See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/321648795

English Academic Writing Convergence for Academically Weaker Senior

Secondary School Students: Possibility or Pipe-Dream

Article  in  Journal of English for Academic Purposes · December 2017

DOI: 10.1016/j.jeap.2017.12.002

CITATIONS

0

READS

44

2 authors:

Some of the authors of this publication are also working on these related projects:

Academic Literacy Catch-Up at a Grade 11 Level in South Africa: Possibility or Pipe-dream? View project

Investigating the teacher shortage crisis in New Zealand: Are we really short of teachers or is the entry into the industry for overseas trained

teachers too big a barrier? View project

Tracey Jane Millin

University of Otago

13 PUBLICATIONS   11 CITATIONS   

SEE PROFILE

Mark Millin

University of Otago

9 PUBLICATIONS   21 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Tracey Jane Millin on 09 January 2018.

The user has requested enhancement of the downloaded file.

Journal of English for Academic Purposes 31 (2018) 1e17

Contents lists available at ScienceDirect

Journal of English for Academic Purposes

journal homepage: www.elsevier .com/locate/ jeap

English academic writing convergence for academicallyweaker senior secondary school students: Possibility or pipe-dream?

Tracey Millin a, *, Mark Millin b

a University of Stellenbosch, School of General Linguistics, Private Bag X1, Matieland, Stellenbosch, 7602, South Africab University of Otago, Department of Economics, PO Box 56, Dunedin, 9054, New Zealand

a r t i c l e i n f o

Article history:Received 9 February 2017Received in revised form 20 October 2017Accepted 7 December 2017

Keywords:Academic literacyAcademic writingReading to learnEnglish for academic purposesScaffolded learningAcademic convergence

Abbreviations: ANA, Annual National AssessmenTests; PIRLS, Progress in International Reading LiterEducational Quality; SES, socioeconomic status.* Corresponding author. Current: University of Ot

E-mail addresses: millintracey@gmail.com (T. Mi

https://doi.org/10.1016/j.jeap.2017.12.0021475-1585/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

This paper discusses key findings of a study which sought to assess the impact of theReading to Learn (RtL) literacy intervention on individual student performance as appliedto senior secondary school students at two schools in the Western Cape of South Africa.The RtL intervention was implemented against a backdrop of serious concerns about thestate of literacy development in schools in South Africa, especially amongst non-nativeEnglish-speaking students from low socioeconomic communities as well as migrantcommunities. By taking each student's written pieces of work, submitted at various stagesthroughout the academic school year, each piece of writing was assessed and codified,which allowed for a detailed examination of various patterns using both descriptive andinferential statistics. The salient finding was that students, who were shown to beacademically weaker pre-intervention, generally exhibited a greater overall improvementin their respective English writing skills throughout the intervention. Thus, with anappropriately targeted intervention (like RtL) a convergence or ‘catch-up’ effect mightlikely occur for classes with large cohorts of non-native English-speaking students, whoare immersed in English medium-of-instruction schools.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Higher education in The Republic of South Africa has shifted from elitism to mass production, resulting in a range ofstudents, with diverse skills, from varied ethnic backgrounds, being given access to university (Fraser & Killen, 2003). Thenumber of students exiting the secondary schooling system with a Bachelor's pass, a pass which provides entrance to uni-versity, has grown steadily over the past number of years. For example, according to the Department of Basic Education (DBE)in South Africa, the percentage of matriculants who qualified to study towards a Bachelor's degree was 14% in 2000, but hasgenerally grown as follows in subsequent years: 20.1% in 2008; 19.9% in 2009; 23.5% in 2010; 24.3% in 2011; 26.6% in 2012;30.6% in 2013; 28.3% in 2014; 20.8% in 2015, and 24.8% in 2016. However, entry into university carries with it the tacit

t; DBE, Department of Basic Education; FET, further education and training; NBTs, National Benchmarkacy Study; RtL, Reading to Learn; SACMEQ, Southern and Eastern African Consortium for Monitoring

ago, Higher Education Development Centre, 65/67 Union Place West, Dunedin, 9016, New Zealand.llin), markmillin@gmail.com (M. Millin).

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e172

assumption that students have the pre-requisite academic literacy skills necessary for university study. But, the skillsdeveloped at school appear not to match the skills required to succeed at university. Other studies which have also reportedon the under-preparedness of undergraduate students in South Africa are: Troskie-De Bruin (1999); Penrose (2002); VanSchalkwyk (2008) and Allardice (2013). According to Groenewald (2005), one-in-every-three students are at risk of drop-ping out of their university studies by the end of their first year due to university under-preparedness. Given that access tohigher education in South Africa has become one of the pre-eminent mechanisms for addressing problems of social injusticeand inequality, a relatively high dropout rate is of great concern. This concern is not limited to the South African contextthough. Other studies which have reported on the under-preparedness of students transitioning into higher educationoutside of South Africa are, for example: Greene and Forster (2003); Maloney (2003); Cukras (2006), and Nakata, Nakata, andChin (2008).

Under-preparedness of students, particularly amongst previously marginalised students, exiting the secondary schoolingsystem and transitioning into university is often seen as one of the biggest barriers to successful learning at university(McGhie, 2012; Van Dyk & Weideman, 2004). For clarity, under-preparedness in the context of this study refers to the weakacademic reading and writing skills being developed in the secondary schooling system. In other words, under-preparednessrefers to a lack of advanced reading and writing skills developed at school to enable students to read to learn independently,and consequently, demonstrate effectively through appropriatewriting conventions, that learning has taken place. Describingsecondary education in South Africa as a “high cost/high enrolment, yet low quality system” (Taylor, 2009, p. 12) is all themore distressing when contextualised against the backdrop of a growing body of research evidence, which paints a rathergloomy picture of literacy performance.

The United Nations Educational, Scientific and Cultural Organisation's (UNESCO) assessment of academic literacy skillsglobally has revealed that more than 75% of South African students in the primary schooling system are performing belowminimum international literacy benchmarks. The Progress in International Reading Literacy Study (PIRLS) shows that 43% ofSouth African Grade 5 students appear not to be developing basic reading skills (Howie, Van Staden, Tshele, Dowse, &Zimmerman, 2012). The latest PIRLS results are even more alarming with 78% of South African Grade 4 learners unable toreach lowest PIRLS reading comprehension benchmarks leaving South Africa placed last out of 50 countries for reading scores(Howie et al, 2016). The Southern and Eastern African Consortium for Monitoring Educational Quality (SACMEQ) reveals that25% of South African Grade 6 students are not acquiring basic literacy skills, and less than 40% of Grade 6 students achievebasic literacy skills only (UNESCO, 2011). Table 1 gives a summary of the more recent South African Annual NationalAssessment (ANA) literacy scores for the earlier years of schooling and the average national performance by languageaffiliation. Table 1 does not paint a positive picture for literacy skills development within the further education and training(FET) phase of the secondary schooling system, especially the First Additional Language cohort, which encompasses most ofthe students in this study. Poor literacy development in the earlier grades means the senior secondary school phase suffers,because insufficient literacy building blocks are developed during the earlier years of schooling.

A long term consequence of the data presented in Table 1 is that tertiary institutions inherit the deficient literacy skills ofthe basic education system, resulting in high dropout rates amongst undergraduate students. For example, during a pilot runof the National Benchmark Tests (NBTs) in 2009 e an entrance exam offered to students at universities across South Africa,46% of students who wrote the test were classified as students in need of an augmented or extended degree programme ifthey were to succeed at their studies. An additional 7% of students were classified as students in need of long-term academicsupport. Hence, 53% of the applicable undergraduate students (who supposedly hold a Bachelor's Pass) were in need ofacademic literacy support at university (MacGregor, 2009). If the support is not given, graduation rates decline as studentseither fail to finish a three-year degree in three years, or fail to graduate at all.

Letseka and Maile (2008) reported that the graduation rate in South Africa is approximately 15% e arguably one of thelowest graduation rates worldwide; hence, the need for greater literacy support. However, support programmes of variouskinds are costly to universities and students alike, further constraining an already over-burdened tertiary education sector.Alternatively, intensive academic literacy development support could be made available to students in senior- or upper-

Table 1South African annual national assessment literacy performance (2012e2014).

