UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The discourse of causal explanations in school science Slater, Tammy Jayne Anne 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2004-931765.pdf [ 19.53MB ]
JSON: 831-1.0078228.json
JSON-LD: 831-1.0078228-ld.json
RDF/XML (Pretty): 831-1.0078228-rdf.xml
RDF/JSON: 831-1.0078228-rdf.json
Turtle: 831-1.0078228-turtle.txt
N-Triples: 831-1.0078228-rdf-ntriples.txt
Original Record: 831-1.0078228-source.json
Full Text

Full Text

T H E DISCOURSE OF C A U S A L EXPLANATIONS IN SCHOOL SCIENCE by T A M M Y JAYNE A N N E SLATER B.A. (Honours), University of Victoria, 1986 • M.A. , University of British Columbia, 1998 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF GRADUATE STUDIES (Language and Literacy Education) We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH C O L U M B I A April, 2004 © Tammy Jayne Anne Slater ABSTRACT Researchers and educators working from a systemic functional linguistic perspective have provided a body of work on science discourse which offers an excellent starting point for examining the linguistic aspects of the development of causal discourse in school science, discourse which Derewianka (1995) claimed is critical to success in secondary school. Yet the work that has been done from this perspective has generally focused on texts in science books and encyclopedias, or in other words, texts written by expert writers (e.g., Mohan et al., 2002; Veel, 1997). A notable exception is Gibbons (1998, 2003), who used data from an elementary ESL science class to illustrate the move from hands-on, context-dependent discourse to the decontextualized forms characteristic of science writing, and the role of the teacher in providing the necessary scaffolding to make this move successfully. No work has yet described the development of causal language by identifying the linguistic features present in oral discourse or by comparing the causal discourse of native and non-native (ESL) speakers of English. The current research responds to this gap by examining the oral discourse collected from ESL and non-ESL students at the primary and high school grades. Specifically, it asks the following questions: 1. How do the teachers and students in these four contexts develop causal explanations and their relevant taxonomies through classroom interactions? 2. What are the causal discourse features being used by the students in these four contexts to construct oral causal explanations? Ethnographic data collection involved recording observations of classroom interactions (244 recorded hours) as well as formal and informal interviews (9+ hours), which were transcribed and coded to reveal (1) how the teachers built up key concepts through their implementation of the two types of linguistic patterning which Halliday (1998) claimed is involved in constructing science knowledge—the creation of technical terms and chains of logical reasoning—and (2) the causal discourse features which were used by the students to construct their explanations. A social practice analysis revealed the similarities and differences which existed among the four contexts studied with regard to the teachers' ways of developing the ability to explain and construct science knowledge, and a small corpus study helped to show the patterns of development across the same four contexts. Concept maps (Novak, 1998), built from the discourse of the classroom interactions, offered graphics to illustrate the knowledge which was constructed through the classroom discourse. The findings of the social practice analysis showed that the teachers in the four contexts differed in their approaches to teaching, with the primary school mainstream teacher focusing largely on the hands-on practice, the primary school ESL teacher moving/rom practice to theory, the high school mainstream teacher movingyrom theory to practice, and the high school ESL teacher relying primarily on theory. Although no causal connections can be made from this study regarding the effectiveness of one approach over another, the findings appear to reflect the popular practice of using hands-on, minds-on approaches to teaching and learning science. The study therefore contributes a new, linguistic perspective to work which has been and continues to be carried out in science education. The findings from the quantitative, small corpus approach suggest that the developmental path of cause which has been identified in the writing of experts shows up not only in written texts but also in the oral texts which learners construct. Moreover, this move appears when the discourse of high school ESL and non-ESL students is compared, suggesting a developmental progression in the acquisition of these features by these students. The findings also reveal that the knowledge constructed, as shown by the concept maps created from the discourse, follows a developmental path similar to the linguistic causal path, from the concrete, hands-on, observable items to more abstract, theoretical concepts. This study is the first systemic functional comparison of the oral discourse of primary and secondary learners as well as the first to compare ESL and non-ESL speakers in this way, and as such it helps map general trends in causal discourse development. iii TABLE OF CONTENTS Abstract ii List of tables ix List of figures xii Acknowledgements xiv CHAPTER 1: INTRODUCTION 1.0 Why causal explanations? 1 1.1 Causal explanations as part of science content 2 1.2 Teaching causal explanations from the science educators' perspective 3 1.3 Teaching causal explanations from the language educators' perspective 4 1.4 Teaching causal explanations in language and content classrooms 5 1.5 Purpose of the study 6 1.6 Research questions 7 1.7 Significance of the study 8 1.8 Theoretical background 11 1.8.1 Theories of causality and causal explanations 11 1.8.2 Theories of language 14 1.8.3 The formalist/structuralist paradigm: Language as rule 14 1.8.4 The functionalist paradigm: Language as resource 15 1.8.5 Theories of learning and teaching 16 1.8.6 Theories of social science research 18 1.8.7 Situating the present study 20 1.9 The format of this thesis 21 1.10 Transcribing conventions 21 CHAPTER 2: REVIEW OF THE LITERATURE 2.0 An overview of the chapter 22 2.1 Reviewing causal discourse and its stake in learning 22 2.2 Science teaching and science language 25 2.3 Teaching content through language and language through content 30 2.4 Science discourse: A systemic functional perspective 35 iv 2.5 Causal explanations and grammatical metaphor 39 2.6 Grammatical metaphor and Halliday's two types of patterning 41 2.7 Mapping the development of concepts and language 42 2.8 The development of causal explanations in school science 45 2.8.1 The development of causal explanations in the classroom 45 2.8.2 The development of causal explanations across grade levels 47 2.9 Summary 52 C H A P T E R 3: M E T H O D S O F INQUIRY 3.0 An overview of the chapter 54 3.1 Research design 54 3.1.1 Interpretive/constructivist (qualitative) research 54 3.1.2 Case study research 55 3.1.3 Discourse analysis 56 3.1.4 Social practice theory 57 3.1.5 Corpus analysis 60 3.1.6 The design of the present study , 61 3.2 Research procedures 63 3.2.1 Sampling procedures 63 3.2.2 Research sites 65 3.2.3 Participants 73 3.3 The role of the researcher 81 3.4 Data collection procedures 81 3.4.1 Observations 82 3.4.2 Interviews 83 3.4.3 Documents 85 3.5 Data analysis and presentation 85 3.5.1 Social practice perspective revisited 87 3.5.2 Corpus analysis revisited 87 3.5.3 Concept mapping revisited 88 3.6 Research design trustworthiness 90 3.7 Summary 91 v CHAPTER 4: MRS. SINCLAIR'S PRIMARY SCIENCE CLASS 4.0 An overview of the chapter , 93 4.1 Mrs. Sinclair's magnet unit 94 4.2 Constructing knowledge in Mrs. Sinclair's class 96 4.2.1 Sequencing the unit 97 4.2.2 Building technicality: Renaming, redefining, and reclassifying 98 4.3 Carrying out the experiments 102 4.3.1 Trouble with technicalizing 103 4.3.2 Opportunities for reasoning 109 4.4 Tracking the construction of three key concepts 118 4.4.1 Building up an understanding of attract 118 4.4.2 Building up an understanding of repel 123 4.4.3 Building up an understanding of north pole and south pole 125 4.5 The final question-and-answer period 128 4.5.1 Questions which concerned the building of technicality 129 4.5.2 Questions which elicited logical reasoning 132 4.6 Summary 136 CHAPTER 5: MRS. MONTGOMERY'S PRIMARY SCIENCE CLASS 5.0 An overview of the chapter 138 5.1 Mrs. Montgomery's magnet unit 138 5.2 Constructing knowledge in Mrs. Montgomery's class 141 5.2.1 Building technicality: Renaming, redefining, and reclassifying 141 5.2.2 The magnet experiments and logical reasoning 148 5.3 Tracking the construction of three key concepts 156 5.3.1 Building up an understanding of attract 156 5.3.2 Building up an understanding of repel 162 5.3.3 Building up an understanding of north pole and south pole 165 5.4 The language of the interviews 168 5.4.1 The students' use of technicality 168 5.4.2 The use of causal discourse in logical reasoning 170 5.5 Summary 176 vi CHAPTER 6: MR. PETERSON'S HIGH SCHOOL SCIENCE CLASS 6.0 An overview of the chapter 180 6.1 Mr. Peterson's chemistry unit 181 6.2 Constructing knowledge in Mr. Peterson's class 182 6.2.1 Building technicality: Renaming, redefining, and reclassifying 183 6.2.2 Prompting logical reasoning 193 6.3 Tracking the construction of three key concepts 200 6.3.1 Building up an understanding of physical properties 200 6.3.2 Building up an understanding of compounds 206 6.3.3 Building up an understanding of mixtures 211 6.4 The language of the interviews 217 6.4.1 Opportunities for technicality: The breadth of process lexis 218 6.4.2 The use of causal discourse in logical reasoning 222 6.4.3 Shifting between the congruent and the metaphoric 228 6.5 Summary 234 CHAPTER 7: MS. ARMSTRONG'S HIGH SCHOOL SCIENCE CLASS 7.0 An overview of the chapter 240 7.1 Ms. Armstrong's chemistry unit 240 7.2 Constructing knowledge in Ms. Armstrong's class 244 7.2.1 Building technicality: Renaming, redefining, and reclassifying 244 7.2.2 Prompting logical reasoning 254 7.3 Tracking the construction of three key concepts 262 7.3.1 Building up an understanding of physical properties 263 7.3.2 Building up an understanding of compounds 273 7.3.3 Building up an understanding of mixtures 278 7.4 The language of the interviews 284 7.4.1 The ESL students' use of technicality and grammatical metaphor 284 7.4.2 The ESL students' resources for logical reasoning 289 7.5 Summary 299 vii CHAPTER 8: FINDINGS AND DISCUSSION 8.0 An overview of the chapter 304 8.1 Responding to research question one 304 8.1.1 The primary classes 305 8.1.2 The high school classes 315 8.1.3 Similarities and differences 322 8.1.4 Summary of research question one 324 8.2 Responding to research question two 325 8.2.1 Summarizing the features of causal discourse used 326 8.2.2 The developmental path of cause: The mainstream English speakers 333 8.2.3 The developmental path of cause: The ESL/mainstream speakers 341 8.2.4 The developmental path of cause: The ESL speakers 344 8.2.5 The use of the negative as a resource for meaning 345 8.2.6 Summary of research question two 347 8.3 The developmental paths of concepts and language 348 8.4 Implications 350 8.4.1 Implications for researchers 350 8.4.2 Implications for educators 354 8.5 Future directions 358 8.6 Reflections on the study 361 References 363 Appendix 1: The station instructions for Mrs. Sinclair's experiments 377 Appendix 2: Mrs. Sinclair's magnet booklets 381 Appendix 3: Mrs. Montgomery's experiments 385 Appendix 4: Key visuals in the writing tasks 388 Appendix 5: Mr. Peterson's students' entities 391 Appendix 6: Ms. Armstrong's students' entities 392 viii LIST OF TABLES Table Title Page Table 3.1 The four research contexts 65 Table 3.2 Mrs. Sinclair's student pairs 74 Table 3.3 Mrs. Montgomery's three classes 76 Table 3.4 The participants in Mr. Peterson's science classes 78 Table 3.5 Mrs. Armstrong's two classes of ESL science 80 Table 3.6 The database 82 Table 3.7 The data as action and reflection discourse 87 Table 4.1 Mrs. Sinclair's plan for the magnet unit 94 Table 4.2 The questions at the stations 96 Table 4.3 Functions of the teacher's unit instructions 97 Table 4.4 Forms of the students' definitions 99 Table 4.5 Processes used to describe attract across the twelve stations 104 Table 4.6 The use of attract by students at Station One 105 Table 4.7 Science participants and their everyday replacements 108 Table 4.8 Teacher/researcher use of question probes 110 Table 4.9 The temporal and causal relations constructed at the twelve stations 112 Table 4.10 Causal discourse features used across the twelve stations 113 Table 4.11 Processes used to talk about magnetism in the final session 132 Table 4.12 Causal and temporal resources in the question-and-answer session 133 Table 4.13 Causal and temporal relations in the question-and-answer session 134 Table 5.1 Mrs. Montgomery's magnet unit 139 Table 5.2 The students'processes of attraction 143 Table 5.3 The students' use of attract in Experiment One 144 Table 5.4 The frequency of Mrs. Montgomery's questions and connections 149 Table 5.5 The students' use of causal language features in the ten experiments 150 Table 5.6 The students'temporal and causal relations 154 Table 5.7 Processes used to talk about magnetism in the interviews 169 ix Table 5.8 Causal language features and their relations in the interviews 171 Table 6.1 Major events in Mr. Peterson's chemistry class 182 Table 6.2 Shifts in the discourse from specific to general 199 Table 6.3 The interviews with Mr. Peterson's students 217 Table 6.4 The 196 processes used by Mr. Peterson's students in the interviews 219 Table 6.5 The average number of occurrences per process 221 Table 6.6 The textual analysis of excerpt one 223 Table 6.7 The textual analysis of excerpt two 223 Table 6.8 The causal features of the mainstream interviews 226 Table 6.9 Temporal and causal relations in the mainstream interviews 228 Table 6.10 Examples of specific and general in Zachary's and Ivan's discourse 232 Table 7.1 Major events in Ms. Armstrong's ESL science class 243 Table 7.2 The interviews with the ESL high school students 284 Table 7.3 The 106 processes in the ESL interviews 285 Table 7.4 The average number of occurrences per process 286 Table 7.5 The causal language features present in the ESL interview data 291 Table 7.6 Tony's explanation of chemical separation 293 Table 7.7 The temporal and causal relations in the ESL interviews 295 Table 8.1 The social practice of learning about magnetism 305 Table 8.2 Action and reflection discourse in Mrs. Montgomery's class 306 Table 8.3 Specific and general reflection in Mrs. Montgomery's class 307 Table 8.4 Action and reflection discourse in Mrs. Sinclair's class 312 Table 8.5 Specific and general reflection in Mrs. Sinclair's class 313 Table 8.6 The social practice of learning about matter 315 Table 8.7 Action and reflection in Mr. Peterson's class 316 Table 8.8 Shifting from specific to general in Mr. Peterson's demonstrations 317 Table 8.9 Reflection on talk 318 Table 8.10 Ms. Armstrong's guided recount of the lab 319 Table 8.11 General reflection as conclusion 320 x Table 8.12 Reflection on the text as action 322 Table 8.13 Conjunctions used in the four contexts 327 Table 8.14 The most popular conjunctions used in the four contexts 328 Table 8.15 The circumstances used in the four contexts 329 Table 8.16 The processes used in the four contexts 329 Table 8.17 Causal and temporal participants used in the four contexts 330 Table 8.18 Nominalizations, abstractions, and specialized terms 331 Table 8.19 Lexical density across the four contexts 332 Table 8.20 The use of negative causal and temporal relations 333 Table 8.21 The path of nominalizations in the mainstream oral discourse 334 Table 8.22 The path of abstractions in the mainstream oral discourse 335 Table 8.23 The path of temporal and causal participants 336 Table 8.24 The path of external temporal conjunctions 336 Table 8.25 The path of external causal conjunctions 337 Table 8.26 The path of causal processes 337 Table 8.27 The path of proof 338 Table 8.28 The path of circumstances 338 Table 8.29 The path of temporal processes 339 Table 8.30 The developmental path of cause for the mainstream students 340 Table 8.31 The developmental path of cause for high school ESL/mainstream 342 Table 8.32 The developmental path of cause for primary school ESL/mainstream 344 Table 8.33 The developmental path of cause for the ESL students 345 xi LIST OF FIGURES Figure Title Page Figure 2.1 Idealized developmental path for cause (Mohan et al., 2002) 49 Figure 3.1 Spradley's variations in research scope 56 Figure 3.2 Social practice, ethnography, sociology, and functional linguistics 57 Figure 3.3 Contexts and the Knowledge Framework 58 Figure 3.4 Mohan's model of social practice (basic level) 59 Figure 3.5 Social practice in learning and teaching (Mohan, 2003) 60 Figure 3.6 The design of the present study : 62 Figure 3.7 Mrs. Sinclair's teaching area 66 Figure 3.8 Mrs. Montgomery's teaching area 68 Figure 3.9 Mr. Peterson's teaching area 70 Figure 3.10 Ms. Armstrong's teaching area 72 Figure 3.11 The process of concept mapping 88 Figure 4.1 Mrs. Sinclair's unit as a concept map 136 Figure 5.1 Visualizing attract 161 Figure 5.2 The graphic representation of the two key processes 163 Figure 5.3 Mrs. Montgomery's unit as a concept map 178 Figure 6.1 The taxonomy of matter presented in Mr. Peterson's class 200 Figure 6.2 The initial taxonomy of physical properties 201 Figure 6.3 Building on the initial taxonomy 202 Figure 6.4 Mr. Peterson's list of physical properties : 204 Figure 6.5 Visually defining compound 209 Figure 6.6 The parallel "translation" of terms in defining solution 213 Figure 6.7 Mechanical mixture separation 215 Figure 6.8 Shifting from events to actions 229 Figure 6.9 Mr. Peterson's unit as a concept map 236 Figure 7.1 Diagram of Mrs. Armstrong's chemistry unit 242 Figure 7.2 Characteristics of Bert 248 xii Figure 7.3 Characteristics of an electron 248 Figure 7.4 The four steps of the lab and accompanying language features 261 Figure 7.5 The taxonomy of matter taught in Ms. Armstrong's class 