Quam Georgetown 0076D 147

HELPING LANGUAGE LEARNERS ALIGN WITH READERS THROUGH NARRATIVE: INSIGHTS INTO THE BREADTH, TARGETS, AND EXPLICITNESS OF EVALUATION FROM APPRAISAL STUDIES OF SECOND LANGUAGE GERMAN WRITERS

A Dissertation submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in German

Justin Erle Quam, M.A.

Baltimore, MD

August 13, 2020

HELPING LANGUAGE LEARNERS ALIGN WITH READERS THROUGH NARRATIVE: INSIGHTS INTO THE BREADTH, TARGETS, AND EXPLICITNESS OF EVALUATION FROM APPRAISAL STUDIES OF SECOND LANGUAGE GERMAN WRITERS

Justin Erle Quam, M.A.

Thesis Advisor: Marianna V. Ryshina–Pankova, Ph.D

ABSTRACT

Language users never communicate in a vacuum. Successful meaning-making through language depends on an awareness of one’s presumed audience and the choice of linguistic tokens appropriate to that interaction. In their engagement with that presumed audience, literate language users are characterized by their ability to express individuality, offer explicit and implicit opinions, assert affiliation with their readers, persuade their readers, and reinforce or challenge socially valued concepts.

In Systemic Functional Linguistics (SFL), these uses of language fall under the interpersonal metafunction, which encompasses the goal of the action performed by the language user and their relationship with an intended audience (Eggins, 2004; Halliday, 1984). One key component of interpersonal meaning making is evaluative language, described within the context of SFL by the APPRAISAL system (Martin & White, 2005), which operates at the level of discourse semantics and facilitates analysis of evaluation patterns beyond the word or sentence level (Coffin, 2002).

Current L2 writing research on interpersonal meaning making is characterized by a lack of studies examining (a) genres outside of academic writing and (b) languages other than

English. To expand instructors’ awareness of (a) students’ command of interpersonal resources at various curricular levels and (b) instructional activities that can foster students’ continued

iii

development, this project explores the use of evaluative language in narrative texts written by L2 learners of German at three proficiency levels.

This project consisted of two rounds of data collection. In the first study, fairy tales written by a small group of students of varying proficiency levels were analyzed in detail; potential patterns in APPRAISAL were flagged for further analysis. In the second study, a larger set of texts were analyzed to assess whether patterns emerging from the first study achieved statistical significance. Results revealed trends in the use of APPRAISAL types across the various levels, with lower proficiency writers relying more heavily on APPRECIATION and narrator- centered evaluations, whereas higher proficiency writers drew more often on AFFECT as a vehicle for implicit evaluation and broadened the scope of their evaluations to include valued behaviors and traits. The thesis concludes with potential instructional activities and remarks on the applicability of the current APPRAISAL scheme to narrative analysis.

