually lost some L1 language and only made very small gains in L2 language. While the learners in the Partly Achieved group showed an 11.3% improvement in L2 reading, they did not show any upward mobility in their L1 language and reading. In contrast, learners who passed Grade 7 were learners who showed improvements in both L1 and L2 reading. Finally, a discriminant analysis was performed with academic performance as the dependent variable and L1 language, L1 reading, L2 language and L2 reading as the predictor variables. A total of 94 cases were analysed. Because the groups were unequal in size, the analysis was computed from group sizes. Univariate ANOVAs showed that the four academic groups differed significantly on each of the four predictor variables, with the value of the discriminant function differing significantly for the four academic groups (w2, 12df, 5 109.39, p 5 0.000). The correlations between predictor variables and the discriminant function suggest that L2 language (.534), L2 reading (.512) and L1 reading (.291) were the best predictors of academic performance. L1 language did not emerge as a significant predictor. Overall, the discriminant function successfully predicted outcomes for 74.5% of cross-validated groups correctly classified. Accurate predictions were made for 63.6% of the Not Achieved learners, 33.3% of the Partially Achieved students, 86.3% of the Achieved students and 82.4% of the Outstanding students. The fact that the Partially Achieved students were less easy to predict is perhaps not surprising: although they show similar characteristics to the learners who failed (no upward mobility of L1 language or reading skills), they showed some features of the achieving learners, namely an increase in L2 reading, albeit a smaller increase. r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 53 Discussion The study reported here is part of a relatively small-scale, 3-year, work-in-progress project. Given that the study was largely project driven rather than research driven and that different tests were used in the pre-and post-tests, the examination of the data collected during the course of the first year was exploratory and the findings must be treated with some caution. The continuation of the reading project over the next 2 years may yet throw up further shifting patterns of relationship between L1 and L2 language and reading skills. Despite these limitations, we feel that the longitudinal data in the study throw up some interesting patterns underlying the complex dynamics of L1 and L2 language and reading proficiency levels in a disadvantaged educational setting that is fairly typical of many schools across South Africa and the rest of the continent. The study also identifies areas for further, more detailed research. Print environment and reading agenda at school The pre-test results are indicative of the kind of L1 and L2 language and reading accomplishments that can be expected in high-poverty schools when learners have limited exposure to texts, and little attention is given to actively nurturing reading abilities. Good readers should show a strong alignment of scores between language and reading performance. The large gap between performance on the language and the reading tests indicates that language proficiency alone, especially in the L1, does not guarantee ability to read in a language. The fact that the learners’ L1 reading was initially marginally better than L2 reading (except for the Outstanding learners whose L2 reading was in advance of their L1 reading) might have been due to the fact that the N. Sotho pre-test was written a month after the English pre-test. By then, the reading project was already into its third month of implementation and learners were starting to read more often. The post-test results are encouraging in that they show that the barriers to learning that poverty imposes can be overcome if learners have access to books and if the school puts reading on the daily agenda. By the end of the year, mean reading levels in both L1 and L2 had improved, even though mean L1 language levels had not increased, and L2 language levels only marginally so. Elley (1991) suggests that the effects of a reading programme may take more than a year before they impact on language and school performance. The changes in reading ability brought about interesting shifts in the relationships between language and reading proficiency in different groups of learners that are relevant to L2 reading research. These changes occurred as a result of changes in the print environment and in the reading agenda at the school. This suggests that any theory of L2 reading needs to be context sensitive. Current formulations of the Linguistic Interdependence and the Linguistic Threshold hypotheses assume a reading context (i.e. access to books and the development of L1 reading skills) that may not generally obtain in the formal learning contexts of the developing world. There are three notable patterns underlying the changes in reading at the school during the year, as discussed below. Teaching factor and negative Matthew effects In spite of having had N. Sotho as LoLT for 3 years in the foundation phase, the Grade 7 learners at Batho Pele had problems with reading comprehension in both N. Sotho and English. These findings point to