262 Figure 7.6 The three states of matter 264 Figure 7.7 Model taxonomy of "everyday" properties 266 Figure 7.8 Description of the atom 267 Figure 7.9 Taxonomies offered for Family One and Family Eighteen 269 Figure 7.10 A web of properties created by the various discourses 270 Figure 7.11 The observation and description activity worksheet 271 Figure 7.12 Polarity in the definitions 279 Figure 7.13 Ms. Armstrong's unit as a concept map 301 Figure 8.1 Connections among experiments, specific, and general discourse 311 Figure 8.2 Social practice and the four teaching styles 322 xiii ACKNOWLEDGEMENTS There are many people whom I would like to acknowledge for their help in bringing this project from its initial inception to its final state. First and foremost, I would like to thank my research supervisor, Dr. Bernie Mohan, for his endless support, intelligence, and enthusiasm. My thanks go out, too, to my two committee members, Dr. Gloria Tang and Dr. Jim Gaskell, for their contributions to the research process. Moreover, I would like to state my appreciation for the support the Department of Language and Literacy Education has offered me over the years, and to the University of British Columbia's (UBC) University Graduate Fellowships, which have given me the financial ability to see this degree through to its completion. This project would certainly not have been possible without the four teachers who agreed to participate in this study. Regardless of the descriptions of the particular situations they found themselves in during the data collection for this thesis, I believe that all four were excellent and caring educators. My appreciation goes out to their students as well—I enjoyed interacting with all of them. At a broader level, I would like to thank the school principals who opened their doors for me to collect data in their schools, and to the administrators at the Saanich School District for allowing me in despite the distance between U B C and the district. I offer a big thank you to the friends who proof-read my thesis—Dasha Malinovich-Rowe, Karin Kellythorne, Corinna Gilliland—to the recent and continuing members of my research discussion group (Carolyn Kristjansson, Angela Wu, Lynn Luo, Sunah Cho, Kecia Yim, Jung Hwang, and Kristy Liang), and to the friends and colleagues who have offered their support and encouragement over the years, in particular Dr. Gulbahar Beckett, Dr. Barbara Harris, Dr. Yan Guo, (Dr. to be) Masaki Kobayashi (and wife, Emi), and (Dr. to be) Michael Levykh. Finally (but certainly not last in my heart!), I would like to thank all the members of my family for their endless support. I extend special thanks to my mother, Jeanette Slater, for ensuring I had every chance possible to finish this degree, and to my sister, Lynn Nash, who played multiple roles in this undertaking. CHAPTER 1: INTRODUCTION 1.0 Why causal explanations? The language of cause and effect permeates almost every academic subject in school, particularly in the upper grades. English literature classes, for example, require students to explain the motivation of characters in works of literature. Discussions in social studies classes revolve around the examination of effects and consequences of various events in history. In science, hypothesizing, predicting, and experimenting clearly involve relations of cause and effect. Painter (1999) maintained that "the ability to infer cause-effect relations is fundamental to notions of 'logical' or 'scientific' thinking, and the fostering of the abilities to reason and hypothesize are prominent educational goals throughout the Western world" (p. 245). Yet causal language, according to Halliday and Martin (1993) is characterized by grammatical metaphor, which is "fundamental to adult uses of language" (Halliday & Matthiessen, 1999, p. 7) and which children typically only begin to handle well at around thirteen years of age. Since this indicates a potential developmental issue, research into causal explanations is important for the purpose of language education, both for native speakers of English and for English as a second language (ESL) students. This study focuses on causal explanations in science, rather than in other disciplines, because of the existing knowledge base in the field. As this chapter will introduce and Chapter Two will detail, there is a solid foundation which can be used to drive theory forward. Most of this has explored causal explanations in science texts written by experts in language and content; thus the results offer a starting point for examining oral causal explanations constructed by learners, both students of English as their first language and ESL learners. Both groups in this study are learning how to construct causal explanations in science using English, and by holding their discourse up to that written by experts, it may be possible to explore more fully any developmental trends in the acquisition of this form of academic language. 1 1.1 Causal explanations as part of science content The Integrated Resource Packages (IRPs) for science, developed by the Curriculum Branch of the British Columbia Ministry of Education, aim to provide "a framework of opportunities for students to become scientifically literate" (e.g., British Columbia Ministry of Education, 1995a, 1995b). According to Karplus and Thier (1967), to be considered scientifically literate, "the individual must have a conceptual structure and a means of communication that enables him to interpret the information as though he had obtained it himself (p. 24). Norris and Phillips (2003) argued that the ability to handle science text is critical to learning and understanding science, and that therefore literacy in its most fundamental sense is central to scientific literacy, although oracy in science "plays an irreplaceable role in the development, critique, and refinement of scientific thought" (p. 233). Throughout the British Columbia Ministry web pages, the IRPs advocate developing science skills and processes which are "the same as those used by scientists at work." Students are expected to develop the knowledge, skills, and attitudes necessary for scientific literacy by "working scientifically," "communicating scientifically," "using science," and "acting responsibly." According to the IRPs, these four processes involve such language functions as asking questions, explaining, defending opinions, discussing limitations, and defining problems, thereby suggesting that language is a main component of science instruction. A brief glance at the learning outcomes for all grade levels further demonstrates the importance of language, as many of those outcomes involve verbs such as describe, communicate, explain, discuss, suggest, and debate. Moreover, the IRPs for all grades list such skills as observing, predicting, controlling variables, measuring, communicating, interpreting data, classifying, hypothesizing, formulating models, designing experiments, and inferring, many of which involve the explicit use of language, as "central to the presentation of all content and the delivery of instruction and assessment activities in classrooms." As inferred from these Ministry documents, language is certainly a key part of science education, but how is language—particularly causal language—being developed in science classrooms? 2 1.2 Teaching causal explanations from the science educators' perspective In the past, language development in science tended to focus primarily on technical terms; after all, as O'Toole (1996) noted, specialist vocabulary tends to be the most noticeable feature of scientific English, and as the Classical Component of the 1991 British Columbia Assessment of Science stated, "an understanding of scientific terminology is necessary for further investigations to take place" (Bateson et al., 1992, p. 169). Yet it has become more recognized that learning to talk science "runs rather deeper than 'simply' learning to articulate the words and phrases of a new speech genre" (Scott, 1998, p. 74), and several recent articles have listed the typical characteristics of science language which make it problematic for students (e.g., Buck, 2000; Carlson, 2000; Simich-Dudgeon & Egbert, 2000). Although an interest in the use of analogy and metaphor to explain scientific concepts has surfaced (Ogborn, 1996), science educators in general have not paid much attention to the nature or use of explanation in their teaching (Lawrence & Pallrand, 2000), and "'explanation' still remains a largely unexplicated notion" (Ogborn, 1996, p. 159). Moreover, even though many authors have advocated the explicit teaching of science language (e.g., Henderson & Wellington, 1998; McGinn & Roth, 1999; McKeon, 2000; O'Toole, 1996; Prophet & Towse, 1999; Ritchie & Tobin, 2001; Yore, Craig, & Maguire, 1998; Zohar & Ginossar, 1998), causal discourse as a key element of scientific explanations has not been addressed. Where causal reasoning in science has been explored (e.g., Borges & Gilbert, 1999, who concluded that more detailed studies of sequential and causal reasoning need to be done), the focus has been on conceptual understanding, and the role language plays in this has not been elaborated on. In fact, the general view throughout science education journals is that it is the science teacher's task to facilitate the acquisition of science concepts; language is recognized as playing a role in this, but conceptual understanding is the primary interest. This recognition is revealed in a comment by Roth (1998), who noted that concepts can be viewed "as the patterns in the language employed by students to describe and explain their science-related experiences, and conceptual change is the change in these descriptions and explanations" (p. 1020). Although students are expected to advance their conceptual understanding and to demonstrate that understanding 3 orally and in written form, causal language development in science classes has not yet been targeted as an area deserving explicit attention. 1.3 Teaching causal explanations from the language educators' perspective Causal language development from the language educators' perspective is surprisingly superficial. From a brief inspection of current popular English as a second language (ESL) textbooks such as the Interchange series (Richards, Hull, & Proctor, 1991) and the Canadian Concepts series (Berish & Thibaudeau, 1998), or English for Academic Purposes (EAP) textbooks (e.g., Campbell, 1995), or Azar's (2000) English grammar Chartbook, causal discourse appears to be limited to isolated lexical items such as the causative verbs make, have, and get, and various causative connectives such as because, consequently, since, and if. The writing of cause and effect discourse is concentrated in only four pages of Swales and Feak's Academic writing for graduate students: A course for nonnative speakers of English (1994). Causal discourse in reference books for native English speakers do not fare much better, and very few articles on the topic have appeared in journals, which is surprising given that causal explanations seem to play such a key role in academic discourse. One exception is Cronnell (1981) who, in a piece dedicated to an overview of cause and effect language, stated that an understanding of cause-effect relations was one factor among many which lead to good reading comprehension. He stated that these relations were common in discourse, but could be quite complex and therefore could pose problems for young readers. His article described "the various kinds of cause-effect relations and constructions that readers must be familiar with in order to comprehend effectively" (p. 155). In eleven pages, Cronnell listed all the examples of cause and effect that he considered necessary to introduce; all the examples offered were either lexical items as in the textbooks or adverbial clause constructions. Mohan (1997) and Mohan and van Naerssen (1997) criticized this sentence-level view of causal discourse as being inadequate for academic language development. Using student recalls of a cause and effect reading passage from a social studies unit, both articles argued that causal meanings are constructed using a combination of rich lexical and grammatical 4 resources and that for students to be able to understand and produce academic discourse, they need to be aware of the subtleties of causal meanings and how to construct them. The development of causal language was the topic of Mohan and Beckett (2001), who illustrated how the interaction between a teacher and a university-level ESL student in a project-based language classroom led to the student's use of more literate, academically valued language. Causal language, as Mohan and Beckett showed, is an important part of academic language, and yet the textbooks which teach language explicitly, as noted above (see also Flowerdew, 1998) , do not go much beyond the sentence-level treatment which Mohan criticized. Little work has been done to explore how these meanings are constructed in science discourse; consequently, little work can be offered to inform the language-teaching textbooks. 1.4 Teaching causal explanations in language and content classrooms Because it has been recognized that non-English-speaking students who arrive in English schools cannot wait until they speak the language fluently before beginning content instruction, much of the work in the combined language and content teaching research has focused on strategies which help reduce the linguistic demands so that these students can access the academic content they need to learn while simultaneously learning English (for a concise discussion of content-based instruction, see Crandall, 1999; for examples of work in the area, see Snow & Brinton, 1997). Although simplifying the language of instruction has been recommended by some authorities (e.g., British Columbia Ministry of Education, 1999) , several educators have instead advocated using graphics to help trigger existing knowledge structures and to show the underlying conceptual relationships in the content (e.g., Carlson, 2000; Early & Tang, 1991; Mohan, 1986, 2001; Tang, 1992, 2001). As Mohan (1986) stated: Much of academic knowledge is knowledge of relations, and these relations can be, and often are, represented by graphics. If learners are able to interpret these graphics, they have easier access to the knowledge represented, (p. 90) Moreover, each knowledge structure visually captured by the graphic, according to Mohan, has specific language associated with it which can be used to construct the discourse of the knowledge structure. 5 Working from Mohan's framework, Early and Tang (1991) reported on a procedure for supporting students' academic reading through key visuals. Tang (1992, 2001) further showed how a social studies unit was presented using these visuals, and how the cause-effect graphics led to student compositions of causal texts. The teacher in Tang's study had to provide the student with the linguistic items of cause and effect, but the resulting student-generated text was a coherent passage. Carlson (2000) presented ideas for and examples of various key visuals that are useful for teaching science concepts in a way that makes the language accessible. Research has shown, however, that the same key visual can produce very different texts. Slater (1998) demonstrated how a visual representation of the water cycle generated a wide variety of text types, and even within the same category of causal explanation, different explanations were judged to be more or less academically literate. What this suggests, then, is that an individual's ability to construct causal meanings may be related to the overall depth or breadth of his or her linguistic resources. Supporting this idea, Mohan (2001) noted that there is a major contrast between skilled and unskilled writers working with the same information. This highlights the importance of looking at resources for causal discourse as academic discourse and reinforces Mohan's observation that sentence-level treatments of causal discourse, although perhaps useful initially, are inadequate for the development of academic language. A closer examination of the topic is very much needed. 1.5 Purpose of the study Given the importance of causal discourse in academic language development as stated above, the present study has been designed to explore the construction of oral causal explanations by ESL and non-ESL students in primary and high school science classes. Specifically, it aims to examine how teachers and their students develop causal explanations in four contexts of school science, and which linguistic resources the students in each of these four contexts use to explain their understanding of cause and effect relationships. It is not the purpose of this study to judge the adequacy or inadequacy of the participants' understanding of science concepts; such a task demands the skills of a science educator. 6 This study also proposes to examine the role which grammatical metaphor plays in the development of causal explanations in these four contexts, and hold this up to previous research on the topic. According to Halliday and Martin (1993), grammatical metaphor is similar to lexical metaphor in that both involve linguistic transformations, but "instead of being a substitution of one word for another,... it is a substitution of one grammatical class, or one grammatical structure, by another" (p. 79, italics in original). Acquiring an ability to use this linguistic resource is a critical step in becoming socialized into an academic discourse community because "articles written for specialists typically display a considerably denser concentration of grammatical metaphor" (p. 14). If it is indeed the goal of science education to apprentice students into the scientific discourse community (McGinn & Roth, 1999; Ritchie & Tobin, 2001), it is necessary for educators to understand the role grammatical metaphor plays in science language and how it can be developed in science classrooms. The study also responds to the more general need for exploring how teachers and learners construct meaning together. In a book devoted to conversational analysis, Markee (2000) claimed that in the field of second language acquisition, theory has "far outstripped empirical verification" about how "second language (L2) learners use talk to learn new language" (p. 13). This study aims to provide empirical data which address how teachers and students are using (or not using) talk to develop resources for causal explanations in science classrooms. 