National Average(2012)

National Average(2013)

National Average(2014)

2012 2013 2014

English Home Language 0e49% 0e49% 0e49%Grade 3 52% 51% 56% 43% 43% 34%Grade 6 43% 59% 63% 61% 32% 23%Grade 9 43% 43% 48% 61% 63% 52%First Additional Language 0e49% 0e49% 0e49%Grade 6 36% 46% 45% 76% 59% 58%Grade 9 35% 33% 34% 79% 83% 82%

Notes: The South African DBE uses seven levels of learner achievement. Levels 1, 2 and 3 comprise learners who achieved a performance score of 0e49%;Levels 4 and 5 a score of 50e69%; Levels 6 and 7 a score of 70e100%. Table 1 shows the national average, and percentage of students who fell within the0e49% range of scores (academically at risk students) across all 9 provinces of South Africa from 2012 to 2014. The summary for the 2015 and 2016 academicyears are not yet available. Source: DBE (2014).

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 3

secondary school (FET phase, Grades 10e12) to ensure students are equipped with the requisite skills needed to transitioninto (and succeed at) university. In an attempt to investigate whether or not the observedmismatch in academic literacy skillsmight be remedied at a school level, this study sought to show howan appropriately targeted intervention, namely Reading toLearn (RtL), might be used to scaffold Grade 11 students' academic writing skills in two diverse school contexts (i.e., one highand one lower socioeconomic status or SES school) in the Western Cape of South Africa. The following research questioninformed the study design: Were there statistically significant improvements in the measurable component of Grade 11students' ‘academic literacy’ skills, as proxied by their (written) ‘literacy’ scores, when the RtL interventionwas used as part ofclassroom pedagogic practice?

2. An overview of reading to learn

RtL's pedagogic strategy was originally developed to address the unsatisfactory literacy performance of indigenousAustralian students, by improving both educational access and success, where academic reading and writing skills appearedto be key determinants of these educational outcomes. The theoretical frameworks which underpin RtL are (but not limitedto): Vygotsky (1978) (learning as a social process); Halliday (1978; 1989; 1996) (language as a text within a social context) andBernstein (1975; 1990; 1996) (education as a pedagogic discourse). Numerous studies in different country contexts haveshowed the effectiveness of RtL at accelerating the development of academic literacy skills (see, for example, Acevedo, 2010;McRae et al., 2000; Millin & Millin, 2014; Rose & Acevedo, 2006). Research shows that, through purposeful scaffolding, thelower (weaker) the base of students' literacy skills the greater the potential gains to be made, effectively democratising theclassroom (Millin & Millin, 2014; Millin, 2015; Rose & Acevedo, 2006; Rose, 2005; Rose, Rose, Farrington, & Page, 2008).

The scaffolding of language awareness takes place through a purposefully designed six-stage RtL cycle (see Fig. 1).1 It isimportant to highlight that the theoretical basis of RtL is what makes this intervention educationally powerful and includes,but is not limited to: mediated learning; scaffolding; working within each student's zone of proximal development; makinguse of both explicit and implicit teaching strategies; differentiated learning and assessment practices; adopting a processapproach to writing, and drawing on systemic functional grammar and genre pedagogy to make student's more aware oflanguage patterns within texts. However, to make these ‘abstract’ theoretical aspects of language teaching more accessible topractitioners, the six-stage cycle acts as a blue print to help teachers follow a more systematic approach to the teaching ofreading and/or writing skills. The six stages were designed to guide teachers, teachers whowere not necessarily familiar withintensive literacy remediation, especially at a senior-secondary school level, where students are often assumed to havealready grasped elementary aspects of academic reading and/or writinge an assumption or expectation of students that is alltoo often wrong. In other words, the RtL cycle guides teachers through the stages of a process approach to teaching readingand writing by breaking down the various tasks into smaller stages with their associated pedagogic activities for each stage.

In practice, not all classes will require the same amount of time to be devoted to each stage. Rather, the time spent withineach stage, is dependent upon student needs. The key here is differentiated learning. For example, if one cohort of studentsdemonstrate weaker sentence construction skills; or, are less able to construct a sound academic argument, then it stands toreason that the teacher would need to spendmore time scaffolding the appropriate skills for this area of writing. Each stage ofthe cycle represents different aspects of reading and writing that need remediation. Based on diagnostic testing, the teacherdecides which stages of the cycle to focus on more to meet the needs of the students. To offer an example from the study, thelower SES school required a greater amount of time to be dedicated to the Preparing beforeWriting stage of the argumentativeessay genre, largely because the class had not finished reading or discussing their set novel (which the assessment tasks werebased on). Thus, an extra lesson or twowas needed to discuss the set novel in more detail. This was important, because if thestudents did not understand what the novel was about; what thematic issues were important and did not understand howthe novel unfolded, then they would, in effect, have very little to write about.

3. Implementation of reading to learn

The classroom practice for this study entailed offering students an overview of two purposefully chosen (curriculumappropriate) genre's e the narrative and academic argument. Each genre was unpacked with a modelled text chosen todemonstrate genre-specific patterns and content (See Appendix B). This provided an opportunity for the overall meaning andpurpose of the text to be highlighted, showing students very clearly that each text's genre needed to fulfil a specific purpose,and that purpose was realised through very careful structuring of the text, as well as the use of specific language (discourse).Once the overall purpose was highlighted, text-based teaching took place making students aware of how each stage andphase of the genre was structured and what language tools could be used to identify each phase within a stage. Once eachphase within a stage was unpacked, students were required to replicate their own phase, which was handed in and marked(Individual Reconstruction stage). Detailed feedback was given, and students would have to attend to the comments toenhance their re-drafts (Independent Writing stage). The re-draft would be handed in alongside the construction of the nextstage of the essay.

1 For a brief explanation of the RtL stages, readers are directed to Millin and Millin (2014).

Fig. 1. Six-stage pedagogic cycle of reading to learn.Source: Acevedo and Rose (2007).

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e174

With detailed coverage of the two genres, aspects relating to field, tenor and mode were unpacked, as well as attention togrammar, spelling, punctuation, presentation, conjunctives, an awareness of referencing techniques (use of pronouns, articlesand demonstratives), and an awareness of language usage related to appraisal. With regards to teaching materials used, thesame narrative essay model text was used for all three classes. Although the model text for the argumentative essay wasdifferent for Classes A, B and Class C, the same essay structure was used for all three classes. The use of different model textswas purposeful so as to bemore context-appropriate and concrete for the students with respect to the different set novel eachclass was reading. Classroom materials are included in Appendix B.

4. Research methodology: design, sample selection & methods

This study made use of a small-scale longitudinal design, in which different classes or cohorts of students were studiedover time. No control groups were available for this study. But, given myriad differences in student characteristics, teachercharacteristics and pedagogic practices found in classrooms throughout the school system, there would arguably be noguarantee that the ‘control group’ would, in fact, function as such, even if one was available. Our concerns about the use ofcontrol groups in educational settings are echoed in Hulstijn (1997) and Taylor and Von Fintel (2016), who point out thevarious debates surrounding the limitations of using control groups in classroom-based research e one of the most difficultmethodological challenges in educational research is the (in)ability to keep all classroom-based variables constant. Accordingto Hulstijn (1997), this is not possible in normal classroom settings with very real heterogeneous language characteristics ofstudents. Further, given that the teachers were present in the classroom duringmany of the RtL sessions and had access to theRtL materials, there was no guarantee they would not adopt RtL strategies in their other grade-specific classes, which mostlikely would have contaminated the control groups, invalidating any inferences based on this type of quasi-experimentaldesign anyway.

This study adopted a purposive sampling procedure. School participants needed to be located close to the research team,because the study was longitudinal in nature and the research team needed to be able to ‘immerse’ themselves in the schoolon a weekly basis for an entire academic year. Given the use of a non-random sampling procedure, we do not attempt togeneralise our findings to the broader grade-appropriate South African school population. But, this being said, our samples ofstudents did have some interesting heterogeneous characteristics (e.g., differences in SES; school environment and language),broadly applicable to a large number of students within the school population.

We made use of three classes of Grade 11 students from two schools willing to participate in the study. The DBE in SouthAfrica currently uses a quintile system to rank schools according to SES, where quintiles 1 to 3 essentially comprise lower SESschools and quintiles 4 to 5 comprise higher SES schools. Two cohorts of students came from a school broadly categorised as aquintile 5 school, whereas the other cohort of students came from a school broadly categorised as a quintile 2 school. Becausepoorer academic literacy performance is prevalent in lower quintile schools, it stands to reason that the researchwas aimed ataddressing unequal literacy performance of largely non-native English speakers emanating from lower SES communities andschools. Hence, the choice of a lower SES school in the sample.

Increasingly, high SES schools in South Africa are providing scholarships to lower SES students as part of each school'sbroader ‘corporate’ social responsibility, whichmeans academic support is also needed for these lower SES students enrollingin high SES schools. This partly motivated our inclusion of a high SES school, but there was a greater purpose: to see if thepatterns of behaviour throughout the intervention would be similar or different for a broadly different group of students. Wewere most interested in the within sample (class) rates of change and used appropriate nonparametric tests (Friedman and

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 5

Wilcoxon signed-rank tests) to make inferences about whether or not there was an improvement in students' literacy scoresby class cohort.