the importance of the teaching factor and the long-term consequences that negative Matthew effects can have on reading. r United Kingdom Literacy Association 2007 54 PRETORIUS and MAMPURU Before the intervention, nothing much was being done about reading at Batho Pele, and the learners had very little access to texts in either the L1 or the L2. At pre-test time, 30% of the cohort of Grade 7 learners were reading below a 20% comprehension level. Several were over-age for their grade level, suggesting a previous ‘Not achieved’ track record during their primary schooling. When a few of the very weak Grade 7 readers were assessed individually, they were found to have poor knowledge of letter–sound relationships and weak word recognition skills, in both L1 and L2. They even had difficulty reading Grade 2 and 3 texts in N. Sotho. This possibly points to an inadequate phonics basis in the foundation phase. Research consistently shows that if sound decoding skills are not well established during the early primary school years, many learners find it difficult to catch up later (Adams, 1990; Chall, Jacobs & Baldwin, 1990; Paran, 1996). The seemingly modest mean gains made by the learners highlight the challenge of trying to simply catch up on basic reading skills as late as Grade 7, a stage when precious time and effort is needed to prepare children conceptually and linguistically for the demands of high school. Furthermore, the class sizes at Batho Pele were large (about 54 per class). The low mean post-test scores point to the challenges that lie ahead. Regular exposure to a variety of L1 and L2 texts is needed to push learners’ reading levels to more acceptable levels. The home language factor did not appear to affect performance. All the learners in the Not achieved group had N. Sotho as a home language; all the Grade 7 learners who were from a non-N. Sotho home background (18%) were in the Achieved group. The learners who failed at this school did so for other reasons, and not because they were not N. Sotho speaking: one of the reasons for failure relates to severe reading difficulties that were never attended to in the lower grades. Whether this pattern obtains at other linguistically heterogeneous schools is obviously an area that merits further research. Factors that predict reading proficiency The findings in this study provide some support for the Linguistic Threshold Hypothesis with regard to L2 reading. However, this hypothesis assumes that readers already possess L1 reading skills and knowledge that can be transferred if the L2 proficiency threshold is reached. The large gap between language proficiency and reading ability, especially in the L1, in this set of data suggests otherwise. In this study, like studies elsewhere, L2 proficiency emerged as a strong determinant of L2 reading initially. However, the longitudinal findings also reveal the shifting dynamics of this relationship; as the learners’ reading skills improved in both L1 and L2, these reading skills seemed to mutually support each other, with L2 proficiency having a smaller effect. But why did the weaker readers (in the Not Achieved and Partly Achieved categories) not make gains in their L1 proficiency or L1 reading over time, yet make some gains in L2 reading during the same period? Does a lack of L1 reading knowledge gradually short-circuit the use of L1 skills over time? If there is little L1 reading knowledge to transfer in the first place, do these learners compensate by developing basic reading skills in the L2 and then use these dominant L2 reading skills to bootstrap themselves to slightly higher levels – provided they have some kind of L2 proficiency threshold (as the Partly Achieved students did, but not the Not Achieved students)? The gap between language proficiency and reading ability in both languages is an area that needs to be further investigated in bilingual reading research in order to better understand the constraints of a proficiency threshold in reading. The findings in this study also lend some support to the Interdependence Hypothesis in that reading ability seemed to transfer across the languages. However, whereas this r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 55 hypothesis generally assumes that transfer occurs from L1 to L2, in this study transfer seemed to occur bidirectionally. As noted earlier, as the learners’ reading skills improved in both L1 and L2, these reading skills seemed to mutually support each other. However, given that the children had access to more and varied texts in the L2 and seemed to read more in the L2, and given too that their L2 reading skills developed more strongly by the end of the year than their L1 reading skills, it seems that their L2 reading was being brought to bear on their L1 reading (particularly among the Outstanding students), rather than the other way round. The study makes a contribution to bilingual reading research in that, surprisingly, the findings indicated that L1 proficiency did not significantly predict L1 reading performance. Instead, L2 reading was a far stronger predictor of L1 reading ability. This may be due to the fact that what little reading these learners do tends to be done in the L2. Reading skills developed in a dominant language can seem to support the development of reading skills in a