1.6 Research questions This study takes a systemic functional linguistic perspective on language, using discourse analysis and concordancing techniques to examine how the participants in four different contexts construct causal explanations. The four contexts reflect different populations (ESL and non-ESL speakers) and different age groups (six to eight years old and fourteen to sixteen years old). The questions which guide this examination of the four contexts are as follows: 1. How do the teachers and students in four distinctively different contexts-primary and high school ESL and non-ESL classes—develop causal explanations and their relevant taxonomies through classroom interactions? 7 2. What are the causal discourse features being used by the students in these four contexts to construct oral causal explanations? The first question is addressed primarily through a discourse analysis of the classroom interactions using observation data (orally recorded data and field notes). The second question is responded to by looking closely at interviews with students from the four contexts and quantifying the causal discourse features they used in their explanations. 1.7 Significance of the study This study is significant in a number of ways. First, it brings together science education and language education and informs teachers and teacher educators about the linguistic resources children use to explain their views of the world. Science educators have typically focused on children's conceptions or misconceptions about science, not attending to the linguistic resources which the child uses to construct those (mis)conceptions and therefore not realizing the potential that science has for developing the child's academic language ability. Sutman (1996) argued that teacher education programs typically do not take the language and science connections into consideration, saying that "too often, the professional practices of teachers ignore the role in science education for language development beyond memorization of science vocabulary" (p. 460). This study aims to show how language development occurs alongside the development of science concepts. Second, as Martin (1972) pointed out, science textbooks frequently require students to explain, yet there has been little systematic inquiry into what acceptable scientific explanations are. More recently, Nieswandt (2001) also raised the question of what constitutes a good explanation at a given level. Brewer, Chinn, and Samarapungavan (2000) noted that "people have relatively clear intuitions about what is or is not an explanation and that these intuitions serve as the foundation of most discussions of explanations" (p. 279). Added to that comment is an observation by Keil and Wilson (2000) that "there are also compelling intuitions about what makes good explanations in terms of their form" (p. 1). Gilbert, Boulter, and Rutherford (1998) reviewed the literature on teachers' understanding of the principles of explanation and concluded that relatively few studies have been carried out in the area. As Ogborn (1996) stated, "giving a better, clearer and fuller account of what 8 explaining is in the science classroom is a current task of urgency and importance" (p. 159). Although this study cannot make judgments about the scientific adequacy of explanations, it can offer science educators a description of the linguistic resources exploited by children constructing causal explanations at two distinct age levels. As far as the researcher knows, a study such as this has not yet been undertaken. Third, although science educators and researchers have frequently observed differences in the ways children "talk science" and have noted that their everyday language is not adequate for learning and explaining science (e.g., Ebenezer & Erickson, 1996; Levine & Geldman-Caspar, 1997; Mqje, Collazo, Carillo, & Marx, 2000; Nieswandt, 2001), these comments have typically been left at the level of observation and a systematic analysis of this language has not been carried out. Many science educators assume that what poses problems for children learning science are aspects of the vocabulary (e.g., Allie, Buffler, Kaunda, Campbell, & Lubben, 1998; Carlson, 2000; Prophet & Towse, 1999). Because this study aims to shed light on how children construct causal explanations in science, it can examine the connections between "vocabulary" and "grammar" in children's developing linguistic ability and thereby offer information about these connections and how they relate to everyday versus science language. Fourth, whereas there have been several research projects which have examined how teachers help construct scientific understanding through oral discourse in the classroom (e.g., Driver, Leach, Millar, & Scott, 1996; Lemke, 1990; Ogborn, Kress, Martins, & McGillicuddy, 1996), these studies involved students who were not designated English as a second language (ESL) students and the focus was on concept development rather than language development. Gibbons (1998, 2003) examined ESL students studying magnetism and argued the importance of teacher scaffolding in developing the children's ability to handle academic discourse. Aside from Gibbons's studies, the researcher is not aware of any other projects undertaken in ESL science classrooms with a focus on language development. Language development is the aim of both the science and the language educator (although it seems that science educators rarely acknowledge this), but it is an explicit goal of ESL science classes where teachers must prepare students for mainstream classes while 9 continuing to teach grade-appropriate science concepts. Because the present study involves ESL science classrooms, it can inform the field of second language acquisition by offering a description of the causal language used by ESL students in content-based language classes. Fifth, in their research of students' ideas about the nature of science, Driver, Leach, Millar, and Scott (1996) developed a framework of three types of student-generated explanations which the authors suggested may be developmental. The present study is significant because it will add to their information by highlighting potential developmental patterns in the explanations of both native and non-native speakers of English. Sixth, this study is significant because it offers insight into the kind of academic science language used to construct causal explanations at widely differing age levels. It has been suggested that the ability to deal with grammatical metaphor, a key aspect of academic discourse, begins at about the time students enter grade eight (Halliday & Martin, 1993), but the discussion of grammatical metaphor has so far been concerned with speakers whose first language is English. Because this study deals with ESL students as well as with students whose first language is English, the findings will add greatly to the second language acquisition knowledge base through its exploration into potential differences between the two age levels under investigation with respect to grammatical metaphor. As far as the researcher knows, the only work that has been done connecting grammatical metaphor and second language development is Mohan and Beckett (2001), which focused on university-level students. The present study should add to this by showing similarities and differences in the use of grammatical metaphor and academic language use between the causal explanations composed by ESL students and those of non-ESL students at two contrasting age levels. Seventh, it is now a commonly held view that it takes second language learners five to seven years to achieve what Cummins (1984) referred to as Cognitive Academic Language Proficiency (CALP). Because this study aims to examine causal explanations produced by ESL and non-ESL speakers at the high school level and analyze the similarities and differences between these two groups, the findings may make it possible to single out key aspects of causal language development which teachers can focus on to facilitate students' 1 0 academic language development across the curriculum. Developing academic language quickly is especially important for the high school age group, which has limited time left to acquire the language necessary to graduate successfully from grade twelve. Furthermore, knowing which areas to focus on may make it easier to respond to Derewianka (1995), who recommended research into teacher intervention in the development of grammatical metaphor in children's language. Finally, as noted in the previous section, this study is significant because it addresses the general need as outlined by Markee (2000) for more empirical verification of how students, particularly ESL students, talk to learn in academic settings. 1.8 Theoretical background Because the researcher's world view influences the questions asked and shapes how the data are collected, analyzed, and presented (Creswell, 1994; Merriam, 1988; Mertens, 1998), it is important to summarize these perspectives as they relate to the study being undertaken. This section will therefore review theories of causality, language, learning, and research, and situate the present study within these theories. 1.8.1 Theories of causality and causal explanations Over the years there have been a number of books written on the topic of causal explanations and causality, and it is not the intention to offer a thorough review here. It is, however, important to offer a brief summary of the ideas held by the researcher as her views may influence the way she selects, interprets, and presents the data. A well-used example of causal theory is Hume's billiard ball example in which ball A strikes ball B, setting it in motion while potentially ceasing its own. The causal relationship between these two balls is not something directly observable: "What we actually observe is one event followed by another event" (Searle, 1983, p. 112). This view, which highlights the consistency of events occurring in a temporal sequence, is referred to as regularity theory (Harre & Madden, 1975). This regularity theory can be contrasted with the power theory, which attributes power to an agent in a causal relationship. Harre and Madden explained 11 the differences in the two views by discussing sedation: Whereas holders of a regularity view of causality would explain that an individual "has been put to sleep by opium because all or most cases of opium taking have regularly been followed by sleep" (p. 85), those holding a power view would look to the nature of opium itself. These two views reflect different perspectives on causality, and the language the individual chooses to construct an explanation may reflect either of these views. As suggested above and closely related to the regularity and power theories is the notion of agency. Harre (1993) raised the question of whether humans are "active agents using their social knowledge jointly to accomplish certain ends" or whether they are "information-processing automata, the behaviors of which are the effects of causal processes" (p. 11). Answers which discussed this agent/automata issue are that "persons are not causes, but their actions can be" (Vendler, 1984) and that "causes are means and tools. People can use them to bring about their effects" (Hausman, 1998). Agent/automata contrasts are apparent in the grammar of English through such pairs as the boat sailed I Mary sailed the boat in which the medium and agent are contrasted (Halliday, 1994, p. 164). Causal explanations in science are often constructed as implication sequences (Halliday & Martin, 1993) or what Hempel (1993) referred to as a genetic explanation in which "each stage must be shown to 'lead to' the next, and thus be linked to its successor by virtue of some general principle which makes the occurrence of the latter at least reasonably probable, given the former" (p. 32). The principle or mechanism which links these events is often not directly observable. Simon (2000) stated that "explanatory theories usually account for phenomena at one level by means of mechanisms drawn from the next lower level of the system structure" (p. 35). Ahn and Kalish (2000) suggested that in the mechanism view of causal explanation, the "mechanism is framed at a different level of analysis that are the cause and the effect. That is, mechanisms involve theoretical constructs that are removed from and underlying the evidential phenomena themselves" (p. 201). Causal explanations, therefore, tend to involve abstract concepts in their linguistic constructions. Another way of examining the notion of causality is by looking at the conditions which bring events about or prevent effects from occurring. Skyrms (1986) noted that 12 the word "cause" is used in English to mean several different things. For this reason, it is more useful to talk about necessary conditions and sufficient conditions rather than about causes.... Being run over by a steamroller is a sufficient condition for death, but it is not a necessary condition. Whenever someone has been run over by a steamroller he is dead. But it is not the case that anyone who is dead has been run over by a steamroller. On the other hand, the presence of oxygen is a necessary condition, but not a sufficient condition for combustion. (1986, p. 84-85) According to Skyrms, people often use everyday language ambiguously to refer to the cause of something; this ambiguity is unable to distinguish between sufficient and necessary conditions or between the signs and symptoms of the effects. He argued that "the precise language of necessary and sufficient conditions is much more useful than the vague language of cause and effect, sign and symptom" (1986, p. 87). The last point to be mentioned here is the existence in the biological sciences of anthropomorphic or teleological explanations denoting a sense of causality. These types of explanations contain notions of function, purpose, and intentionality (von Wright, 1971) and are similar to each other in that purpose and intentionality are considered human attributes, yet these goal-oriented outcomes are typically offered as explanations for natural phenomena (Zohar & Ginossar, 1998; Zuzovsky & Tamir, 1999). There are issues regarding the appropriateness and legitimacy of using these types of explanations when teaching and learning science (Lemke, 1990; Taber & Watts, 1996; Zohar & Ginossar, 1998). For a recent review of the types of causality which surface in scientific explanations and the nature of the development of these in children, see Grotzer (2003). The views in this section suggest that defining causation and causal explanation may be a difficult task given that philosophers have been debating the meaning of causality and explanation for many years (Kiel & Wilson, 2000; Ogborn, 1996; Salmon, 1998). This study adheres to a definition of causation which follows that of Lakoff and Johnson (1999) and is centrally based on volitional human agency via direct physical force: At the heart of causation is its most fundamental case: the manipulation of objects by force, the volitional use of bodily force to change something physically by direct contact in one's immediate environment, (p. 177) 13 This view of prototypical causation extends to a number of different kinds of causation such as means-end relations. Included in the study's definition is Halliday's distinction between external causation—a causes x to happen—and internal causation or proof —h proves y or b causes one to think y (Halliday, 1993, p. 64). Moreover, the study will adopt the broad definition of a causal explanation proposed by Christie, Gray, Gray, Macken, Martin, and Rothery (1992): An explanation is a piece of writing that tells a reader how and why something happens as it does or how and why something is as we find it. (p. 7) Given that the current research involves oral explanations, however, the word 'writing' in the definition above is understood to mean 'discourse', and 'reader' to include 'listener'. 1.8.2 Theories of language In the field of linguistics there are two primary paradigms: the formalist/structuralist paradigm and the functionalist paradigm (Derewianka, 1999; Halliday, 1994; Martin, Matthiessen, & Painter, 1997; Schiffrin, 1994). The assumptions about the nature of language which each paradigm makes and the way each views language learning and teaching are strikingly different. 