There was another good reason for choosing three relatively heterogeneous groups of students (even Class A and Class Bfrom the same school site, had significantly different home-language characteristics e see the following paragraph for a briefexplanation). To test the reliability of our findings, it would be necessary to replicate the intervention among several samplesof students to alleviate concerns about any improvements in literacy outcomes purely being a chance occurrence. In otherwords, all things being equal, if the pattern of behaviour was broadly similar and replicated for all three classes over theintervention, one could reasonably attribute such change to RtL, and not some random occurrence.

The language of teaching, learning and assessment in both schools was English as stipulated by the DBE e both schools'language-in-education policy prescribed English as the medium of instruction. For the high SES school, one class comprisedfirst-language (home language) speakers of English (Class A), whereas the other class comprised second-language speakers ofEnglish (Class B). To elaborate further, unlike Class Awheremost of the students were English native speakers of English, ClassB students were either Afrikaans, isiZulu, or isiXhosa home language speakers. English was their second language. For thecohort from the lower SES school, all students were classed as second-language speakers of English (Class C), with isiZulu andisiXhosa being the students' home language. Within the South African Curriculum Assessment Framework, Classes B and Cuse the same assessment standards, because both are categorised as English First Additional Language classes. To offer acontextual explanation, all South African secondary schools use English as the medium of instruction, regardless of whether astudent is a native speaker of English or not. But to account for the difference in English language skills, English nativespeakers are enrolled in English Home Language classes. Any student that identifies as a non-native speaker of English isoffered the choice to enrol in an English First Additional Language class. The curriculum assessment standards are slightlydifferent between English Home Language and English First Additional Language classes to account for the difference inEnglish proficiency skills.

5. Data collection

The study set out to test the impact of RtL on individual student performance as one of many possible solutions to thegrowing literacy crisis in South African schools. More specifically, the study sought to investigate whether students wouldshow a statistically significant improvement in their academic writing skills as measured by the numerical scores (quanti-tative data) assigned to each piece of writing (qualitative data) submitted for assessment. Students' writing was assessed pre-intervention; during the process of implementing RtL, and post-intervention. An obvious question that readers might have atthis juncture, relates to the process of generating numerical scores fromwriting e a practice all too common in the everydayclassroom. In a day-and-age of ‘measurability’ and ‘standardisation’ of educational outcomes, this practice is commonplace,but nonetheless, represents an abyss of debate in research methodology circles. To explain, students' assessed pieces ofwriting were collected at various stages throughout the implementation of RtL. These pieces of writing were then assessedusing an analytical rubric (see Appendix A), which made use of codified categories to convert the qualitative information intoquantifiable measures (numerical scores). Doing so would facilitate the comparison with other studies of RtL using a similarmethodological approach and identical marking rubric.

We are aware of the debate surrounding the use of marking rubrics and issues related to subjectivity in awarding grades (aform of bias). Choinski, Mark, and Murphey (2003) claim that the use of marking rubrics can present an objective frameworkfor teachers to conduct assessment. However, Shay (2004; 2008) questions whether assessment can really be impartial,because the process of assessment involves socially situated decision making. The ‘literacy’ scores were generated using aspecially designed marking rubric, which in itself represented both objective and subjective elements for grading a student'sperformance. For instance, although the sole marker for this study had clearly defined (objective) categories from which tochoose, the marker still had to make a subjective choice about which category best accorded with a student's performance.Hence, elements of subjectivity are always likely to present a problem. However, it must be noted that measures were used toavoid subjectivity in marking as far as possible. These included ‘blind’ marking a batch of scripts, and, thereafter, randomlyselecting a number of those scripts to be double marked to check for marking consistency. Blind marking meant that eachpiece of writing submitted was graded blind of previous scores awarded, or knowledge about student identity. Numericalcodes were assigned to student writing portfolios.

The classes allocated to the researcher for the teaching of academic writing (RtL) took place during curriculum timealready designated for extended writing skills development. The time allocated for the teaching of each genre at the twoschool sites was identical e the RtL syllabus ensured that each class spent the same number of lessons working through thesame materials (see Appendix B). For example, for the teaching of the argumentative essay, 10 individual lessons were setaside. Lesson onewas set aside for the pre-test. Students were given a topic and asked to respond bywriting an argumentativeessay with no assistance from the research team or class teacher. This served as both a diagnostic test measure and baselinemeasure. Lesson two entailed the teaching of the introduction (stage 1) of the argumentative essay. Students spent lessonthree writing their own draft of an introduction. Lesson four entailed the teaching of the body of the argumentative essay(stage 2) and students spent lesson five attending to the feedback from their introduction, rewriting the introduction anddrafting their body. The same pattern followed for the conclusion. The final lessonwas a post-test where all three classes weregiven the same topic and asked to draft an academic argument in response to the topic given.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e176

Table 2 gives a description of each piece of writing submitted, which collectively made up the writing portfolios for eachstudent. Given that each student was required to write multiple drafts of each genre of writing, and each genre comprisedgathering pieces of writing before (baseline diagnostic), during (the process of re-writing) and after (the final product) theintervention, 837 individual pieces of writing (qualitative data) were collected (the sum of n (Valid) in Table 3). To bypasspossible concerns about inter-rater reliability, each piece of writing was marked by the same person to control for possiblediscrepancies or inconsistencies in marking standards that might have occurred if different markers (raters) were used. Inother words, inter-rater reliability was not a problem, because only one marker was used. All marks allocated were noted onthe marking rubric sheet only, and were filed separately from the student portfolios. No marks were recorded in each stu-dent's writing portfolio, which meant the marking of the subsequent piece of writing was done blind of previous scoresawarded, and problems relating to marker bias (i.e., awarding progressively higher marks for each successive piece of writingsubmitted) were, therefore, effectively negated.2

6. Findings from the descriptive statistics

Research globally tends to make use of pre- and post-intervention scores only in drawing conclusions about the efficacy ofRtL. However, this study alsomakes use of the scores obtained during the intervention, to give amore nuanced, process-basedperspective of student performance, which aids in highlighting any discrepancies during the implementation process, whichmay need attention. Using pre- and post-intervention scores only may show an improvement in students' overall perfor-mance. However, students may have ‘dipped off’ in performance at various points throughout the intervention, due to time,curriculum or pedagogic constraints. For example, we noticed a reduction in performance for all 3 classes in the finalassessment task for each genre of writing (see, where applicable, N4 compared to N3 and A4 compared to A3 in Table 3),which could largely be explained by the greater degree of complexity required for the task in each genre's case. But, thesereductions were relatively minor, not weakening or detracting from the general (positive) conclusions drawn.

Table 3 reports the descriptive statistics for all three classes. Class A comprised 32 students; Class B, 35 students and ClassC, 36 students. For each piece of assessed writing (i.e., N0 to N4; A0 to A4), the number of students who actually submitted apiece of writing for assessment is given by n (Valid) and those who did not, by n (Missing). The maximum possible score was42, where each of the 14 categories in themarking rubric (i.e., purpose; staging; phases; field; tenor;mode and so forth) couldbe awarded a maximum possible score of 3 out of 3 (see Appendix A). Although both the mean and median statistics arepresented for the sake of completeness, henceforth, we refer to the median statistics to establish certain patterns ofbehaviour, simply because of the relatively small class sizes and the (non-normal) distributional properties generally asso-ciated with these smaller samples of data.

The general improvement in students' (written) literacy performance across both genres as proxied by their scores foreach piece of writing submitted is noteworthy. The improvement was not only exhibited by the comparison of pre- and post-intervention scores for all three classes (N4 or N2 > N0; A4 > A0), but was also generally evidenced throughout the process ofintervention (N3 > N2 > N1 > N0; A3 > A2 > A1 > A0). In other words, owing to the lack of evidence to the contrary (i.e., thatwe were aware of, no other literacy intervention was being run concurrently and none of the classes or students concernedwere being subjected to any other academic literacy support intervention at the time), the data patterns exhibited couldreasonably be attributed to the implementation of RtL. However, this judgement is made with caution, because one is neverreally able to control for all other extraneous variables, which might have had an intervening or confounding effect onstudents' writing performance.