ACKNOWLEDGMENTS

I am deeply grateful to my advisor, Marianna Ryshina–Pankova. Even before I asked her to take on that role, she offered her time and insight whenever I found myself twisted around a concept or research problem. Without her encouragement, I would never have considered a Ph.D a real possibility; when my first year of teaching full time left me rudderless, she believed more strongly than I did that I would still finish this project. Большое спасибо за то, что ты меня поддерживала, вдохновляла и побуждала меня к развитию на протяжении всего периода моих исследований. Я счастлив, что ты была моим куратором, и я надеюсь на нашу совместную работу в будущем. I offer my deepest appreciation to my committee members, Joe Cunningham and Lourdes Ortega, as well as Mary Helen Dupree, who helped assess my oral comprehensive exams. A professorship entails a slew of duties one is expected to take on without additional recompense, which can become onerous at the best of times; for volunteering their time on my behalf during an ongoing global health crisis, I thank and honor them. I owe a debt of thanks to the faculty, staff, and graduate students of the entire GUGD, a learning community rich in curiosity, insight, and solidarity. I recognize Courtney Feldman, for remembering our birthdays; Anja Banchoff and Astrid Weigert, for their constructive criticism on my teaching and their support for my musical endeavors; Peter Pfeiffer, for his counsel during my job search and his well-chosen Goethe references; and Friederike Eigler, for her sympathy after an unexpected medical setback. I also learned a great deal from my fellow graduate workers in other departments, and I applaud the Georgetown Alliance for Graduate Employees for their tireless work to secure protections and stability for the entire graduate community. For their direct assistance in turning this dissertation from notions into pages, I am grateful to many extremely patient friends, students, and colleagues, including Julia Goetze and Michael O’Donnell, for their advice on coding methodology; James Fallows, for recommending the Scrivener word processing app; Raphael Kabo and Shane McGarry, for developing elegant workflows I used to manage my citations; Christopher Tabisz, for offering feedback on my study materials; Sandra Digruber, for coding even more of the data than I asked her to; the participants in my studies, who gave of their time and creativity, including Ann–Kathrin Koster, Julia Goetze (again), Helene, Jürgen, Lilly, Mara, and Max; Megan Brett and Charlotte Hagerman, for studying alongside me and keeping me on track; Jimmy Rothschild, for insightful comments on a previous paper; Emma Rackstraw, for making time on short notice for a final dry run; and Christina Castelli and Julia Davidson, for offering to proofread my manuscript, catching the typos I swore I had eliminated, and reminding me when I wasn’t making any sense. Had it not been for Waldsee, I doubt I would have learned German or discovered my love of languages. For setting me on that path, I am grateful to my first teachers: Xia, Jan, Monika, Ökorette, Gabi, Michi, Ilona, Dani; my mentors-turned-colleagues: Ellen, David, Detlef, Heidi, Julia, Resi, Steffi, Udo(?); Berndt and Karl; and far too many students and colleagues to name. (I tried to, and then I realized I could fill this thesis and still forget far too many. You know who you are.) I wrote the final chapters of this dissertation in relative isolation during the 2020 coronavirus pandemic. Maintaining my productivity (and my mental health) proved unexpectedly challenging. For the camaraderie they shared in our virtual game nights, film screenings, e-mail check-ins, and long chats over coffee, I am grateful to my quarantine

buddies—Bill Golba and Christina Tricou—and to Amanda Chang, Andrea Chapin, James Conway, Sarah Hulbert, Antje Krueger, Scott Lewis, Kellie Nance, Allison Parker, Kelly Rolfes–Haase, Emily Tucci, Ted Watson, Anke Woodsmall, and Stephen Wirth; Elmar, Lizzy, and Ian Schärfig; Em Skow, Brad Potter, and Zach Mark; and the authors Daniel Abraham, Tana French, and Max Gladstone, whose worlds of ink and whimsy allowed me to escape the four walls of my apartment even during a stay-at-home order. Throughout my high school and college years, I was fortunate to study with curious, kind, challenging teachers. For their generosity and willingness to nerd out, I am particularly grateful to Lissa Angus, Tom “Bob” Martinez, David Eberhard, R.J. Kasicki, Rebecca Fornelli Bredle, Leland Livingston, Jennifer Winokur Biegen, Bob McBride, Anders Winroth, Paul Kennedy, Paul Lagunes, Akhil Reed Amar, Vasili Byros, Stephen Latham, and the late Kent Youngren. For their collaborative spirit and welcoming arms, I am grateful to the faculty and staff of the German International School of Portland and the St. Paul’s Schools in Brooklandville, Maryland. Throughout my studies, I have taken solace in music. For their insightful direction and patient conducting, I am thankful for Max Blum–Campo, Sharon Corbin, Don Devany, Jeff Douma, Ross Heise, Keiji Ishiguri, Joanne May, Ben Olinsky, Joanna Powell, John Smedstad, Dileep Srihari, Cynthia Viise, and my fellow singers in the choirs of St. Paul’s (IL and MD), the Yale Glee Club, Out of the Blue, the Whiffs, the Capital Hearings, and the 18th Street Singers. I owe an incalculable, uncategorizable debt of thanks to: — Ezra Deutsch–Feldman and Agnes Mazur, for reminding me via political trivia questions that I have a safe harbor in D.C. (I swear, I will watch The Simpsons one of these days); — Justin Jee, for inspiring me by tireless, perceptive example; — Debra Keane, for honesty, greeting cards, and introducing me to Eddie Izzard; — Patrick Rothfuss, who wrote the two books that got me through my darkest hours; — Mike Rowan, best of bassy roommates, for never minding that my books were strewn over every possible surface; — Neil Weyhing, for constantly encouraging me to finish my dissertation (whenever he was not pursuing the more urgent goal of recommending a worthy sci-fi series); — The Waldsee 2020 Betreuerschaft, who sat through as many as three iterations of my defense, for reminding me of the rewards of looking inward; — Evi, Ilse, and Zenzi, my Tuesday night Märchenwald game night partners, for laughter in dark times; — Victor, without whom I never would have tried the most important new things; — Ross Kellan, who taught me how to teach by example; — My might-as-well-be family: Carole, Elise, Justin, Renate, Roger; — …and, first among all: Cassandra, Joel, Nate, Seth, and Vivian. Like Westley, they know there are many ways to show love and support — from invitations to play King of Tokyo and Wingspan, to Harry Potter references and Star Trek crew drafts, to Bovinity Divinity ice cream.