1.8.3 The formalist/structuralist paradigm: Language as rule The formalist structuralist paradigm (e.g., Chomsky, 1957, 1965) highlights the form of language by focusing on the universality of rules which allow us to understand and create novel sentences. From this perspective, language development and teaching concerns the individual's ability to acquire and manipulate these rules. Much language instruction, both first language development and English-as-a-second-language (ESL) instruction, targets this manipulation to internalize correct linguistic patterns. Although Chomsky's standard theory of grammar is considered too abstract to be particularly useful in the ESL language classroom, "most instructors... cite the insight into patterns of English, into knowing what is rule-governed behavior and what needs to be memorized, into what structures are similar and different, into knowing what goes together as very useful in their teaching" (Paulston, 1998, p. 714, italics in original). In summary, the prominent view from this perspective is that we 14 know a particular language when we know the rules which form the basis of that language (Fromkin, Rodman, Hultin, & Logan, 2001). In other words, the focus from this view is on form. 1.8.4 The functionalist paradigm: Language as resource Rather than seeing language as a biologically or neurologically formed system, linguists working within the functionalist paradigm (e.g., Halliday, 1994) see language as a vast system from which we choose and construct meanings to achieve specific social goals. Functional grammar examines how language has evolved in a particular culture to enable us to accomplish these social goals within that culture. Its focus is on the text as a whole, not on syntax, although smaller units within the text can be highlighted as they relate to the whole text. Systemic functional linguistics (SFL), the functional perspective this research follows, considers that all languages are internally organized into three metafunctions: ideational, which allows speakers to represent experience; interpersonal, which enables them to set up and maintain relationships; and textual, which allows them to create connected, coherent discourse (Christie & Unsworth, 2000). Because SFL maintains that language is connected to social purposes, text is always interpreted in its context, and context is in turn always a part of the text. There are two inter-related levels of context: context of situation and context of culture. The former is described in terms of field, which is the content or topic of the activity; tenor, which refers to the nature of the people involved; and mode, which refers to the medium used in the situation. These three variables relate to the three metafunctions as field/ideational, tenor/interpersonal, and mode/textual, and each variable draws from the resources available in its accompanying metafunction. The context of culture, according to Christie and Unsworth, also influences language choice in that "cultures evolve recognizable ways by which members can achieve their social purposes in the range of situations they typically experience" (p. 4). These cultural practices—or genres—have their own characteristic text structure, and becoming a member of a particular culture entails learning how to structure appropriate genres, such as a scientific causal explanation, within the context of a particular situation, such as a grade nine 15 science lesson on atoms. Unlike a grammar in the formalist/structuralist view, SFL does not focus on correctness of form, although it does not ignore this aspect, but looks at meanings which are or can be constructed from the resources of the language to meet particular needs in a social/cultural context. Barker (2001) raised the question of whether Chomsky's theory of universal grammar is simply affording "us all a "lowest common denominator" of verbal expression" (p. 421) and postulated that it is the environment—particularly environments which support reading and writing—which is "responsible for the profound individual differences in thought and consciousness that accompany differences in word usage" (p. 422). This perspective reflects to some extent Halliday's notion of language as a resource for making meaning in that Barker is suggesting that students' resources are expanded by engagement with text and that each student's engagement with the textual environment determines the extent of his or her resources. To summarize, this study, with its focus on a specific social genre and how its meanings are constructed by members of a particular culture within certain social contexts and interactions, is best approached through the theoretical framework of SFL, which views language as a resource for constructing meaning. 1.8.5 Theories of learning and teaching Over the decades there have been several trends in the field of language learning (Brown, 2000) and of learning in general (Anderson, 2000; Barker, 2001; Bransford, Brown, & Cocking, 1999). Structuralism/behaviorism looked at the observable, breaking it into small units which could be scientifically described and put together again into a whole. Among psychologists, behavior was seen to be something which could be objectively perceived, recorded, and measured. From this paradigm, learning involved being conditioned to respond in certain ways based on positive or negative reinforcement. Anchored to this perspective is the educational practice referred to as transmission which, according to Miller and Seller (1990) is "philosophically allied with an empiricist world view |and| psychologically allied with behaviorism" (p. 56). This position sees knowledge 16 as fixed content which students are expected to learn in contexts which emphasize "direct instructional techniques such as lecture and recitation" (p. 43) and in which the student "merely responds to a structured learning situation" (p. 56). From the transmission position, the purpose of education "is to transmit facts, skills, and values to students" (p. 5), typically using traditional teacher-fronted approaches which see language as a conduit through which this knowledge can be poured. The view of learning reflected in the Chomskian view of language builds on structuralism, but is not simply interested in the objective measurement of units. From a language perspective within this view of learning, Chomsky's generative-transformation school of linguistics aimed to look beyond a simple description of language to underlying explanations for language acquisition, including notions of innateness and universal grammar. Looking at learning from a psychological view, cognitive psychologists turned to rationalism, looking for reasons to explain why humans behave and learn as they do. In the later part of the twentieth century, the constructivist paradigm emerged. According to Spivey (1999), "constructivists view people as constructive agents and view the phenomenon of interest (meaning or knowledge) as built instead of passively "received" by people whose ways of knowing, seeing, understanding, and valuing influence what is known, seen, understood, and valued" (p. 3). This paradigm led to a more transactive view of teaching and learning in which "education is viewed as a dialogue between the student and the curriculum in which the student reconstructs knowledge through the dialogue process" (Miller & Seller, 1990, p. 6). Within this constructivist paradigm are Piaget and Vygotsky, who differed in their ideas about the role which social context plays in learning and development. In the words of Cole and Wertsch (1996), "according to the canonical story, for Piaget, individual children construct knowledge through their actions on the world.... By contrast, the Vygotskian claim is said to be that understanding is social in origin" (p. 250). Although the authors argued that this "story" is in dispute, language, as a primary cultural artifact, seems to play a much larger role in Vygotsky's view of learning than it does in Piaget's. Tomasello (1996) noted: In general it may be said that Vygotsky accorded to language an active and formative role in intellectual development, as children rallied their cognition 17 around the communicative conventions of mature members of their cultures, whereas Piaget always subordinated language to cognition, especially the operative aspects of cognition that derive from children's physical actions on the physical world (later internalized into logical operations carried out mentally), (p. 269) It is this focus on language and dialogue in the constructivist paradigm that is reflected in the notion of language socialization, a concept articulated by Schiefflin and Ochs (1986; see also Cazden, 1999). This concept views language acquisition as "socialization through language and socialization to use language" (p. 14); in other words, linguistic knowledge is embedded in sociocultural knowledge, and language development is thus seen as socialization into the particular cultural habits and actions of the social group in which the learner is involved. Because it can be argued that the learners in the present study are both learning to use language (being apprenticed into the language used by scientists) and learning through language (using language as a medium to learn science), the language socialization perspective is a useful lens through which to view the discourse interactions regardless of the educational practices (transmission or transaction) which appear to characterize the research contexts. 1.8.6 Theories of social science research According to Mertens (1998), there are three main paradigms in research, each determined by the ontological, epistemological, and methodological assumptions it makes. The first of these paradigms, the dominant one which has for many years guided research in education and psychology, Mertens called positivism/postpositivism. Positivists hold the view that there is a truth or reality which awaits discovery and the researcher's job is to "uncover the facts and to understand the laws or principles that account for those facts" (Palys, 1997, p. 13, italics in original). Postpositivists, on the other hand, argue that although a reality or truth exists, researchers' theories can never prove them; they can only eliminate competing theories, thereby strengthening their own (Mertens, 1998). The reality that positivists and postpositivists aver exists can be held up for experimentation and manipulation, and the experimenter in this paradigm, as Harre (1993) noted, "is to look for 18 correlations between elementary stimuli and elementary behaviours, usually by the use of statistical analyses to identify central tendencies" (p. 14). Quantitative researchers prefer these nomothetic trends and aggregated data. They typically begin with a hypothesis which they then set out to support or refute (the hypothetico-deductive method), and they maintain social distance from those they are researching (Palys, 1997). The second major paradigm in research, according to Mertens (1998), is the interpretive!constructivist paradigm, which covers ethnographic research methodology (e.g., Spradley, 1980). Whereas positivist/postpositivist—or quantitative—researchers deliberately manipulate the contexts or objects they are studying and observe the outcomes of those manipulations, researchers working from this "qualitative" paradigm are concerned with process (Palys, 1997) and attempt to become a non-interfering part of the natural events around them, all the time recognizing that their presence may alter the situation they are observing (Bassey, 1999). The interpretive/constructivist paradigm holds the view that knowledge is socially constructed by the individuals involved in the research process, and that "human beings construct meanings for the events in which they participate" (Griffiths, 1998, p. 36). There is no objective truth or reality; the researcher's goal is to "understand the multiple social constructions of meaning and knowledge" (Mertens, 1998, p. 11). Researchers within this paradigm aim to describe, interpret, or explain social actions within their natural contexts, beginning not with hypotheses which await testing, but with ideas and questions which guide inquiry and build theory from the ground up (Bassey, 1999; Mertens, 1998; Palys, 1997). The third of the major paradigms which Mertens listed is the emancipatory paradigm which "arose because of dissatisfaction with the dominant research paradigms and practices and because of a realization that much sociological and psychological theory had been developed from the White, able-bodied male perspective and was based on the study of male subjects" (1998, p. 15). Whereas much of the ontological, epistemological, and methodological assumptions are similar to the interpretive/constructivist paradigm, the emancipatory paradigm highlights the influence of value-laden views of society, politics, gender ethnicity, and other potentially oppressive structures and policies. 19 The assumptions made by the researcher in the present study, along with the questions being asked, fall clearly into the interpretive/constructivist paradigm. Data collection strategies involve natural, social contexts and interactions with participants who "actively perceive and make sense of the world around them, have the capacity to abstract from their experience, ascribe meaning to their behaviour and the world around them, and are affected by those meanings" (Palys, 1997, p. 16). As a discourse analyst within this paradigm, the researcher is "interested in language and texts and sites in which social meanings are created and reproduced" (Tonkiss, 1998, p. 246), and uses "a common set of tools to examine how different discourses present their versions of the social world" (p. 249). Moreover, she is keenly aware of the observation made by Hitchcock and Hughes (1995) that "data does not speak for itself but only through the interpreter" (p. 324). 1.8.7 Situating the present study As noted in section 1.3, this study adopts the perspective on language shared by those working in systemic functional linguistics, and uses discourse analysis and concordancing techniques. The researcher describes this study as a qualitative (interpretive/constructivist) multiple case study design using the rationale that each case is a particular group defined by specific characteristics which bind that group together making it distinct from other groups, and that the phenomenon under study within that group can be explored by the researcher's involvement with that group as an observer and participant in its natural context (Bassey, 1999; Hitchcock & Hughes, 1995; Merriam, 1988; Mertens, 1998; Palys, 1997; Stake, 1994; Yin, 1994). Each of the four cases in this study is one classroom, an example of one social practice of teaching and learning. These four social practices, or 'cases', can be explored by looking at the discourses which construct them, thereby combining discourse analysis from an SFL perspective and ethnography, as Chapter 3 will describe. The goal of studying these groups is to examine how the participants construct causal explanations, an examination which may reflect some or all of the theories summarized above. 20 1.9 The format of this thesis Chapter 1 has introduced the topic of this thesis, stated the research questions which guide the investigation, and offered some information about the theories which both inform and influence the research. Chapter 2 will review previous studies which are relevant to the line of argument presented in the current study. Chapter 3 will describe the data collection procedures and the frameworks for the analysis of the data. The next four chapters will provide a detailed description of the four contexts which have been studied. These include two cases at the primary school level, one mainstream science and one ESL science, and two cases at the high school level, one mainstream and one ESL. These chapters include both a qualitative "thick" description of the classroom interactions and a quantitative examination of causal language features which surfaced in interviews with the students in each of these four cases. Finally, Chapter 8 will provide an analysis and discussion of the findings from the four contexts and will offer implications and directions for future work. 1.10 Transcribing conventions The discourse transcriptions in this study have been presented as clearly as possible for the reader, without heavy reliance on symbols. Some clarification may however be useful and is therefore presented below: |text] interjection by an interlocutor Speaker 1: [ overlapped speech Speaker 2: [ (italics) comment on or clarification of the action by the researcher (xx) words could not be understood for the transcription short, somewhat unnatural, pause by the speaker — sudden, abrupt change of direction in content or thought; sudden stop 21 CHAPTER 2: REVIEW OF THE LITERATURE 2.0 An overview of the chapter In the first chapter, it was briefly noted that although causal explanations are a key part of the language of science and of academic discourse in general, little work has been done to explore the development of these in science and language classes. This chapter will continue that line of argument by reviewing literature on causal discourse to establish its importance in learning and education (Section 2.1). This will be followed by a review of science teaching and science language (Section 2.2) and content-based instruction, primarily as it relates to the teaching of language through science and science through language (Section 2.3). Section 2.4 will then discuss the systemic functional perspective on science writing, particularly how its written form has developed historically, how knowledge of this development may help develop students' ability to deal with this written discourse, and how science knowledge is brought into existence in science classrooms. This leads to a summary of the work on grammatical metaphor, a key area of causal discourse (Section 2.5), followed by a discussion of Halliday's two types of patterning and their relation to science teaching (Section 2.6). Section 2.7 presents Novak's concept mapping and its connections to Halliday's two types of patterning, linking concept development to language development. Section 2.8 discusses the key studies which frame the present study and reveal the research which has been attempted regarding the development of causal discourse. Section 2.9 summarizes the chapter and restates the research questions which guide this study. 