Figs. 2 and 3 give a visual representation of the general data patterns for the narrative and argumentative essays presentedin Table 3. What should be most noticeable is that Class C (the weakest cohort of students) had the ‘steepest’ rate ofimprovement; Class B (a relatively stronger cohort of students) had a relatively ‘flatter’ rate of improvement and Class A (thestrongest cohort of students) had the ‘flattest’ rate of improvement e as shown by the steepness of the steps of the bars byclass. The figures show visually the general pattern of behaviour observed within classes, which we refer to as the phe-nomenon of ‘convergence’ or ‘catch-up’. In general, within all three classes of students, gains were made in terms of thosefeatures that we systematically measured. However, for both genres of writing, Class C generally showed the fastest rate ofimprovement in the demonstration of their written literacy skills.

7. Findings from the inferential statistics

In the analysis to follow, we were most interested in whether or not there was a statistically significant improvement instudents' literacy scores within classes. Owing to the fact the sample (class) sizes were small and various samples of scoreswere not necessarily normally distributed, nonparametric testing procedures were deemedmost appropriate. To this end, we

2 To be clear, the use of a marking rubric to assess the written abilities demonstrated by students comprises issues of construct validity and internalvalidity. For instance, construct validity refers to whether the ‘construct’ e the rubric as assessment tool e correctly measures what students have beentaught (and have learnt) throughout the intervention. Because the rubric used is RtL intervention-specific, we assume this to be true. However, internalvalidity might refer to how the assessment tool is applied in each case; whether the (same) assessment rubric is applied the same way or consistently eachtime. Using the same person to mark all pieces of writing is arguably more likely to minimise any inconsistencies with which the rubric is applied in each-and-every case.

Table 2Description of the writing samples.

Data Type Writing Sample Description Sample Size

Pre-intervention Narrative essay: baseline(Task 1: N0)

Introductory narrative essay writing assignment used for diagnostic or baselinetesting purposes, and representing Task 1. Also referred to as the pre-test.

92

Process Narrative essay: orientation(Task 2: N1)

Extended piece of writing forming the orientation stage of a narrative for Task 2.Referred to as assignment N1. A narrative text was used as the text model.

93

Narrative essay: complication(Task 2: N2)

Extended piece of writing forming the complication stage of a narrative for Task2.Referred to as assignment N2. A narrative text was used as the text model.

90

Narrative essay: resolution & coda(Task 2: N3)

Extended piece of writing forming the resolution and coda stage of a narrativefor Task 2. Referred to as assignment N3. A narrative text was used as the textmodel.

49

Post-intervention Narrative essay: re-write(Task 3: N4)

Extended piece of writing forming the final (summative) re-write of a narrativetext. This included all three stages of the narrative and was referred to as Task 3.A narrative text and students' narratives from assignments N1-N3 (Task 2) wereused as the model text. Also referred to as the post-test.

59

Pre-intervention Academic essay: baseline(Task 4: A0)

Introductory academic essay writing assignment used for diagnostic or baselinetesting purposes, and representing Task 4. Also referred to as the pre-test.

85

Process Academic essay: introduction(Task 5: A1)

Extended piece of writing forming the introduction stage of an academic essayfor Task 5. Referred to as assignment A1. An academic essay was used as the textmodel.

88

Academic essay: body(Task 5: A2)

Extended piece of writing forming the body stage of an academic essay for Task5.Referred to as assignment A2. An academic essay was used as the text model.

90

Academic essay: conclusion(Task 5: A3)

Extended piece of writing forming the conclusion stage of an academic essay forTask 5.Referred to as assignment A3. An academic essay was used as the text model.

98

Post-intervention Academic essay: re-write(Task 6: A4)

Extended piece of writing forming the final (summative) re-write of anacademic essay. This included all three stages of the academic essay and wasreferred to as Task 6. A purposefully written academic essay and students'academic essays from assignments A1-A3 (Task 5) were used as the model text.Also referred to as the post-test.

93

Notes: A drop in the total number of students participating was due to a discontinuation of the narrative lessons with Class C after N2. School managementconstraints played a large part in this decision being taken. For instance, the respective class teacher was away on compassionate leave over this period,meaning regular access to the students was disrupted.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 7

only use SES and linguistic background as ways to qualitatively describe the different classes (Class A, Class B and Class C), andnot as a way to generate a predictive model explaining the ‘outcome’ variable (students' literacy scores).

For our purposes, in conjunction with the descriptive statistics of the literacy scores, we use a Friedman test e anonparametric test for joint significance of difference in the distribution of the ranks of the literacy scores across all applicableassessments by class (see Table 4) e and then associated post hoc Wilcoxon signed-rank test e a nonparametric test forpairwise significance of difference in the distribution of the ranked differences of the literacy scores across pairwise as-sessments by class (see Tables 5e7). Both tests are based on dependent samples. They are performed under the null hy-pothesis that the distributions being tested are drawn from the same population. Where multiple Wilcoxon pairwise tests areperformed for each genre by class, a Bonferroni correction to the critical level of significance must be made (Bonferroni-corrected a ¼ 0.05/number of tests) to account for inflated family-wise error rates (Field, 2009). A corrected critical level ofsignificance of 0.01 (0.05/5) applies to each genre in Tables 5e7, and 0.017 (0.05/3) applies to the Narrative Essay genre forClass C in Table 7.

Where a significant difference is shown to exist among paired samples using Wilcoxon's test, in conjunction with theassociated descriptive statistics (see Table 3), the sum of, number of andmean positive versus negative ranks can also be usedas statistics to gauge whether or not an overall improvement, stagnation or decline in literacy scores has occurred. Put moresimply, Figs. 2 and 3 give pictorial evidence of what the various statistical measures and tests show. In other words, thevarious Wilcoxon tests and associated statistics are useful at showing where improvements have occurred as part of aprocess-based assessment procedure. From a teaching practitioner's perspective, doing so makes sense, because a teacherrunning an intervention is constantly engaged in reflexive practice, and would ideally want to know where significantchanges have occurred. Although we do so in a post hoc fashion here, such knowledge would be useful for the practitioner toinform a change in pedagogy, for example, in future such interventions.

The results from Table 4 were generated using the Skillings and Mack (1981) version of the Friedman test, as oper-ationalised in Stata by Chatfield and Mander (2009). The Skillings-Mack test is appropriate for unbalanced or incompleteblock designs with a missing-data structure, whether the missing data be by design or completely at random (Chatfield &Mander, 2009), and is unlike the more conventional version of the test as operationalised in Stata by Goldstein (1991;2005), which cannot be performed when faced with missing data values e as is the case with our dataset. What the re-sults show is that all sample comparisons are highly significant. In other words, the samples of literacy scores for each genre ofwriting (Narrative Essay or Argumentative Essay) by class (Class A or Class B or Class C) were significantly different from one

Fig. 2. Median performance for the narrative essay by class.

Table 3Descriptive statistics.

CLASS A

Narrative Essay Argumentative Essay

N0 N1 N2 N3 N4 A0 A1 A2 A3 A4

Mean 26.29 33.3 36.26 36.92 35.39 29.84 33.78 35.18 36.87 35.29Std Dev. 5.197 4.772 1.723 3.393 5.123 4.079 4.852 4.691 2.986 3.046Median 26 34 36 37.5 37 31 35 36 37 35IQR 8 7 3 3 4.5 5 8 4 5 4Min. 17 22 33 26 21 19 24 18 30 30Max. 35 40 39 41 40 36 41 40 41 41

n (Missing) 1 2 5 6 4 7 9 4 1 1n (Valid) 31 30 27 26 28 25 23 28 31 31

CLASS BNarrative Essay Argumentative EssayN0 N1 N2 N3 N4 A0 A1 A2 A3 A4

Mean 20.3 31.71 33.1 37.43 33.48 28.54 32.16 35.97 35.71 33.79Std Dev. 3.871 4.362 4.221 2.233 4.582 2.546 5.298 3.431 4.428 4.350Median 21 32.5 34.5 38 35 28 33 37 37 35IQR 6 7 5 3 6 1.5 10 3 7 5Min. 12 20 21 31 22 24 23 28 26 24Max. 27 38 39 40 40 36 39 40 40 39

n (Missing) 8 7 5 12 4 7 3 3 1 2n (Valid) 27 28 30 23 31 28 32 32 34 33

CLASS CNarrative Essay Argumentative EssayN0 N1 N2 N3 N4 A0 A1 A2 A3 A4

Mean 19.24 27.31 34.64 18.91 25.45 33.57 35.48 35.14Std Dev. 4.321 4.262 4.980 2.821 4.684 3.989 4.324 2.973Median 20 26 36 19 26 34 37 36IQR 5 7 3 2.5 8 6 4 3Min. 6 21 14 12 19 22 22 26Max. 29 36 40 24 34 39 41 38

n (Missing) 2 1 3 4 3 6 3 7n (Valid) 34 35 33 32 33 30 33 29

Notes: Std Dev. refers to standard deviation. IQR refers to interquartile range, and is the difference between the values for the 75th and 25th percentiles. Thesum of n (Missing) and n (Valid) gives the total sample size per class.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e178

another. These results give credence to running the various post hoc pairwise comparisons using the Wilcoxon signed-ranktest.