language in which reading is not often done. This is obviously an area that merits further research. It may also be fruitful to look at the subtle interactions of language and reading factors among different achievement groups in bilingual settings, in both urban and rural areas. The issues around L1 proficiency and literacy in developing countries also need to be more closely examined – what is taught, how it is taught, how it is assessed, how dialectal variety is affected or accommodated in relation to standardised versions, whether learners become diglossic (‘mixed township slang’ for everyday interactions and ‘schooled’, standard N. Sotho for classroom use) and how these factors impact on reading in the home language are all issues that await future research. Oller (1979, p. 58) points out that scores on a dictation test correlate highly with numerous other language tests. Whether these correlations also obtain in education contexts characterised by linguistic diversity are issues that need closer scrutiny. These findings highlight the importance of examining more closely the ways in which factors such as quality of teaching, time spent on task, class size, availability of and access to books affect language and reading accomplishment in the L1 and L2. More carefully controlled quasi-experimental studies, the use of a variety of assessment tools, qualitative research involving classroom observations and case studies of individual learners would help us gain greater understanding of language and reading interactions in bilingual reading research in learning contexts that are so vastly different from those that generally prevail in more affluent developed countries. Factors that predict academic performance Both the regression and the discriminant analyses showed that while L2 proficiency is an important determinant of academic performance, reading ability is also crucially important. This applies to both L2 and L1 reading. There was clear evidence of positive Matthew effects in reading: only when gains were made in both L1 and L2 reading did learners make it into the Outstanding band of academic performance. These were also the learners who started to narrow the gap between language proficiency and reading ability in both languages. In addition, their mean L2 comprehension levels were well above 60%. Although the Achieved learners showed a 13% increase in L2 reading, they were still reading well below 50% in both languages and the gap between language and reading proficiency was still quite large. Findings such as these point to fruitful areas for future research. These findings underline the importance of linking school-based reading r United Kingdom Literacy Association 2007 56 PRETORIUS and MAMPURU research to academic performance, especially in developing countries where a strong literate culture is not yet firmly embedded within the school system and reading is regarded as an ‘add on’, to pass the time when all the important work has been done. Implications for education Soon after the newly elected democracy in South Africa in 1994, the basis was laid for a new outcomes-based approach to education, which forms the foundation of the transformed curriculum in South Africa. According to the revised curriculum, competence in the home language should be developed in order to provide ‘a strong curriculum to support the language of learning and teaching’ (Department of Education, 2002, p. 4). Despite explicit and official attention given to the development of home language literacy in principle, national policy does not yet translate well into action in the classroom. Various factors that are usually associated with literacy accomplishment do not seem to seem to come together felicitously in many schools in developing countries. There are still schools without libraries, schools where the delivery and distribution of textbooks is highly problematic, where little time is spent on reading tasks, where teachers use inappropriate instructional methods, where reading problems develop in the early grades and are left unattended through the higher grades and where children have minimal access to print materials. All these factors require urgent attention. Special planning and budgeting in particular needs to go into resolving the problem of availability of books in the home languages. Even if language policy provides for an additive bilingual programme, literacy development cannot occur in a print vacuum. Learners need to read in order to become good readers, and in order to read, they need exposure to a variety of print material. The magnitude of this problem is reflected at a micro level at Batho Pele. Even though advocacy for reading in the L1 is a central feature of the reading project, and even though there is funding to purchase L1 books, the fact is that the learners simply do not have exposure to as wide a variety of texts in the L1 as in the L2. The relevance of language policy in literacy development should not obscure the critical role of reading in the learning context and the need for constant exposure to texts. Doing one’s schooling in a home language does not necessarily make one a good reader or a good student. The quality of an education system depends vitally on access to books. The notion of learning to play football without a ball is