2.1 Reviewing causal discourse and its stake in learning The first chapter indicated that causal discourse from a teaching or developmental perspective has been treated in a fairly superficial manner. As mentioned in that section, Cronnell (1981) offered a list of cause and effect relations and constructions which he considered useful to teach in order to help readers comprehend text. More recently, Moreno (1997) examined a corpus of English and Spanish business and economics research articles to compare their use of what she referred to as causal metatext, lexical items which she lists 22 in the appendix. Flowerdew (1998) examined explicit cause and effect markers in a small corpus of expert and learner written texts and compared them "to ascertain the overuse, underuse and misuse of these markers on both a syntactic and semantic level" (p. 330). Her findings suggested that English as a second language (ESL) students tend to rely on a small set of linguistic devices—most noticeably conjunctions such as because, since, and as, and adverbs such as so, therefore, and thus—to construct causal texts. The author also examined texts designed to teach English for academic purposes (EAP) to see how they presented linguistic devices for constructing causal discourse. She found that many of the explicit causal devices used by expert writers were ignored by the EAP textbooks often in favor of the ones which students were overusing. Flowerdew recommended comparing the list of resources that students typically use with what the experts use, and then focusing on teaching the linguistic devices which the students typically underuse or misuse. This dependency on sentence-level markers of causality, as Mohan (1997) and Mohan and van Naerssen (1997) stated, seems inadequate given the number of studies which have examined the importance of causal structure in the recall of information in narratives. In fact, it has been posited that the findings from these studies suggest that causal relations are a critical part of narrative discourse structure (van den Broek, Linzie, Fletcher, & Marsolek, 2000). Trabasso and Sperry (1985) examined the question of "what makes a statement "important" in a text" (p. 545) and found that a statement's importance was determined by the number of connections it had in the causal network of the story and whether the event was in a causal chain from the beginning of the story to the end. O'Brien and Myers (1987) studied the effects of causal text structure on memory by measuring how long it took to retrieve concepts in the causal connections of narratives. Their findings indicated that the physical position of a concept in the text was not a factor in recall time; the results supported previous work that highlighted the importance of causal connections in narrative writing. Trabasso, Secco, and van den Broek (1984) stated that the extent to which individuals are able to represent in memory the information they hear or read, and draw upon this knowledge later, is dependent on the logical and causal cohesion of the events in the story (see also Trabasso & van den Broek, 1985). Events anchored in a causal chain in 23 the narrative were remembered better than events which had no causes or consequences, attesting to the importance of constructing causal chains in reading comprehension. The authors suggested that in teaching reading, a focus on discovering cause and effect relations would appear to be a useful skill to develop. Moreover, the authors recommended that writers should endeavor to make it easy for readers to infer causal relations in their writing. Further research on the connections between reading comprehension/memory and the causal structure of the narratives has been carried out by Fletcher and Bloom (1998), Myers (1988), Trabasso, van den Broek, and Suh (1989), Sanford (1988), van den Broek (1988a, 1988b), and Vonk and Noordman (1988). Playing devil's advocate, Giora (1996) warned that causally connected text is not necessarily coherent and easily comprehended. She cited an example and argued that the text must demonstrate other forms of discourse coherence as well as causal connectedness. In an attempt to see if writers could make a text easier to comprehend, Linderholm et al. (2000) examined the effect of revisions on the causal structure of easy and difficult reading passages with more-skilled and less-skilled readers at the college level. Quantitative and qualitative findings suggested that by revising difficult texts so that (1) their temporal line was straightforward (rather than when consequences precede their antecedents), (2) the goals of the text were made explicit, and (3) the causal coherence was clarified, readers were better able to learn from the text, as indicated by their ability to recall events and answer comprehension questions correctly. Similar revisions to easier texts, however, were found to be ineffective for the more-skilled readers. Overall, the results of the study suggested that when a text's causal structure is not clear, revisions to clarify it can benefit readers. The study was carried out with history texts, and the authors warned that "repairing the causal structure of a scientific text, for example, may not result in findings similar to the ones reported here" (p. 548). The importance of causality and causal connectedness in writing surfaces in van den Broek et al. (2000), in which the authors stated that "causality is clearly one of the constraints writers use to produce and connect new ideas to existing narrative text" (p. 718). In this study, the authors examined how writers build on what they have already written to 24 see what kind of causal relations they established as they continued their constructions of the narratives. The findings suggested that writers were constrained to maintain a causal connection as they generated new ideas in the narrative. Moreover, "writers did not simply write any action that constituted a causal consequent. Rather, they selected ideas that established a relation of causal necessity, either alone or in conjunction with sufficiency, while largely avoiding sufficiency alone" (p. 714, emphasis in original). The causal continuations which the writers offered were most often linked to the last event in the story's causal chain, no matter where that event was located temporally or physically in the story. This section has revealed three main points which impact strongly on the present research. The first is that regarding the explicit teaching of causal discourse, there appears to have been an emphasis on sentence-level or lexical markers, and there has been little if any discussion of the combination of resources which construct causal text as a whole. The second point concerns the high level of emphasis which the literature has placed on causality and the textual structure of causality in spite of the lack of work on their linguistic resources for construction. It appears evident that causal connectedness in text affects comprehension—and consequently the potential learning—of texts. The third point is that although there has been some research on causal discourse in business and economics texts, some on causal revisions in history texts, and a relatively large amount on the causal structure of narratives and how causal connectedness relates to comprehension, memory, and learning, there has been very little work on the causal structure of science texts, beyond what will be presented in the upcoming sections. 2.2 Science teaching and science language It was stated in the first chapter that language development in science classes has tended to focus primarily on the teaching of technical terms, yet it has become more recognized that learning to talk science "runs rather deeper than 'simply' learning to articulate the words and phrases of a new speech genre" (Scott, 1998, p. 74). Duran, Dugan, and Weffer (1998) insisted that the acquisition of science vocabulary takes second place to mastering the patterns and linguistic expressions for relating ideas in a variety of semiotic forms 25 in scientific contexts. Research into the 'literacy' aspect of 'science literacy' has been expanded, illuminating the depth of connection between language and science learning. Several studies will be reviewed here. (For a broader review of the last 25 years of language arts and science research, see Yore, Bisanz, and Hand, 2003.) Lemke (1990) examined the discursive practices of science classrooms. Using transcripts of oral interactions and descriptions of the context in which these interactions occurred, Lemke revealed various activity structure patterns—patterns of dialogue which regularly occur in classrooms, such as triadic dialogue, bids to question, and so on—focusing on the thematic patterns which comprise science discourse. These thematic patterns are constructed from semantic relations: "The thematic pattern of the dialogue is the pattern in which these [semanticj relationships are joined together" (p. 14). Lemke stated that it is the patterning of these semantic relationships which define science, and that frequently difficulties in understanding the content stem from differences in the semantic relationships held by the various individuals in the class rather than by the words themselves: In fact, the same scientific ideas can be expressed in many different ways, because the semantics of a language always allows us to use grammar and vocabulary in different ways to express the same meaning. The wording of a scientific argument may change from one book to the next, one teacher to the next, even one day to the next in the same classroom. But the semantic pattern, the pattern of relationships of meanings, always stays the same: That pattern is the scientific content of what we say or write, (p. x) The role of science educators, Lemke argued, is to apprentice students into the use of new thematic patterns, or new ways of meaning. This combination of meaning patterns and language was brought up again more recently by Roth (1998), who noted that concepts can be viewed "as the patterns in the language employed by students to describe and explain their science-related experiences, and conceptual change is the change in these descriptions and explanations" (p. 1020). Yet it is often the case that these patterns are left implicit, as Lemke noted, and some students fail to understand the science in them. Lemke's observation that the same scientific ideas can be expressed in a variety of ways plays a key role in the argument in favor of explicit causal language development in science because, as Carre (1981) noted, the ability to use the scientific register "is positively 26 correlated with the impression pupils give of their ability in the subject as assessed by teachers" (p. 11, italics in original). In other words, it appears that the linguistic choices a student makes may play a role in how the teacher views that student's conceptual understanding; a more literate explanation is equated with better conceptual understanding. Yet it appears that science educators in general are often satisfied to make sure that the concepts themselves are understood, and that the various ways of expressing these concepts will develop naturally and unaided alongside this conceptual understanding. Lee and Fradd (1998), for example, advocated having ESL students do hands-on science to facilitate language development, stating that "while students describe and explain their observations in science activities, they acquire the discourse of literacy and the language of science" (p. 18). Buck (2000) promoted student collaboration in ESL science classrooms, arguing that "oftentimes, one fourth-grade student could explain any concepts in ways that are more easily understood by another fourth-grade student" (p. 40). Yet this approach seems to be contrary to the many comments made that children's everyday language is not adequate for explaining science concepts (e.g., Ebenezer & Erickson, 1996; Levine & Geldman-Caspar, 1997; Moje, Collazo, Carrillo, & Marx, 2000; Nieswandt, 2001). Moreover, it has been noted that when learning science, students of all ages tend to give surface-level explanations of what they have observed if they do not have a deeper understanding of the topic (Grotzer, 2003). The notion suggested above that the language of science can be picked up during instruction also contradicts earlier views such as Solomon (1986), who in her article on children's explanations observed that "it is left to the students to pick up appropriate ways of explaining by the ostensive example of the teacher. There are always some that fail" (p. 43). Verelas, Pappas, Barry, and O'Neill (2001) believed that explicit teaching and reading of information books do not necessarily result in students picking up scientific understandings, but that teachers "mediate between the texts and the students" (p. 29). Christie (1986) argued that the failure of students to "master the skills, capacities, and knowledge of schooling goes hand and hand with an inability to handle the language structures necessary to make such mastery possible" (p. 239). 27 Although the register of scientific discourse has been discussed in the science education journals, and several recent articles have listed the characteristics typical of this register which make it problematic for students, such as the use of the passive, technical vocabulary, and nominalization (e.g., Buck, 2000; Carlson, 2000; Simich-Dudgeon & Egbert, 2000), the primary focus from the science education field has revolved around developing children's conceptual understanding using language which is familiar to the students as the basis for constructing new meanings, and two major works in this area will be discussed here (see Ogborn, 1996, for a condensed review of the research in science education). Ogborn, Kress Martins, and McGillicuddy (1996) developed a theoretical framework for examining explanations in science classrooms, based on the assumption that explanations are accounts of "how things are" (p. 7). This framework contained three main propositions: Explanations in science are analogous to stories with protagonists and actions (no matter how abstract and unfamiliar the entities might be); meaning-making in explanations consists of creating differences, constructing entities, transforming knowledge, and putting meaning into matter; and there are variations and styles of explanations to choose from. Explanation, in Ogborn et al., appears to be synonymous with the construction of scientific knowledge. Their framework suggests a regularity view of causality, with a somewhat positivistic view of scientific reality as being "out there," and it is the scientist's—or at least the science educator's—job to describe what occurs, and through that description, explain. The texts they present for examination include the semantic relations presented by Lemke (1990), but the authors do not refer to any of these relations explicitly. Many of their examples of explanations fit a pattern of description (e.g., X has the attribute Y), and in fact the authors defend the inclusion of description, labeling, and defining by saying that these must be done to build entities which can participate in explanations and are therefore part of those explanations. Ogborn et al. did not examine how the language itself was being developed or taught; their discussion appeared to reflect the earlier mentioned notion that the students' linguistic ability would develop naturally alongside their understanding of the concepts under study. Yet, as previously mentioned, not all students pick up the language of science, and as Carre (1981) noted, "pupils must come to grips with the science register if they are to succeed in the subject" (p. 11). 28 In a similar look at how students learn science concepts through the language of the classroom, Driver, Leach, Millar, and Scott (1996) first discussed their views about the nature of science and followed this with a presentation of their research, which was undertaken to explore students' ideas about the nature of science. The data were collected through interviews of pairs of students at three age levels (nine, twelve, and sixteen). The probes used in the interviews aimed to uncover, among other things, insights into how these children reasoned and the connections they made between observation and explanation. The authors developed a framework of reasoning which divided the types into three categories: phenomenon-based reasoning, in which "explanation is seen as a redescription of the phenomenon and, as such, it is seen as an unproblematic portrayal of 'how things are'" (p. 114); relation-based reasoning, in which "the explanation is seen as a generalization emerging from the data" (p. 141); and model-based reasoning, in which the explanation is "expressed in terms of a different theoretical system" (p. 115). The authors suggested that the three types may be developmental; the majority of the students in their study fell into the relation-based category, yet many of the younger students used phenomenon-based reasoning, and of the few students who used model-based reasoning, all were from the oldest group. Driver et al., while raising interesting notions of the use of models in students' explanations and the connections between observable "evidence" and explanation, exhibited a very non-linguistic analysis of explanations. Whereas this might not have posed a major problem, their discussion of the language of observations and explanations in the three reasoning types became somewhat confusing. For example, the authors commented that in relation-based reasoning, features "are described in the same language categories as observations" (p. 115). What do they mean by describe? How do they define "language categories"? In their discussion of model-based reasoning, they stated "explanations in this case... are expressed in a different language from the language of observations; the language used describes the behavior of the theoretical entities posited" (p. 116). The authors' unfortunate use of phrases such as "different language from" and "same language categories" suggest that they have examined characteristics of the discourse which distinguish, for 29 example, explanations from observations, a task which typically represents the linguist's point of view (Cloran, 1999), yet they offered no linguistic analysis to illustrate what they meant. Although an interest in the use of analogy and metaphor to explain scientific concepts has surfaced (Ogborn, 1996) and the issue of the appropriateness of anthropomorphic language in explanations has been raised (Taber & Watts, 1996; Zohar & Ginossar, 1998), science educators, as stated in chapter one, have not generally paid much attention to the nature or use of causal explanations in their teaching (Lawrence & Pallrand, 2000), and "'explanation' still remains a largely unexplicated notion" (Ogborn, 1996, p. 159). Several authors have advocated the explicit teaching of science language (e.g., Henderson & Wellington, 1998; McGinn & Roth, 1999; McKeon, 2000; O'Toole, 1996; Prophet & Towse, 1999; Ritchie & Tobin, 2001; Yore, Craig, & Maguire, 1998; Zohar & Ginossar, 1998), but causal discourse as a key element of scientific explanations remains relatively unresearched in favor of research on conceptual understanding (see Ogborn, 1996, for a review of the research in this area). The linguistic element has rarely been mentioned. As Leach and Scott (2000) noted, learning science involves learning how scientists explain concepts, and although students are expected to advance their conceptual understanding and to be able to demonstrate it linguistically—Gruenwald and Pollak (1984) insisted that progression to more abstract tasks in science will not occur until a student can use language to express understanding of a concept—causal language development in science classes has not yet been targeted by the science educators as an area deserving explicit attention. 