From Table 5, one might be inclined to view RtL with some scepticism as only 4 of the 10 pairwise differences across bothgenres for Class A were significant at the corrected critical level of significance or lower. Weaker differences were observedthroughout the process of intervention (for both genres). In some sense, these results comprise the least convincing findings

Fig. 3. Median performance for the argumentative essay by class.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 9

for the three classes. Regardless, there was still significant evidence of progression (albeit weaker in places) and highlysignificant overall improvement with respect to pre- and post-intervention comparisons (N4-N0; A4-A0).

Table 6 shows the written literacy performance for Class B. One can see that 8 of the 10 pairwise differences across bothgenres were significant at the corrected critical level of significance or lower, implying a meaningful (positive) difference instudents' performance, both throughout the process of intervention and when comparing pre- and post-intervention scores.Irrespective of the insignificant difference in distributions noted (N2-N1; A3-A2) the general trend of improvementthroughout the entire process of intervention across both genres of writing was clearly evident for Class B.

Of the three cohorts, the results for Class C, as shown in Table 7, were the most convincing (see the very low p-values). Theresults show that the sample distributions in all pairwise comparisons were statistically (meaningfully) different from oneanother, except for the comparison A4-A3. The most salient finding from Table 7 is that students from Class C showed strong,significant improvements throughout the process of intervention, which was exhibited across both genres of writing.Although only a partial interventionwas administered for the narrative genre, there was still a similar strong, positive patternof improvement throughout.

A common finding, attributable to all three cohorts and seemingly against expectation, was that in pairwise comparisons(differences) N4-N3 and A4-A3, the common trend where the number of positive ranks outweighed the number of negativeranks was reversed (see Tables 5e7). However, this finding can be explained by the fact that the topic for the final piece ofwritten assessment (the post-intervention essay assignment) was new and required a technically more complex demon-stration of literacy skills. For example, students were not given any guidance with the creation of this new essay. Studentswere meant to make use of guidance from all previous written assessments to see if they had internalised skills taughtthroughout the intervention. Despite a slight drop-off at these stages of the process, the general trend of marked improve-ment was observed for all three classes and both genres of writing, showing evidence students had nowgenerally developed aform of academic autonomy. This finding is confirmed by the highly significant (positive) difference in students' literacyscores when compared purely on a pre- and post-intervention basis (see pairwise comparisons N4-N0 for Class A and Class Bor N2-N0 for Class C, and A4-A0 for all three classes).

8. Discussion

To contextualise the aforementioned findings bear inmind that Class C startedwith the lowest academic writing skills (seebaseline scores, N0 and A0). This could have been directly related to their inability to successfully assimilate into the ‘D/discourse’ of the school, or factors associated with lower socioeconomic circumstances (e.g., poorer linguistic stimulation,fewer reading materials, curriculum resource shortages, weaker pacing of curriculum coverage, insufficient teaching of ac-ademic writing etc.). RtL is designed to assist in these educational situations. Students from Class C also had greater difficultyusing English as the medium of instruction, largely due to a mismatch between the school's official language policy (English)and the language used for actual teaching and learning in the classroom (isiXhosa). A policy-practice gap exists in most SouthAfrican classrooms. For instance, official policy says that English should be used as the medium of instruction in secondaryschool classrooms throughout South Africa, but very often, informal policy prescribes that teachers make use of students'home language (which is not English) to supposedly facilitate the process of teaching and learning, which arguably onlyfurther exacerbates problems related to English language development and resultant (poorer) English academic literacy skillsof said students. Despite Class C having the lowest level of literacy pre-intervention, they performed on par with Class A andClass B (the stronger cohort of English First Additional Language students) in terms of the demonstration of their Englishacademic writing skills, post-intervention. The observed pattern of written literacy improvement being greater the weakerthe students' (initial) level of literacy skills was a notable finding of the study.

Table 4Friedman test results for Class A, Class B and Class C by genre of writing.

ClassCohort

Sample Comparisons

NarrativeEssay

ArgumentativeEssay

N0 N1 N2 N3 N4 A0 A1 A2 A3 A4

Class A 61.860(0.000)

38.019(0.000)

Class B 71.533(0.000)

59.645(0.000)

Class C 59.216(0.000)

101.254(0.000)

Notes: For each class cohort (Class A, Class B or Class C), the test statistic and associated p-value in parentheses are reported for each set of sample com-parisons. The associated probability values in each case represent the simulated null distribution for computation of the empirical p-value, because wherethere are tiese as was the case in all sample comparisonse the Stata programme (Skilmack) routinely runs a simulation. For the Class C Narrative Essay, onlyN0, N1 & N2 were used for the applicable sample comparisons test.

Table 5Wilcoxon signed-rank test results for Class A.

Pairwise Comparisons

N1-N0 N2-N1 N3-N2 N4-N3 N4-N0 A1-A0 A2-A1 A3-A2 A4-A3 A4-A0

z-statistic 4.467 2.826 1.769 �1.219 4.469 1.375 2.058 1.700 �2.499 4.010p-value 0.000 0.005 0.077 0.223 0.000 0.169 0.040 0.089 0.012 0.000Negative Ranks 1 7 4 13 2 7 7 8 20 1Positive Ranks 29 17 14 9 26 9 15 14 9 23Mean Negative Rank 15.5 9 16.88 14.69 3.5 6.71 10 14.81 18.68 10Mean Positive Rank 15.5 16.76 12.54 11.78 15.35 11.67 13.67 19.04 13.28 12.61Sum of Negative Ranks 15.5 63 67.5 191 7 47 70 118.5 373.5 10Sum of Positive Ranks 449.5 285 175.5 106 399 105 205 266.5 119.5 290Ties 0 2 4 2 0 1 1 6 2 0

n (Valid) 30 26 22 24 28 17 23 28 31 24

Table 6Wilcoxon signed-rank test results for Class B.

Pairwise Comparisons

N1-N0 N2-N1 N3-N2 N4-N3 N4-N0 A1-A0 A2-A1 A3-A2 A4-A3 A4-A0

z-statistic 4.379 2.154 3.707 �3.237 4.377 3.358 4.386 �0.047 �3.172 4.319p-value 0.000 0.031 0.000 0.001 0.000 0.001 0.000 0.963 0.002 0.000Negative Ranks 0 7 2 16 0 5 3 13 21 2Positive Ranks 25 16 20 2 25 20 27 15 6 25Mean Negative Rank n/a 13.57 6.5 12.81 n/a 8.6 8.17 20.12 21.26 4.75Mean Positive Rank 13 17.06 12 10 13 15.35 17.43 17.1 15.58 14.74Sum of Negative Ranks 0 95 13 205 0 43 24.5 261.5 446.5 9.5Sum of Positive Ranks 325 273 240 20 325 307 470.5 256.5 93.5 368.5Ties 0 4 0 3 0 1 1 4 6 0

n (Valid) 25 27 22 21 25 26 31 32 33 27

Table 7Wilcoxon signed-rank test results for Class C.

Pairwise Comparisons

N1-N0 N2-N1 N2-N0 A1-A0 A2-A1 A3-A2 A4-A3 A4-A0

z-statistic 5.020 4.653 4.847 4.854 4.677 3.265 �1.818 4.715p-value 0.000 0.000 0.000 0.000 0.000 0.001 0.069 0.000Negative Ranks 0 1 1 0 0 4 16 0Positive Ranks 33 31 30 30 27 22 8 29Mean Negative Rank n/a 15.5 1 n/a n/a 14.63 18.34 n/aMean Positive Rank 17 16.53 16.5 16.5 16 15.66 15.81 15Sum of Negative Ranks 0 15.5 1 0 0 58.5 293.5 0Sum of Positive Ranks 561 512.5 495 495 432 344.5 126.5 435Ties 0 0 0 1 2 2 5 0

n (Valid) 33 32 31 31 29 28 29 29

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e1710

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 11

All students, regardless of ability level as determined by pre-intervention baseline assessments, showed evidence of ac-ademic writing improvement. The greatest improvement was in the broad categories of ‘context’ and ‘discourse’ (seeAppendix A for clarification). For ‘context’, nearly all students made marked improvements with the schematic structuring oftheir essays together with aspects related to ‘field’. This meant there was also a marked improvement in the creation of textssuitably appropriate for the ‘purpose’ of the text's genre. Importantly, with the improvement of schematic structuring, thestudents' writing wasmore cohesive and coherent, allowing for clearer articulation of thoughts, ideas and arguments relevantto the essay questions. This alsomade the reading of students' writing easier to follow as onewas less inclined to read, re-readand have to revert to preceding elements of the essay to follow a student's line of argument. This finding provides someevidence that RtL might improve students' access to the ‘discourse’ of formal writing, because patterns of language usagewere made explicit to students during classroom teaching. For instance, access to the ‘discourse’ of formal writing wasimproved largely through intensive support and scaffolding, achieved through text modelling (examples of genre-specifictexts were given to students with aspects of the structure and language tools made explicit during classroom teaching e seeAppendix B), intensive assessment feedback of writing drafts, and differentiated processes and pedagogic practicesthroughout the implementation of RtL. For example, differentiated processes and pedagogic practices were formulated basedon an on-going needs analysis (process-based writing scores) for the three classes at the two different schools.