preposterous to anyone familiar with soccer – it would cause outrage among the soccer fraternity. Yet the fact that many hundreds of thousands of learners are expected to acquire literacy skills without books is a common occurrence in many schools in developing countries, and one that hardly raises an eyebrow. This study indicates that putting the reading ball onto the educational playing field not only enhances skill but also makes the game more meaningful. Acknowledgements The ‘Reading is FUNdamental’ project, from which this research derives, is funded by the D.G. Murray Trust. This material is also based upon work supported by the National Research Foundation. We would also like to express our sincere gratitude to all the learners and staff at the school for participating so generously and willingly in the assessments. Thanks are also due to Dr L. Fletcher from the Department of Statistics, r United Kingdom Literacy Association 2007 PLAYING FOOTBALL WITHOUT A BALL 57 Unisa, who unstintingly helped with statistical advice. The feedback and suggestions by three anonymous reviewers are also gratefully acknowledged. Note 1. Not the real name of the school. References Adams, M.J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press. Alderson, J.C. (1984). Reading in a foreign language: A reading problem or a language problem? In J.C. Alderson & A. Urquhart (Eds.), Reading in a foreign language. (pp. 1–27). London: Longman. Bernhardt, E.B. & Kamil, M.L. (1995). Interpreting relationships between L1 and L2 reading: Consolidating the linguistic threshold and the linguistic interdependence hypotheses. Applied Linguistics, 16, 15–34. Bossers, B. (1991). On thresholds, ceilings and short-circuits: The relations between L1 reading, L2 reading and L2 knowledge. In J.H. Hulstijn & J.F. Matter (Eds.), AILA Review, 8, 45–60. Brisbois, J.E. (1995). Connections between first- and second-language reading. Journal of Reading Behavior, 27, 565–584. Chall, J.S., Jacobs, V.A. & Baldwin, L.E. (1990). The reading crisis: Why poor children fall behind. Cambridge, MA: Harvard University Press. Cummins, J. (2000). Language, power and pedagogy: Bilingual children in the crossfire. Clevedon: Multilingual Matters. Department of Education (2002). Revised National Curriculum statement grades R–9. Pretoria: Department of Education. Education Library Information and Technology Services (2003). A response to the recommendation of Grade 3 systemic evaluations. Study by ‘Quality assurance’. Elley, W.B. (1991). Acquiring literacy in a second language: The effect of book-based programmes. Language Learning, 41, 375–411. George, D. & Mallory, P. (2003). SPSS for windows. Boston, MA: Allyn & Bacon. Grade 6 Systemic Evaluation Report (2005). Pretoria: Department of Education. Matjila, D.S. & Pretorius, E.J. (2004). Bilingual and biliterate? An exploratory study of Grade 8 reading skills in Setswana and English. Per Linguam, 20, 1–21. McCormick, S. (1995). Instructing students who have literacy problems. Englewood Cliffs, NJ: Merril. Mothibeli, A. (2005). Cross-country achievement results from the SACMEQ II project – 2000 to 2002. A quantitative analysis of education systems in Southern and Eastern Africa. EduSource Data News No. 49/October 2005. www.edufound.org.za Oakhill, J. & Cain, K. (1998). Problems in text comprehension: Current perspectives and recent research. In P. Reitsma & L. Verhoeven (Eds.), Problems and interventions in literacy development. (pp. 177–192). Dordrecht: Kluwer Academic Publishers. Oller, J.W. (1979). Language tests at school. London: Longman. Paran, A. (1996). Reading in EFL: Facts and fictions. ELT Journal, 50, 25–34. Serpell, R. (1989). The cultural context of language learning: Problems confronting English teachers in Zambia. In C. Kennedy (Ed.), Language planning and English language teaching. (pp. 92–109). London: Prentice Hall. Thomas, W.P. & Collier, V.P. (2002). A national study of school effectiveness for language minority students’ long-term academic achievement. Berkeley, CA: Center for Research on Education, Diversity and Excellence, University of California. http://www.crede.org/research/llaa/1.1es.html. United Nations Report. Statistics Division – Millennium Indicator Database (2003). http://unstats.un.org/unsd/ mi/mi _series_resultsd.asp?rowID=591 (accessed 18 June, 2004). Wafawarowa, B. (2000). Book development policies in Africa. Meta-info Bulletin, 9, 15–16. Webber, B.L. (1980). Syntax beyond the sentence: Anaphora. In R.J. Spiro, B.C. Bruce & W.F. Brewer (Eds.), Theoretical issues in reading comprehension. Hillsdale, NJ: Lawrence Erlbaum. Walter, S.L. & Davis, P.M. (2005). The Eritrea national reading survey. Dallas, TX: SIL International. r United Kingdom Literacy Association 2007 58 PRETORIUS and MAMPURU Williams, E. (1996). Reading in two languages at Year five in African primary schools. Applied Linguistics, 17, 182–208. Williams, E. & Cooke, J. (2002). Pathways and labyrinths: Language and education in development. TESOL Quarterly, 36, 297–322. Received 3 April 2006; revised version received 2 August 2006. Address for correspondence: Elizabeth J. Pretorius, Department of Linguistics, University of South Africa, PO Box 392, Pretoria 0003, South Africa. E-mail: [email protected] r United Kingdom Literacy Association 2007
© Copyright 2025 Paperzz