2.3 Teaching content through language and language through content The first chapter briefly introduced the use of key visuals to make academic content accessible to readers and to serve as an organizer to help students compose academic texts. These key visuals are graphic representations of knowledge structures, and they aim to help develop thinking skills and language (Mohan, 1986; 2001) while helping the student learn content through language. According to Mohan, there are six core knowledge structures—description, sequence, choice, classification, principles, evaluation—which 30 are common to most, if not all, academic content areas at all levels. These six make up the Knowledge Framework, a theoretical framework for analyzing discourse and social practice, based on Halliday's systemic functional grammar and grounded in a language socialization perspective. The Framework is useful for discussing language and content teaching and learning because education itself is a social practice mediated to a great extent through language. What the Knowledge Framework offers teachers, therefore, is a tool for breaking a lesson's activity into pieces which highlight particular thinking skills, or knowledge structures. By focusing on particular knowledge structures in isolation, teachers can help develop students' language ability within that structure as well as help students access the structure of the knowledge being taught. Several reports of successful implementation of the Knowledge Framework have been presented (e.g., Early, 1989, 1990, 1991a, 1991b; Early, Mohan, & Hooper, 1989; Early & Tang, 1991; Tang 1991, 1992, 1997, 2001). Stemming from the same functional view of language and strongly influenced by Mohan's Knowledge Framework is the Project Framework (Beckett & Slater, in press), a visual planning tool for use in project-based instruction. The Project Framework, which helps make explicit the connections between content, language skills, and thinking skills, was developed and tested in a university-based, second-year academic, content-based language class with a group of Japanese students learning English in a one-year exchange program. The students reported that the Project Framework made the connections between content and language explicit and thereby helped them understand how their self-initiated research projects were promoting language development simultaneously with content learning. Without the framework, these connections were left implicit, and at the high school level, some students expressed dissatisfaction with project work in their ESL classes because they were not convinced of the language development potential (Beckett, 1999). Cantoni-Harvey (1987) discussed the teaching of science and language, claiming that "students who learn to write science reports accurately and appropriately can apply this ability to other content areas" (p. 167). She advocated hands-on science experimentation for limited English proficient (LEP) students with the rationale that having the visual context would allow them to extract meaning more easily. Students in the higher grades, the author 31 stated, are at a disadvantage because the language becomes much less contextual. If a learner is unable to participate in a bilingual program or get private tutoring in science, the author suggested, it might be necessary "to interrupt her study of science until she becomes able to resume it in a class taught entirely in English" (p. 166) because it would be unreasonable to have students attempt to read scientific texts in English or to have teachers reteach more elementary concepts to LEP students. Cantoni-Harvey's emphasis is on learning strategies and activities to promote language development through the use of the four language skills of reading, writing, listening, and speaking. As the author noted, "the linguistic and cognitive ability he [the student] gains through reading and listening prepare him for advanced academic tasks that require receptive as well as productive skills" (p. 20). Rupp (1992) also stressed the importance of hands-on discovery learning in science for LEP students because of the context for language support which this teaching approach provides. The author noted that the cognitive abilities of second language learners may be more advanced than their language use suggests and emphasized the importance of dialogues between students and between the teacher and students during these hands-on experiences as an essential part of learning science and science language. Rupp did not provide examples to show how this language learning might occur. Parkinson (2000) described a theme-based language course for teaching science and technology at a South African university, advocating the teaching of language through science rather than through a general language course. By grounding language teaching in science teaching, the author argued, the needs and interests of science students are addressed. Students become familiar with the genres and literacies associated with science and the register used in the genres while moving forward in content learning. O'Malley and Chamot (1990; also Chamot & O'Malley, 1992) developed the Cognitive Academic Language Learning Approach (CALLA) , an instructional model which is not intended to duplicate the mainstream curriculum, but is designed to prepare ESL students in upper elementary and secondary schools for the vocabulary, structures, and functions of English they will encounter in mainstream classes. Developing academic language skills and learning strategies is the primary focus of C A L L A ; content appropriate to the grade 32 level is made comprehensible by "providing additional contextual support in the form of demonstrations, visuals, and hands-on experiences, and by teaching students how to apply learning strategies to understand and remember the content presented" (p. 194). Academic language, the authors claimed, is particularly difficult because it is context reduced and cognitively complex, and because science is taught using a discovery approach with context-embedded, hands-on activities, the authors recommended it as the best content area to begin with. O'Malley and Chamot made no reference to the difficulties of science language which others have noted, but suggested that the move from context-embedded language to the context-reduced form would be done "through a whole language approach in which all language skills are applied and integrated for all areas of the curriculum" (p. 196). It would appear that content-based language programs are useful bridges to mainstream classes. Kasper (1997) provided quantitative support for the use of content-based ESL instruction as a way to ease students' successful transition to mainstream classes at the college level. The study involved 152 students of which 73 were in the experimental group. The findings suggested that the experimental group generally did better academically than the control group, leading Kasper to conclude that content-based ESL courses "provide ESL students with the linguistic and academic tools they need to succeed in the mainstream college curricula" (p. 318). Gaffield-Vile (1996) also recommended sheltered content courses because they motivated ESL students more and introduced them to the academic culture of English-speaking cultures through the types of assignments they contained. Many of the authors of content-based instruction in ESL science appear to target the teaching of science vocabulary as the main need. For example, Straw, Sadowy, and Baardman (1997) stated that although students require social language to function in school, the development of "subject-specific vocabularies" is critical "to make students contributing members of the school and academic community" (p. 39). This is echoed in the research questions which guided Carroll and Gallard (1993) in their inquiry into whether students were learning to mimic teachers' scientific vocabulary or whether there was real understanding of the concepts being taught. Vocabulary as a key issue for learners is also clearly highlighted in the British Columbia Ministry of Education's 1999 ESL learners: A 33 guide for teachers. This document advised teachers to be conscious of the vocabulary they use and to teach the vocabulary of the subject. It also recommended simplifying sentence structure to facilitate comprehension. While language through content programs can be useful bridges, academic language development must go beyond simple vocabulary teaching and continue after the students' promotion to the mainstream classes because many of these students need continued support to develop their academic language ability, particularly given the five to seven years that it takes to catch up with their native English speaking peers (Cummins, 1984). Key visuals offer this support by helping make explicit the structure of the knowledge, yet as the first chapter suggested, key visuals can produce a variety of different texts, depending on the linguistic resources the speaker/writer chooses. Mohan (1989, 2001) offered comparisons of classification texts written by skilled and less skilled writers and demonstrated how the two texts reveal "something of the complex, and only partly conscious, discourse decisions a skilled writer makes, and an unskilled writer needs to develop" (2001, p. 119). Key visuals offer a way for students to "communicate about information while learning to shape text" (1989, p. 113), but to become skilled composers of discourse, students need continued development and expansion of the linguistic resources available in the English language. With causal discourse, this involves looking beyond sentence-level features, yet this does not seem to be a major area of research in the content-based language learning literature. The brief review in this section has aimed to show that although there are a variety of perspectives on content-based instruction in the literature, very few acknowledge the role that the development of causal discourse has in the academic language proficiency which students—both ESL students and those who speak English as a first language—need to succeed in higher-level studies, and several approaches advocate the same inadequate sentence-level focus that has been introduced in earlier sections of this thesis. Although teaching approaches based on the Knowledge Framework (Mohan, 1986) address this issue to some extent, a deeper exploration into the causal language students are currently using and how that language might be developed is much needed. 34 2.4 Science discourse: A systemic functional perspective Probably the richest source of information on causal discourse in science is Halliday and Martin (1993), which offered a rich description of the language scientists use. Halliday and Martin argued that the language of science—discourse which can be challenging and alienating to both children and adults alike—reflects the evolution of scientific knowledge itself. The authors suggested that "physical scientists led the way in expanding the grammar of the language, as they found it, so as to construct a new form of knowledge" (p. 67). The authors argued that this "new" form of knowledge which is taught in schools replaces common-sense understanding, offering an alternate interpretation of the world. They showed how science organizes knowledge in ways that go beyond the observable and argued that for the scientific register of English to be effective in constructing technical taxonomies, it became characterized by grammatical metaphor and in particular by the changing of clauses into noun phrases. As a nominal group, any happening could be defined, classified, or related causally to other happenings in new clauses. This evolution is schematized in the following manner (p. 66): From a happens; so x happens because a happens, x happens that a happens causes x to happen happening a causes happening x To happening a is the cause of happening x As well as technical taxonomies and grammatical metaphor, Halliday described five further categories of scientific English which can pose problems for readers, suggesting that there are other features which could also be added. Interlocking definitions create difficulties because the terms are often defined by other terms within the same text, and these other terms may also require definition. Special expression can also be problematic, according to Halliday, although examples tend to be more common in mathematics than in science. Science discourse also tends to be lexically dense with syntactic ambiguity and semantic discontinuity. A l l of these features work together to create the distinctive quality of scientific discourse. 35 The two main genres in science discourse, according to Halliday and Martin, are report, used to construct taxonomies to describe how the world is organized, and explanation, which explains why the world is organized that way. The primary difference in the two genres "is that reports focus on things while explanations focus on processes" (p. 206). In that light, explanations tend to contain more action verbs than reports do, and the actions in explanations "are organized in a logical sequence" (p. 191). Martin referred to these logical sequences as "implication sequences," a term which others following Halliday and Martin have adopted in discussing scientific explanations. Rose (1997, 1998), for example, showed how the sequence of events in written technological and scientific explanations link together with each step representing an effect or outcome of the preceding step, with causality either implicitly or explicitly stated. A similar model of explanation was discussed in Wignell (1998), who suggested that an explanation is a sequence of events linked temporally, causally, or conditionally. The clarity of an explanation is also reflected in the causal resources which the author chooses to construct the implication sequence (e.g., Unsworth, 1999). Unsworth (2001a) examined written science explanations about coal and sound using three types of analysis: from a genre perspective, by looking at conjunctive relations, and by exploring the use of nominalization. He showed how 'events' are packaged into 'things' and how these things are unpacked into events as the implication sequences, part of the schematic structure of the explanation genre, unfold and technicality is built. He advocated helping students understand how this process is carried out by discussing with them not only how to unpack highly nominalized text but also how to transform the more congruent clauses back into the grammatically metaphoric so that students would be able to construct and deconstruct similar explanations. Young and Nguyen (2002) compared teacher-talk and textbook discourse on the topic of mirrors in a twelfth-grade physics class, using systemic functional grammar. The authors revealed several differences, including (1) the use by teachers of the first person with material processes in the active voice compared to the passive and third person visible in the written text, (2) more mental processes in the written mode than in spoken, and (3) the 36 teacher's use of explicit statements of cause and effect versus the writer's style of explaining through the description of a process. Young and Nguyen cautioned against generalizing their findings noting that "there are certainly teachers whose style of presentation is much closer to the discourse of the textbook and there may be textbook authors who attempt a more interactive style of presentation" (p. 365). Schleppegrell (1998) used systemic functional linguistics to analyze science "description" written by grade seven and eight students from ethnically diverse backgrounds. She discovered that the students made a variety of grammatical errors, most often concerned with inflectional endings. They also relied heavily on a small set of verbs for their descriptions and exhibited problems with basic sentence structures. With the exception of be and have, verbs were frequently in tenses other than the timeless present, connected to the specific context which the students were in, such as in the example "in the picture, they are drinking water" (p. 200). Finally, the thematic choices made by the writers reflected different, sometimes non-scientific themes, such as person or specific rather than generic themes. Schleppegrell concluded by suggesting that a basic understanding of functional grammar can help teachers help students improve their writing in science by making explicit the language features which are characteristic to specific genres and registers. She argued this further in Schleppegrell (2001), stating that "knowing how to make the linguistic choices that realize appropriate texts is an aspect of sociolinguistic competence" (p. 536). Continuing along a similar line, Schleppegrell (2002) analyzed science lab reports written by one native English speaker and three ESL speakers in an upper division university course in chemical engineering. She found that the lack of linguistic resources available to the ESL students led to texts which were less authoritative with obscure meanings. It was not the grammatical errors that the ESL students made that were responsible for the problems with their texts; the difficulties were frequently in the choices the students made from the interpersonal and textual metafunctions, including the grammatical metaphor involved in the construction of logically progressing discourse. The author concluded that ESL writers rely heavily on resources to construct ideational meaning without realizing the impact that their interpersonal and textual choices have on their writing, and that assistance needs to be 37 provided to help these students learn how their grammatical and lexical