The term ‘convergence’ (or ‘catch-up’) can be used to explain the phenomenon of academic writing skills improvementfound during the study, whereby the gap between weaker- and stronger-performing students was narrowed within classes.The observed convergence phenomenon shows that, through the use of RtL, students who were initially deemed weaker intheir respective academic writing skills, were given an equal opportunity of improving their skills to a level on par withstronger-performing students, thereby illustrating RtL's ability to ‘democratise’ or ‘equalise’ the classroom. However, if left tonormal classroom pedagogic practices, initially weaker-performing students would arguably stagnate in their academic skillsdevelopment, exacerbating the chasm of educational division in the classroom. In other words, RtL provides a progressiveapproach to transforming the literacy development classroom.

Current classroom practice in South Africa tends to make use of an incremental learning model formalised through theresearch of Piaget. The incremental learning model provides student support according to the intellectual abilities of indi-vidual students, which arguably maintains the status quo between students e the abilities gap between stronger- andweaker-performing students persists. In contrast, RtL aims to provide students with differentiated support, meaning thatweaker students are given additional, differentiated support compared to stronger students, so that tasks which require anintellectual outcome significantly beyond weaker students' current abilities, can be achieved. This is contrary to an incre-mentalist approach. Therefore, RtL might best be described as a convergence learning model of academic literacy skillsacquisition.

9. Conclusion

There can be little doubt that a literacy crisis exists in South African education today. The contemporary evidence is quitepejorative concerning literacy development across all grades, with a widening gap between students from poorer versusmore privileged socioeconomic communities. This paper reported on both the descriptive and inferential statistical findingsof a broader study, which sought to test and assess the impact of the RtL literacy intervention as applied to several differentsample cohorts (or classes) of senior or upper-secondary students from two diverse socioeconomic school contexts in theWestern Cape of South Africa. Each class generally had divergent levels of written English language abilities, pre-intervention.The goal thenwas to assess whether RtL could significantly improve all students' (especially the most marginalised students')demonstration of their English language skills as proxied by the scores awarded for their various pieces of writing submittedacross two genres of writing (narrative and argumentative) throughout one academic year. The empirical evidence derivedfrom this study seemed to be supportive of convergence in English academic literacy skills development within classes ofstudents.

RtL can be described as a convergence (as opposed to an incremental) learning model of literacy skills development. HowRtL goes about achieving the fundamental goal of convergence is by providing access to discourses not normally madeavailable to students from low-literate homes, and, in doing so, provides access to categories of language awareness normallyassociated with middle and upper social-class homes. At a time in South African education where divergence in Englishliteracy skills across school contexts is the rule rather than the exception, RtL seems to be well placed to address the growingproblem. From a small-scale (longitudinal) perspective, the results were encouraging. However, further longitudinal researchacross a greater number of schools and socioeconomic contexts, would be a welcome addition to the growing body of evi-dence concerning RtL as applied to South Africa. Doing so, would then make it more appropriate to proffer broader gener-alisations about the efficacy of RtL in South Africa.

This paper adds to the dearth of literature on academic literacy skills development of secondary school students in SouthAfrica at a time where there is great concern about the literacy abilities of these students transitioning into higher levels ofeducation. But, these concerns are not unique to South Africa, and arguably encompass a broader global problem. Inaddressing problems related to literacy inequality, this paper offers insights into a tried and tested academic literacy inter-vention that can be used not only in South Africa, but myriad global educational contexts. In a day-and-age where globali-sation, among other things, has resulted in changing migratory patterns, it is becoming more and more common to findmultilingual classrooms with large cohorts of non-native speakers of English struggling to access teaching and learning,

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e1712

where English is the medium of instruction. In predominantly English-speaking countries, this means more resources andgreater effort needs to be devoted to offering remedial English classes formigrant learners to enable these students to developappropriate English academic literacy skills to participatemore fully in classroom activities. With growing evidence of literacyinequality globally, academic literacy interventions, such as RtL, that aim to remedy this problem are likely to be very muchneeded.

Funding sources

This work was supported by Doctoral funding from the Graduate School, Stellenbosch University, and the Department ofHigher Education and Training, South Africa. Postdoctoral funding was through the Oppenheimer Memorial Trust e which,without this generous fellowship, would not have resulted in the publication of this paper.

Acknowledgments

Professor Anthonissen was the principal supervisor for the Doctoral study which this paper originates from. As such, shegave valuable comments throughout the study, and spent hours reading various drafts of the thesis. Her input and expertisewere greatly appreciated. We also thank the reviewers for their valuable comments on the original manuscript.

This paper is written & published in memory of 2 students, Luxolo & “K-Way”, who passed away during our time withthem. May you both rest in peace! The teacher and students of ‘Class C’ have an ongoing impact on our lives.

Appendix A. Reading to learn marking rubric

Student no: Category Description 3/3 2/3 1/3 0/3

Context Purpose Appropriate for genreand writingassignment?

Excellent grasp of genreconventions. Meetspurpose of genre.

Adequate grasp ofgenre conventions.Working towardsmeeting purpose ofgenre.

Weak grasp of genreconventions.Extended writingdoes not meetpurpose of genre.

No awareness ofgenre conventions.Extended writingdoes not meetpurpose of genre.

Staging Are there appropriatestages? How welldeveloped is eachstage?

All stages of genrepresented and verywell developed.

Most stages of genrepresented andadequately developed.

Limited presentationof stages of genre.Needs furtherdevelopment.

No awareness ofstages of genre.Needs significantdevelopment.

Phases How well developed iseach phase?

All phases of genrepresented and verywell developed.

Most phases of genrepresented andadequately developed.

Limited presentationof phases of genre.Needs furtherdevelopment.

No awareness ofphases of genre.Needs significantdevelopment.

Field Does the writerconstruct the plot,settings and characters(narratives); explainthe field (factual texts),or describe the issues(argument)?

Excellent developmentof points of argument.

Adequate developmentof points of argument.

Limited developmentof points of argument.

No evidence of pointsof argumentpresented.

Tenor How well does thewriter engage thereader (narratives) orpersuade the reader(argument)?

Excellent, objectiveargument beyondpersonal reason andlinked to broader topic.

Adequate, nearobjective argument andlinked to broader topic.

Mostly a subjectiveargument with slightevidence ofobjectivitydeveloping.

Weak argument dueto subjective natureof argument and notlinked to the broadertopic.

Mode How well written is thetext for the school age?Is it spoken or writtenlanguage?

Well written, advancedlevel for age andlanguage affiliation.Evidence of appropriatewritten style.

Adequate level for ageand language affiliation.Emerging evidence ofwritten style oflanguage.

Poorly written for ageor languageaffiliation. Mostlyspoken style oflanguage withevidence of emergingacademic writtenstyle.

Poorly written for ageor languageaffiliation. Mostlyspoken style oflanguage.

Discourse Lexis Content words relatedto the field.

Excellent use of contentwords appropriate tothematic developmentof novel.

Adequate use ofcontent wordsappropriate to thematicdevelopment of novel.

Limited use ofcontent wordsappropriate tothematicdevelopment ofnovel.

Poor evidence ofknowledge of contentwords related tothematicdevelopment ofnovel.

(continued )

Student no: Category Description 3/3 2/3 1/3 0/3

Appraisal How well is appraisalused to engage/persuade/evaluate?

Excellent use ofappraisal techniquesappropriate to genre.

Adequate use ofappraisal techniquesappropriate to genre.

Inconsistent and pooruse of appraisaltechniques.

Very poor use ofappraisal techniquesfor genre.

Conjunctives Is there a clear logicalrelationship betweensentences?

Excellent use ofconjunctives.

Adequate use ofconjunctives.

Poor use ofconjunctives.

Erroneous use ofconjunctives (orabsence ofconjunctives).

Reference Is it clear who isreferred to in eachsentence (pronouns,articles,demonstratives)?