choices construct meaning in science. Deconstructing causal discourse into predictable words and patterns has been criticized by some authors such as Watkins (1999), who claims it risks advocating a rigid formula for recreating science genres. Along a similar line, Sawyer and Watson (1995) presented three arguments against adopting a genre approach to teaching science writing. The first challenged the idea that science discourse constructs science meaning, stating that it is instead "a continuum of registers and styles which does not include the scientific content as a variable" (p. 69, italics in original). Secondly, the authors argued that the type of science language described by Halliday and Martin is not suitable for school texts; rewriting the texts is necessary rather than "inducting pupils into the linguistic features of expert-to-expert scientific prose" (p. 70). Thirdly, they questioned the model of learning theory that the genre school adopts, asserting that a genre approach to teaching science equals "the rejection of a constructivist view of learning and that... the ideology of the genre school is firmly within a Transmission model long ago discredited as an effective model for learning" (p. 75). Sawyer and Watson's view regarding the lack of constructivist learning and teaching was based strongly on work done in 1976 by Douglas Barnes, which was cited heavily in their argument. The more recent literature which they cited takes a somewhat kinder view concerning the connections between science language and science content. Still, the authors maintained a critical stance against the genre approach to teaching science, omitting reference to how teachers use constructivist ideas to bridge students' existing language features and understandings to those held by the field of science, and instead insisting that the features which Halliday and Martin and the genre school describe are "not necessary to the conveying of scientific knowledge or modes of thought" (p. 71, italics in original). Despite arguments that science knowledge can be conveyed— note the idea of transmission rather than construction which Sawyer and Watson used in the above quote-without the language which Halliday and others have described as characteristic of the field, it is nonetheless important for teachers to consider these linguistic features because, as Derewianka (1990) stated, "if children have an explicit knowledge of what language 38 resources are available, they are in a better position to make informed choices when developing texts of their own" (p. 5). Yet by looking at the views of both the "genre school" and authors such as Sawyer and Watson, it becomes apparent that both students and teachers need to be aware that simply presenting the resources in a grammatically correct manner does not result in the construction of literate scientifically sound discourse, a point which Halliday emphasized: Whenever we interpret a text as 'scientific English', we are responding to clusters of features.... But it is the combined effect of a number of such related features, and the relations they contract throughout the text as a whole, rather than the obligatory presence of any particular ones, that tell us that what is being constructed is the discourse of science. (Halliday & Martin, 1993, p. 56) From the combined perspectives, therefore, when teaching science, language and content should not—cannot—be dichotomized into either a focus on language forms or a focus on science content. Instead, teachers and learners must work together to construct appropriate scientific meanings, introducing and using language features which are appropriate for the students' level of development, and constructing these meanings through their linguistic interactions in the classroom. In other words, developing students' ability to construct causal meanings in science involves much more than simply offering students a list of characteristic lexical items and grammatical structures; although these resources are indeed important, teaching students to read and write academic text involves socializing students into new ways of looking at the world and new ways of linguistically constructing causal relations. The ability to handle grammatical metaphor plays an important role in these constructions. 2.5 Causal explanations and grammatical metaphor In the first chapter, the term grammatical metaphor was introduced and defined as being similar to lexical metaphor in that both involve linguistic transformations, but "instead of being a substitution of one word for another,... it is a substitution of one grammatical class, or one grammatical structure, by another" (Halliday & Martin, p. 79, italics in original). Whereas in lexical metaphor there is a literal meaning which is different from the metaphorical term(s), with grammatical metaphor, the non-metaphorical construction is 39 referred to as being the more congruent form (Halliday, 1994). Considered by Derewianka (1995) to be a vastly undertheorized notion, grammatical metaphor is poorly handled in discussions of academic language development. Typically its only representation is nominalization, which is not surprising given that it has been common to treat grammatical metaphor and nominalization as interchangeable (Derewianka, 1995). Nominalization is frequently defined as a characteristic of written language, often used in school textbooks "to achieve economy of expression" (Crowhurst, 1994, p. 33). A summary of the arguments in favor of and against the use of nominalization is offered in Perera (1984). Yet nominalization is only one type of grammatical metaphor (Eggins, 1994). In a longitudinal study of her English-speaking son from age five to fourteen, Derewianka (1995) documented the development of various kinds of grammatical metaphor. Her study offered empirical evidence for the suggestion that adult language and child language differed primarily in the use of grammatical metaphor and showed how different types of metaphor emerged at different times, with a dramatic increase in use at around nine or ten years old. Her findings support Halliday's observation that "students well into secondary school may still find it difficult to comprehend, even if they have been educated throughout in the English medium" (Halliday & Martin, 1993, p. 82). With regards to the shift from the more congruent clause constructions to the more grammatically metaphoric, Painter (1999) discussed the development of causal relations in her son, Stephen, from age two-and-a-half to five years. Interestingly in light of Halliday's observation above, she found that initial expressions of reason involved the hypotactic linking of two processes, and the last to occur were "metaphorical within-clause expressions" (p. 312). Both Painter and Derewianka also noted that their data revealed the social nature of language learning as they modeled the use of grammatical metaphor for their sons. In Derewianka's recommendations for further study, she advised exploring teacher intervention which might facilitate the development of grammatical metaphor. Mohan and Beckett (2001) responded to this by discussing recasts in causal explanations. Using examples from interactions in a project-based language and content classroom, the authors argued that the 40 teacher's recasts, by using grammatical metaphor, were able to model a more literate way of meaning by turning the students' more congruent forms into ones which were less congruent. The students were not initially able to understand the recasts, but the teacher had created a "zone of negotiation" (p. 151) in which language development could occur. The students' rephrasing of the teacher's recasts revealed the successful development of less congruent language. 2.6 Grammatical metaphor and Halliday's two types of patterning Halliday (1998) argued that grammatical metaphor, in particular nominalization, plays a powerful role in making meaning in science. It "creates a universe of things, bounded, stable and determinate; and... of relations between the things" (p. 228). In other words, these "things" can be technicalized (renamed and reclassified), and processes can be used to relate the "things" in reasoned arguments.1 Learning science involves these two types of patterning: creating new technical taxonomies, which differ from everyday understandings, and then relating the participants in the taxonomies to each other and to other classifications. Wignell, Martin, and Eggins (1993) defined the process of technicalizing as involving two steps: renaming everyday terms and reclassifying them into scientific taxonomies. This is done using grammatical resources such as projection and elaboration. For example, a technical term can be introduced by a projecting naming process such as we say that X or we call this Y, or by an elaboration through an identifying relational clause such as X is defined by Y. The authors stated that technicality "refers to the use of terms or expressions... with a specialized field-specific meaning" (p. 144) and that "different fields will name, reorder, or reclassify similar things differently according to what is 'emic' (meaningful or relevant) to that field" (p. 139). Teaching the field of science from this perspective, therefore, involves renaming and reclassifying everyday things to create new technical taxonomies. Carroll and Gallard (1993) argued that the introduction of new technical terms promotes new learning. They stated that if the teacher talks about a scientific event using terms from 1. Hartnett (2001) presented several factors involved in the use of nominalization, including the use of previous knowledge to build new knowledge, rhetorical organization, inclusiveness, and efficiency. She also stated that nominalization use changes as language changes, and that this use may be on the decline. 41 the students' everyday vocabulary, the students are already familiar with the concept and therefore no negotiation of meaning is required and no new meaning-making occurs. Halliday (1998) showed how the resources of the grammar are used to build sequences of reasoned arguments and the role grammatical metaphor plays in that process. In constructing arguments, the author stated, the grammar construes both experience, through the ideational meaning, and the grammar itself, by creating a cohesive and coherent piece of discourse. He described "the 'general drift' of grammatical metaphor" (p. 211), which sees movement from relator (conjunctions) to circumstance (adverbial phrases) to process (verbs) to quality (adjectives) to entity (nouns). Halliday's two types of patterning offer a way to look systematically at the teaching of science language and content. Classroom interactions can be examined to see how teachers are creating technicality and modeling new, more literate ways of arguing. Analyzing the discourse from this perspective can also help researchers and educators see the language resources students are using to construct causal explanations in science and may perhaps be exploited to judge to some extent how satisfactory these explanations are for their grade level. 2.7 Mapping the development of concepts and language Novak has for several years used concept maps to investigate concept learning. In his 1998 book, he offered concept maps drawn from interviews with a boy named Paul about his understanding of matter, done in both grade two and grade twelve. In Paul's earlier interview, the concepts which appeared in his map numbered twenty, with only one technical term ("oxygen") included. By grade twelve, there were fifty-one concepts, of which more than half were either technical or metaphorical terms. Novak noted that the later effort showed both "quantitative and qualitative growth in Paul's conceptual/ propositional knowledge about forms of matter" (p. 67). Novak (1988) provided a similar example using two maps of matter drawn by a student named Phil in grades two and twelve. Given the notion that "conceptual change is an ongoing process in which the child, in collaboration with a teacher or other student, integrates everyday concepts into a coherent system of concepts" (Howe, 1996), several authors have also examined the use of concept 42 mapping to reveal concept learning. Harrison, Grayson, and Treagust (1999) noted that the participant in their case study, Ken, had increased the number of entries and connections on his concept map by the end of forty periods of studying heat and temperature. The authors attributed this increase to Ken's greater understanding of the topic. Jones, Carter, and Rua (2000) presented concept maps drawn by their participant, Cary, to show the differences between the grade five student's understanding of heat and convection before and after being taught the topic. The authors also presented a scientist's concept map of the same topic which contained only nineteen concepts. They argued that the number of concepts may in fact be lower in maps drawn by specialists in the field than by learners. The lower number of concepts in specialist maps can be explained easily by considering Novak's (1998) explanation of representational learning. The author stated that once a child learns that all dogs have certain common characteristics, he or she has acquired the concept dog. Similarly, children may recognize similarities between dogs, cats, lions, and tigers long before they learn the word carnivore to label or represent this group of flesh-eating animals, (p. 37) A specialist may include a word like carnivore, whereas non-specialists may not have this term in their linguistic resources and may instead include examples of the class word. What is therefore important to note about this type of technical term when considering the concepts which exist in a particular map is that the word itself may be constructing more than a simple one-to-one representational meaning. Hence, the number of terms presented in a concept map may be fewer in one created by a specialist in the field, but the meanings which the terms incorporate create a much larger network of ideas. Pollak (1994) captured this idea when he stated that as science matures, its conceptual structure becomes more efficient, and is capable of accommodating vastly more information... it is no longer necessary to remember vast amounts of detailed information. In fact, a fundamental goal in the development of models and theories is inclusiveness—to account for more and more on the basis of the fewest and simplest fundamental ideas.... By virtue of greater sophistication, as a science matures things become simpler, not more complex, (p. 96) Indeed, the differences between the preconcept and postconcept maps in the Jones et al. study revolved around the types of concepts included. Whereas the preconcept map 43 included few technical terms or metaphoric entities, the postconcept map contained several. The number of everyday concepts dropped in the latter. In the scientist's concept map, there were no everyday terms; the concepts captured technical terms, laws, theories, and nominalizations of processes, each of which involve knowledge which goes beyond the simple representational connections between observable ideas or things and their labels, yet the scientist obviously felt no need to include this knowledge because he already had a technical or metaphorical term that captured the knowledge which had already been constructed. It is this type of difference which seems to play a key role in capturing the development of conceptual knowledge in the concept maps. Aside from the numbers and types of terms included in the concept map, another difference can be found in the relations or propositions which were constructed around and amongst the concepts. Novak (1998) constructed Paul's grade two map using connections such as is made of, as in, is, into, by, can, and when. Paul's grade twelve map, in comparison, included such language as is made of, but also makes up, causes, produces, aids, comes from, means, and forms. In other words, more causal relations were evident in the more mature concept map. A similar pattern emerges when the three maps in the article by Jones et al. (2000) are examined. The preconcept map in their study contained relational processes such as is, gets, turns, and has, as well as circumstances such as in oven and into the air. The postconcept map added to the variety of relational process with involves and is called, but also included more causal processes and circumstances of cause, such as can be made by. The scientist's map included not only causal processes, but processes of evidence and metaphorical constructions: causes, explains, leads to, is due to, depends on, involves, drives, is measured in, and is the source of. It could be suggested, then, that the scientist's concept map includes (1) a much higher level of technicality, with abstractions such as theories and models, and nominalizations representing and containing more detailed but invisible conceptual structures, as well as (2) more causal and metaphoric ways of reasoning among them. This finding, as suggested by examining the concept maps presented in papers such as Novak (1988, 1998) and Jones et al. (2000), parallels Halliday's two types of patterning, revealing how the development 44 of science concepts and science language appear to go hand in hand. But how has the development of causal explanations—which involve these two types of patterning—been researched from a linguistic perspective? The next section will review the key studies which have been carried out regarding the development of causal explanations in school science. 2.8 The development of causal explanations in school science The development of causal explanations has been explored, although not in adequate depth, both from the classroom level, where teachers interact with students to deepen their understanding of the topic, and from the curricular perspective, through texts written at various levels by experts. This section will review the available literature of both areas. 