Excellent grasp of in-text referencing.

Adequate grasp of in-text referencing.

Limited use of in-textreferencing.

Erroneous use of in-text referencing.

GraphicFeatures

Grammar Have grammaticalconventions of formalEnglish been usedaccurately?

Excellent use ofgrammar.

Adequate use ofgrammar.

Weak grammar skills. Erroneous grammarskills.

Spelling How accurately arecore words and non-core words spelt?

Excellent spelling. Adequate spelling. Weak spelling. Non-existentspelling.

Punctuation Has punctuation beenused accurately andconsistently?

Excellent punctuation. Adequate punctuation. Weak punctuation. Non-existentpunctuation.

Presentation Paragraphs? Clearlayout?

Excellent presentation. Adequate presentation. Weak presentation. Very poorpresentation.

TOTAL

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 13

Appendix B

Classroom materials

Narrative essay model text

The crazy tiger. Once upon a time, in a dark,misty forest jogged a youthful petite framed girl with her equallyminuscule, short-haired Jack Russell. The girl might have been young, but she was energetic. She was a good runner and loved to spend hoursafter school running through the pine scented forest, leaping off rocks and sliding down the steep mountain bike tracks.Timbo, her Jack Russell, was quick and agile, and darted from rock to track, back and forth, back and forth as his attentive noseguided him. Hewas a loyal dog and loved nothingmore than running and exploringwith hismaster, Lucy. The forest; however,was dark and cold and as the leaves of the trees whispered during the erratic wind storms, one could be forgiven for thinkingthat they were talking about something - something that should be made known to all those who set foot in the forest.

Perhaps Lucy should have paid closer attention to the untamed whisperings of the trees because suddenly, withoutwarning, Timbo stopped dead in his tracks. Lucy, not used to sudden stops on her runs nearly fell hard as she tripped overTimbo in front of her. Lucy carefully and slowly got up and looked in the direction of Timbo's glance. There it was - tall, slenderand brown rosetted. Magestic! Timbo growled as it slowly walked towards them. Lucy, partially blinded by the airborne dustfrom her fall, was terrified to the core. Where on earth did the infant tiger come from?Was it going to eat Timbo? Would shebe able to out run it? Lucy's headwas spinning. Perhaps she bumped her head as she fell? She tried to think of away out of thesituation, when, out of the blue, she heard a loud, low ….….….… MEEEOOW.

Lucy thought she was now hearing things. Since when do Tiger's meow? Lucy plucked up the courage to have a closer lookand you will never guess what her tiger was. It was in fact a young male Bengal kitten. Lucy was ecstatic and overcome withjoy as she realised the tiger cat was in fact her neighbours courageous and inquisitive Bengal cat, Tarzan.

Lucy made a mental note to herself to stop reading stories of tigers and bears attacking unsuspecting hikers before bed inthe future. Lucy realised that this experience also made her appreciate the importance of why her mom told her not to runalone in the deserted forest.The narrative genre schematic structure for all three Classes.

Stage 1: Orientation

Phase 1.1: Setting (presents people, activities, places etc)Phase 1.2: Description (describes the people, places &things)

Stage 2: Complication

Phase 2.1: Problem (sets up the unexpected event that will create the problem/tension) Stage 3: Resolution Phase 3.1: Solution (describes how the problem is resolved or fixed)

Phase 3.2: Reaction (explains how the characters react to the problem being solved)Phase 3.3: Narrator's Comments (the narrator's thoughts on the events that just happened: not mandatory)

Stage 4: Coda (moral of the story)

Phase 4.1: Reflections (the characters thoughts about what has happened, often offered as a moral to the story)

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e1714

Example of the academic essay model text

Catching Fire, The Hunger Games

Essay question. In no more than one page (500 words), argue whether the novel ‘Catching Fire’ by Suzanne Collins isrelevant to the Grade 11 curriculum.

‘Catching Fire’ is the second novel within Suzanne Collins' ‘Hunger Games’ trilogy. The novel is set within a dystopian,post-apocalyptic society driven by a ruthless, power-hungry government. The ‘Capitol’ is extremely wealthy; whereas, thetwelve districts outside of the Capitol have different types of poverty. Katniss Evergreen, the novel's protagonist, survives thefirst round of the hunger games and finds herself, once again, at the mercy of the Capitol when she is re-selected to representher district e Appalachia. Dystopian societies are often defined as communities full of misery, oppression and disease (insertcitation here). Given the dictatorial rule of the Capitol in ‘Catching Fire’ and its relevance to present-day political disruption,this essay will argue that the novel is relevant for the Grade 11 curriculum. In doing so, this essay will explore the idea oftyrannical rule in the novel.

A tyrannical ruler is a leader who exercises total power and authority over a group of people and does this in a cruel, unjustand unfair way. For example, the Capitol and its leader, President Snow, illustrate characteristics of a tyrannical leader in howhe controls the way the various Districts are run and governed (p. 100). The Capitol also controls the availability of food andresources to the Districts despite plenty resources being made available to the people in the Capitol (p. 200). This can oftenlead to feelings of misery and hopelessness which may ultimately lead towards a rebellion. For example, Katniss, after herrebellious victory at her first appearance in the ‘Hunger Games’, becomes the symbol (mocking jay) of a rebellion (p. 250).Rebellions are not always a bad thing but can lead to an unnecessary loss of life, bringing turmoil, pain and grief alongwith theprocess. Given the ‘unstable’ political context of South Africa, it is important to understand how rebellions can start and whattheir consequences can be, rendering a reading of this novel appropriate for the grade 11 curriculum.

To conclude, this essay argued that the novel is relevant for the Grade 11 curriculum. This is largely due to the relevance ofthe use of tyrannical rulers in the novel which highlights the cruel, unjust and unfair treatment of marginalised people bythese types of rulers. The description of President Snow and his harsh rule over the poorer people of the districts illustratesvery well how feelings of misery and hopelessness can easily give rise to rebellions which is a stark reminder of the politicalcontext of South Africa, rendering a reading of this novel relevant.

The academic essay (Exposition) genre schematic structure for all three Classes.

Stage 1Introduction

Phase 1.1: Background Information Controversial StatementDefinition/sContextual Information

Phase 1.2: Thesis StatementPhase 1.3: Controlling Ideas (firstly; secondly, thirdly etc)

Stage 2Body(Paragraph 1)

Phase 2.1: Topic SentencePhase 2.2: EvidencePhase 2.3: Elaboration/ExplanationPhase 2.4: Sign-post Forward

Body(Paragraph 2)

Phase 2.1: Topic SentencePhase 2.2: EvidencePhase 2.3: Elaboration/ExplanationPhase 2.4: Sign-post Forward

Stage 3Conclusion

Phase 3.1: Restatement of Thesis StatementPhase 3.2: Short/Brief Summary of each ParagraphPhase 3.3: Recommendations

Reading to learn teaching schedule

The teaching schedule presented below was used to maintain consistency with regards to Reading to Learn imple-mentation across all three classes.

Unit one e narrative essay

� Welcome students.� Introduce the project aims and objectives and provide a rational for student participation.� Students to complete the assent forms.� Hand out work books and discuss how programme implementation will run.� Students to write baseline assignment (Task 1 e N0).

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 15

Unit two e narrative essay

� General feedback on baseline assessment.� Unpack stage one (orientation) of the narrative in detail.� Students to construct own stage one based on model text (orientation) (Task 2 e N1).� Completed drafts to be handed in at end of each unit for detailed feedback before the next lesson.

Unit three e narrative essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Unpack stage two (complication) of the narrative.� Students to make use of feedback from stage one (orientation) and rewrite stage one as well as construct stage two basedon model text (complication) (Task 2 continued e N2).

� Completed drafts to be handed in at end of each unit for detailed feedback.

Unit four e narrative essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Unpack stages three and four (resolution and coda) of the narrative.� Students to make use of feedback for stage one and stage two (orientation and complication) and rewrite stage one andtwo as well as construct stages three and four based on model text (resolution and coda) (Task 2 continued e N3).

� Completed drafts to be handed in at end of each unit for detailed feedback.

Unit five e narrative essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Hand out final post-intervention topic.� Students to compile post-intervention narrative (Task 3 e N4) e all students were given one lesson only for this task.

Unit six e academic essay

� Hand out academic essay topic.� Students to write baseline academic essay (Task 4 e A0).

Unit seven e academic essay

� General feedback on baseline assessment.� Unpack stage one (introduction) of the academic essay in detail.� Students to construct own stage one based on model text (academic essay) (Task 5 e A1).� Completed drafts to be handed in at end of each unit for detailed feedback before the next lesson.