2.8.1 The development of causal explanations in the classroom Haneda (2000) explored the interactions between a teacher and two grade three Chinese-Canadian students as they conducted and discussed an experiment on refraction. The author was particularly interested in examining how the students participated in the interactions, and in the connections between the talk and the students' subsequent writing task. Haneda noticed differences between the two students with regards to their involvement with the topic in both the interaction and writing tasks, with the girl—Jasmin—using different types of talk as the interactions progressed from a recount to an explanation attempt. The author further noted that talk which was concerned with procedure saw the children taking the lead, but when the task shifted to explanation, the teacher scaffolded the students' understanding so that Jasmin could begin to reason logically about the topic. Alex, the boy in the study, remained at the level of doing and observing, with the author inferring that the interactions were beyond what Vygotsky termed the zone of proximal development (Vygotsky, 1978). Haneda concluded that although it was evident that interactions between teachers and students can help students learn, more research should be undertaken, particularly with regards to probing how these interactions can promote deeper thinking. Gibbons (1998) examined the language development of nine- and ten-year-old ESL students learning about magnetism by following the progression from the hands-on activities 4 5 which Veel (1997) stated begins an investigation of a topic, through to the students' written reflection on their learning. She addressed this development from a register perspective, describing in detail the moves which occur in the classroom during three stages: the hands-on activity of lab experiments, the teacher's scaffolding of decontextualized recounts of the labs, and the more generalized written discourse. She focused on the move between these stages, highlighting the way the language shifts from the interpersonal to the ideational, and elucidating the role that the teacher plays in helping students adopt the more academic registers of science. Gibbons offered a valuable description of the natural teaching progression from experimentation to teacher-guided oral discussion to writing, discussing the use of the teacher's recasts and encouragements in helping the students appropriate the new science-specific lexis and a more decontextualized way of talking. She argued in favor of having the students come to some understanding of the topic through hands-on activities before introducing and reinforcing the new science language, yet there is no comparative data to see what the discourse might look like if the new terms (e.g., repel) are presented before the students become engaged in the action. Is it the order of presentation or the teacher's strategies for connecting experience to language that becomes important in the teaching? Are there more strategies than the recasts and encouragements the author mentioned? As Gibbons stated, more classroom-based research into how students learn language in school is needed. In Gibbons (2003), the author provided deeper insights into the strategies her teachers used to bridge the students' experiences and everyday language to a register appropriate for school science. Slightly different from her 1998 article, her research participants in this paper were eight- and nine-year-old mostly ESL students and their teachers who were trained ESL instructors teaching language and content simultaneously. Although the author framed the study along a mode continuum from the oral, context-dependent language of group work to the written, highly context-independent discourse of a science encyclopedia written for youths, this article focused on the stage she referred to as "teacher-guided reporting" (p. 256), in which the bridging of students' existing understandings and abilities and the new 46 target knowledge occurred. Gibbons discussed four key ways the teacher used to mediate language learning (p. 257): 1) mode shifting through recasting, 2) signaling to learners how to reformulate, 3) indicating the need for reformulation, and 4) recontextualizing personal knowledge. These four ways were supported with discourse examples analyzed to reveal the moves the teachers were making. Particularly interesting was the example of mode-shifting, in which the teacher's discourse was divided into columns indicating "situationally embedded," "everyday," and "formal," making it easy to see how the teacher was drawing parallels between the students' current linguistic forms and the target school language. Gibbons concluded her paper by stating that her examples were not unusual and that "similar interactions between teachers and students probably occur daily throughout hundreds of classrooms without teachers being explicitly aware of the nature of their responses" (p. 268). Her data, however, were limited to two trained ESL teachers working in two mainstream classes heavily populated with ESL students and containing eight- and nine-year-old children. Would teachers without specific language training use the same strategies to teach science content and language? Would the same strategies be used at different age levels? Further qualitative research needs to be carried out in different contexts, both ESL and non-ESL and with younger and older students, to paint a more complete pictures of what the interactions in classrooms look like and how they promote—or even fail to promote— language and content learning. 2.8.2 The development of causal explanations across grade levels Veel (1997) noted that the school curriculum reflects a progression similar to the evolution of science language, from the more observable, sequential treatment of science content to the more abstract, causal language. He claimed that this "idealized knowledge path" (p. 189) mirrors the way that children acquire language. Explanations for younger students tend to be sequential accounts of observable events, and it is only when the student can deal with more abstract or theoretical concepts that the explanations progress beyond 47 the language of sequence. Veel proposed that there are four linguistic indicators which mark the development of content and move students from the younger, sequential explanations towards "the abstract, technical and 'transcendental' kinds of meaning we expect of adult, educated discourse" (p. 188). These four indicators—an increase in lexical density, a higher number of nominalizations and abstractions, a shift from temporal to causal conjunctions and a move from external to internal text organization—were demonstrated in Veel's selection of written genres. Although Veel makes his claims using only four short texts from four different topics-making sugar, sea breezes, physical weathering, and buoyancy and density—his work is valuable because it provides a clear set of hypotheses which he illustrates well using these four texts. Yet there are difficulties with these hypotheses in that beyond suggesting that there will be an increase in lexical density and nominalizations, Veel focuses on temporal and consequential conjunctions as well as internal and external conjunctions, therefore putting a great emphasis on the role of conjunctions in the development of his knowledge path. His basic hypotheses about these are: • Temporal conjunctions decrease • Consequential conjunctions increase • External conjunctions decrease • Internal conjunctions increase Moreover, Veel does not address how the temporal and consequential conjunctions or the external and internal conjunctions are related to lexical density or nominalizations beyond suggesting that "there are recognizable syndromes of language features, and that these features work to produce a kind of knowledge path along which ideal pedagogical subjects will move into fully fledged scientific discourse" (1997, p. 190). Rather than putting such a strong emphasis on logical relations as Veel has done, a closer examination needs to be taken to see the connection between these conjunctions and the move towards grammatical metaphor. To explore Veel's hypotheses more fully and to elaborate on the role grammatical metaphor may play in the knowledge path, Mohan, Slater, Luo, and Jaipal (2002) used a computer concordancing application combined with hand analysis to examine discourse 48 samples from a science encyclopedia for learners aged eight to fourteen and from one targeted for older, university-level students. The features for analysis were taken from lists of causal items provided by previous concordancing studies (e.g., Fang & Kennedy, 1992; Flowerdew, 1998). To address the issue of the move from conjunctions—or relators as Halliday (1998) terms them—to the more grammatical metaphoric constructions, the authors proposed that there be two axes of development working together, as shown in Figure 2.1. Rather than the knowledge path developing primarily along the temporal/causal axis through conjunctions with entities (nominalizations and abstractions) situated outside of this axis as Veel has described, Mohan et al. offered a schematized developmental path which moves out from the lower left corner at a roughly 45-degree angle. This model suggests, as does Veel, that there is a shift from temporal conjunctions towards ones which signify causality and proof at the same time that there is a shift from external to internal conjunctions. But the model goes beyond Veel's idea to suggest that there is also a move away from conjunctions as the primary marker of causality towards the more grammatically metaphoric constructions such as circumstances, processes, qualities, and entities. The order of these more metaphoric categories stem from Halliday (1998), who described this progression as "the 'general drift' of grammatical metaphor" (p. 211), from the clause complex, through to clause, and finally to nominal group, the most metaphoric construction. Figure 2.1: Idealized developmental path for cause (Mohan et al., 2002) proof (internal) causal (external) temporal relator circumstance process quality entity 49 When Mohan et al. tallied their findings from the corpus analysis and held them up against Veel's hypothesis and their own idealized developmental path, they found that there was a tendency for the encyclopedia written for the younger group to have more temporal external conjunctions than the discourse constructed for older readers, confirming Veel's hypothesis that temporal conjunctions decrease. Contrary to Veel's findings, however, external consequential conjunctions also decreased. With regards to the shift from external conjunctions to internal ones, when only temporal and causal conjunctions were considered, there was a move, as Veel had hypothesized. However, when the total number of internal conjunctions were brought into the picture (i.e., temporal, consequential, additive, and comparative, as Veel outlined on p. 186), there appeared to be no difference between the encyclopedias written for younger and older audiences, thereby refuting Veel's hypothesis. Mohan et al. went on to track the frequencies of various processes, qualities, and entities in the two corpora as well. What they discovered was that whereas the numbers of causal processes dipped slightly in the encyclopedia for older students, the number of proof processes increased, suggesting a more metaphoric move through processes from external to internal. Moreover, the frequencies of causal qualities and causal entities also rose. It should be noticed, however, that not all categories of causal features were examined by Mohan et al. Temporal processes, qualities, and entities, and all circumstances as well as the proof qualities and entities were not reported. Furthermore, only items which had been listed in the previous concordancing literature were counted even though there were potentially other items existing in the data. Despite this, though, their findings appear to suggest that between these two encyclopedias there is a move from the use of conjunctions to more metaphoric ways of constructing meaning. In general, the findings from their corpus study suggest that the developmental path between temporal and consequential conjunctions does not work out as Veel had hypothesized, but that a path can work if Halliday's "general drift" toward grammatical metaphor is built into the theory, focusing on time/cause relations. Veel (1997) and Mohan et al. (2002) have attempted to show developmental moves at the curricular level of school science, but in his chapter, Veel inferred that this knowledge path is also noticeable at the level of the teaching unit. He stated: 50 The investigation of a topic in the classroom, for example, will frequently commence with physical activities such as experiments and observations, proceed to more generalized 'bookish' study of the topic and conclude with an investigation of how the topic in question affects people's lives. In terms of written genres, this will involve a shift from procedures and procedural recounts to explanations and reports and then to expositions and discussions, (p. 174) Veel's observations about genres and classroom investigations are also captured in Unsworth (2000, p. 249; 2001b, p. 125), who charted the progression of the principal genres involved in "doing science," "explaining events scientifically," "organizing scientific information," and "challenging science." Unsworth further noted that oral language typically uses conjunctions to construct logical relations whereas in writing, the same meanings are constructed using nouns and verbs. His observations, which were supported with examples in Unsworth (1999, 2001b), were concerned with a mode continuum from oral to writing, and the register shift which accompanies this shift. Is there a difference in the use of conjunctions versus the more metaphoric features within the oral mode as the understanding of field deepens? Unsworth's work, while valuable in showing the general and distinctive differences between spoken and written science discourse, did not examine how or if the language changes as understanding of the field grows, even when the mode remains constant. Section 2.8 has discussed the literature which impacts strongly on the current research project. Veel (1997) and Mohan et al. (2002) examined the development of causal discourse over the school science curriculum. Both studies served to highlight the observation made in Derewianka (1995) that the development of grammatical metaphor is critical to success in secondary school. But both studies were concerned solely with texts written by experts. Unsworth (1999, 2001b) described the differences between oral and written science discourse, but was concerned with native English speakers/writers and did not address the development of academic register within the oral mode alone (or even within the written mode). Haneda (2000) detailed the development of two students learning through interactions with the teacher, but her study raised questions as to why one student benefited from the interactions while another didn't, and how interactions might help all students understand. Gibbons (1998, 2003) presented discourse evidence to show the 51 role her ESL teachers played in helping the students learn more academically appropriate registers, but her research focused on eight- to ten-year-old students taught by ESL language and content specialists. How do other teachers teach science language and content to other groups? How do the students in other classes talk science after studying the topic under investigation? How does the use of oral language features compare with that of the written features described by Veel or Mohan et al.? How does the use of oral language features compare between ESL and non-ESL students at different ages? Do different teachers use similar interaction strategies? These few studies raise many research questions, some of which the current study aims to respond to. 2.9 Summary This chapter has reviewed several areas of the literature which are relevant to this research. From the review above, it was revealed that except for the work of those in the area of systemic functional linguistics, the treatment of causal discourse has generally been centered on sentence-level lexical markers. Furthermore, although causal connectedness has been acknowledged as playing a critical role in text comprehension and learning, very little research has to date examined the development of this connectedness or explored how students learning the language of cause and effect in science classes construct their texts. Moreover, from a teaching perspective, a review of the literature from the science educators has suggested that the explicit development of causal discourse in science classes is not a focus for instruction, so it would seem that students must be acquiring it implicitly. Yet how implicit is this causal language teaching in mainstream classes? Is it more or less explicit in content-based language classes? The literature has suggested that in content-based language classes, there has been an effort to develop students' language ability through content teaching, particularly in classes which use Mohan's Knowledge Framework. Based on the literature, however, much of the explicit language teaching has been limited to sentence-level linguistic devices which help students reach a level where they can move into mainstream classes, but does not necessarily help them reach the same level of academic literacy that their native English-speaking peers are at. To reach this level of literacy, students need to 52 develop their ability to handle grammatical metaphor, a linguistic ability which research on native English-speaking individuals has shown distinguishes adults from children (Derewianka, 1995), but which has had very limited research from the second language acquisition field. How is grammatical metaphor being developed in ESL science classes and in mainstream science classes? How are the participants in these classes constructing causal explanations? Are causal explanations constructed by ESL students qualitatively different from those constructed by non-ESL students? The current study uses concordancing techniques and discourse analysis to examine how the participants in four different contexts develop causal explanations and the taxonomies from which the explanations d