Unit eight e academic essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Unpack stage two (body) of the academic essay.� Students to make use of feedback from stage one (introduction) and rewrite stage one as well as construct stage two basedon model text (body) (Task 5 continued e A2).

� Completed drafts to be handed in at end of each unit for detailed feedback.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e1716

Unit nine e academic essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Unpack stage three (conclusion) of the academic essay.� Students to make use of feedback from stage one and stage two (introduction and body) and rewrite stage one and two aswell as construct stage three based on model text (conclusion) (Task 5 continued e A3).

� Completed drafts to be handed in at end of each unit for detailed feedback.

Unit ten e academic essay

� Hand out work books and go over general feedback.� Allow time for individual student queries.� Hand out final post-intervention topic.� Students to compile post-intervention academic essay (Task 6 e A4).

References

Acevedo, C. (2010). Will the implementation of reading to learn in stockholm schools accelerate literacy learning for disadvantaged students and close theAchievement gap? Sweden: Multilingual Research Institute.

Acevedo, C., & Rose, D. (2007). Reading (and writing) to learn in the middle years of schooling, pen, 157 (pp. 1e8). Sydney: Primary English TeachingAssociation.

Allardice, S. (2013). Student preparedness for academic writing: An evaluation of the perceptions of preparedness for academic writing of school leavers takingEnglish 178 at Stellenbosch University (Stellenbosch: Unpublished Masters Dissertation).

Bernstein, B. (1975). Class and Pedagogies: Visible and invisible. London: Routledge.Bernstein, B. (1990). The structuring of pedagogic discourse. London: Routledge.Bernstein, B. (1996). Pedagogy, symbolic control and Identity: Theory, research, critique. London: Taylor and Francis.Chatfield, M., & Mander, A. (2009). The Skillings-Mack test (Friedman test when there are missing data). Stata Journal, 9(2), 299e305.Choinski, E., Mark, A., & Murphey, M. (2003). Assessment with Rubrics: An efficient and objective means of assessing student outcomes in an information

resources class. Libraries and the Academy, 3(4), 563e575.Cukras, G. (2006). The investigation of study strategies that maximize learning for underprepared students. College Teaching, 54(1), 194e197.Department of Basic Education. (2014). Report on the annual national assessment of 2014: Grades 1 to 6 & 9. http://www.education.gov.za/Curriculum/

AnnualNationalAssessments(ANA).aspx. (Accessed 10 June 2017).Field, A. (2009). Discovering statistics using SPSS (3rd ed.). London: Sage.Fraser, W., & Killen, R. (2003). Factors influencing academic success or failure of first-year and senior university Students: Do education students and

lecturers perceive things differently? South African Journal of Education, 23(4), 254e263.Goldstein, R. (1991). snp2: Friedman's ANOVA & Kendall's coefficient of concordance. Stata Technical Bulletin, 3, 26e27. (Reprinted in Stata Technical Bulletin

Reprints, 1, 157e158. Stata Press: College Station, TX).Goldstein, R. (2005). Software update: snp2 1: Friedman's ANOVA test and Kendall's coefficient of concordance. Stata Journal, 5(2), 285.Greene, J., & Forster, G. (2003). Public high school graduation and college readiness rates in the United States. Education working paper 3. september. New York:

Center for Civic Innovation.Groenewald, J.. (2005). Unpublished Proceedings from the Stellenbosch University's Teaching Day, Stellenbosch.Halliday, M. (1978). Language as a social semiotic. London: Edward Arnold.Halliday, M. (1989). Spoken and written language. Oxford: Oxford University Press.Halliday, M. (1996). On grammar and grammatics. In R. Hasan, C. Cloran, & D. Butt (Eds.), Functional Descriptions: Theory in practice (pp. 1e38) (Amsterdam:

Benjamins).Howie, S., Combrinck, C., Roux, K., Tshele, M., Mokoena, G., McLeod Palane, N., & PIRLS Literacy. (2016). South African highlights report, centre for evaluation &

assessment. Preotira: University of Pretoria.Howie, S., Van Staden, S., Tshele, M., Dowse, C., & Zimmerman, L. (2012). PIRLS 2011: South African Children's reading literacy Achievement: Summary report.

Pretoria: Centre for Evaluation & Assessment, University of Pretoria.Hulstijn, J. (1997). Second-language acquisition research in the Laboratory: Possibilities and limitations. Studies in Second Language Acquisition, 19(2),

131e143.Letseka, M., & Maile, S. (2008). High university drop-out rates: A threat to South Africa's future: HSRC policy brief, pretoria. Human Sciences Research Council.MacGregor, K. (2009). Shocking results from university test. University World News.Maloney, W. (2003). Connecting the texts of their lives to academic Literacy: Creating success for at-risk first-year college students. Journal of Adolescent &

Adult Literacy, 46(8), 664e672.McGhie, V. (2012). Factors impacting on first-year students' academic progress at a South African university (Stellenbosch: Unpublished Doctoral Thesis).McRae, D., Ainsworth, G., Cumming, J., Hughes, P., Mackay, T., Price, K., et al. (2000). Education and training for indigenous Students: What has worked (and will

Again): The IESIP strategic results project. IESIP SRP national coordination and evaluation team. Canberra: Australian Curriculum Studies Association.Millin, T. (2015). Scaffolding academic literacy using the reading to learn Methodology: An evaluative study (Stellenbosch: Unpublished Doctoral Thesis).Millin, T., & Millin, M. (2014). Scaffolding academic literacy using the reading to learn Intervention: An evaluative study of a tertiary education context in

South Africa. Per Linguam, 30(3), 26e38.Nakata, M., Nakata, V., & Chin, M. (2008). Approaches to the academic preparation and support of australian indigenous students for tertiary studies. The

Australian Journal of Indigenous Education, 37(Supplementary), 137e145.Penrose, A. (2002). Academic literacy perceptions and Performance: Comparing first-generation and continuing-generation college students. Research in the

Teaching of English, 36(4), 437e461.Rose, D. (2005). Democratising the classroom: A literacy pedagogy for the new generation. Journal of Education, 37, 131e167.Rose, D., & Acevedo, C. (2006). Closing the gap and accelerating learning in the middle years of schooling. Literacy Learning: The Middle Years, 14(2), 32e45.Rose, D., Rose, M., Farrington, S., & Page, S. (2008). Scaffolding academic literacy with indigenous health sciences Students: An evaluative study. Journal of

English for Academic Purposes, 7(3), 165e179.Shay, S. (2004). The assessment of complex performance: A socially situated interpretive act. Harvard Educational Review, 74(3), 307e329.Shay, S. (2008). Beyond social constructivist perspectives on Assessment: The centring of knowledge. Teaching in Higher Education, 13(5), 595e605.

T. Millin, M. Millin / Journal of English for Academic Purposes 31 (2018) 1e17 17

Skillings, J., & Mack, G. (1981). On the use of a friedman-type statistic in balanced and unbalanced block designs. Technometrics, 23(2), 171e177.Taylor, N. (2009). The state of South african schools Part 1: Time and the regulation of consciousness. Journal of Education, 46, 9e32.Taylor, S., & Von Fintel, M. (2016). Estimating the impact of language of instruction in South african primary schools: A fixed effects approach. Economics of

Education Review, 50, 75e89.Troskie-De Bruin, C. (1999). Academic Development: Bridging at a South African university (Stellenbosch: Unpublished Doctoral Thesis.).UNESCO. (2011). The hidden Crisis: Armed conflict and education. Paris: UNESCO Publishing.Van Dyk, T., & Weideman, A. (2004). Switching Constructs: On the selection of an appropriate blueprint for academic literacy assessment. SAALT Journal for

Language Teaching, 38(1), 1e13.Van Schalkwyk, S. (2008). Acquiring academic literacy: A case of first-year extended degree programme students at Stellenbosch university (Stellenbosch:

Unpublished Doctoral Thesis).Vygotsky, L. (1978). In M. Cole, V. John-Steiner, S. Scribner, & E. Souberman (Eds.), Mind in Society: The development of higher psychological processes.

Cambridge, MA: Harvard University Press.

Tracey Millin is currently a Visiting Academic Fellow at the University of Otago - Higher Education Development Centre. She has a special interest inEmancipatory Education and Educational Linguistics. She specialises in teaching academic writing to undergraduate students, and the training of teachers tobecome literacy teachers across the curriculum.

Mark Millin is currently a full-time Doctoral candidate at the University of Otago - Department of Economics. His research interests are in the field of TheEconomics of Education e global public spending on education patterns, and Literacy Development. Mark has lectured undergraduate Economics at theUniversity of Natal/KwaZulu-Natal, and the University of Otago.

View publication statsView publication stats