TABLE OF CONTENTS

CHAPTER 1. INTRODUCTION ...... 1 CHAPTER 2. LITERATURE REVIEW ...... 4 2.1 Nature and Function of Interpersonal Resources ...... 5 2.2 Trends in the Understanding of Interpersonal Resources ...... 9 2.3 Overview of Approaches to Interpersonal Resource Analysis ...... 11 2.4 Modality ...... 13 2.5 Stance: Perspectives from Corpus Linguistics ...... 17 2.6 Voice: Perspectives from Rhetoric and Composition ...... 21 2.7 Evaluation: Perspectives from Pragmatics and Discourse Analysis ...... 24 2.8 Systemic Functional Linguistics ...... 27 2.9 The Interpersonal Metafunction within Systemic Functional Linguistics ...... 30 2.10 APPRAISAL ...... 33 2.10.1 APPRAISAL Categories ...... 37 2.10.2 Benefits of the APPRAISAL Approach ...... 39 2.10.3 ATTITUDE ...... 40 2.10.4 GRADUATION ...... 54 2.10.5 ENGAGEMENT ...... 56 2.10.6 Explicit and Implicit ATTITUDE ...... 57 2.10.7 Elaborations to the APPRAISAL Framework ...... 59 2.10.8 Theoretical Cross-Pollination: Application of APPRAISAL to other Interpersonal Resource Frameworks...... 64 2.10.9 Limitations and Criticisms of the APPRAISAL Model ...... 65 2.11 Applications of Interpersonal Resource Frameworks ...... 67 2.11.1 Academic Writing ...... 67 2.11.2 Variations in APPRAISAL across Disciplinary Communities...... 69 2.11.3 Student Writing (L1) ...... 72 2.11.4 Second Language Writing...... 75 2.11.5 Narratives ...... 80 2.11.6 Pedagogical Applications...... 83 2.12 Limitations and Avenues for Further Study...... 86 2.13 Reasons for Undertaking this Project ...... 88 2.14 Affordances of Narratives ...... 90 2.15 Focus on Evaluation ...... 93 CHAPTER 3. METHODOLOGY ...... 96 3.1 Research Design...... 96 3.1.1 Curricular Context ...... 99 3.1.2 Study 1 (Exploratory) ...... 100 3.1.3 Study 2 (Cross-Sectional) ...... 101 3.2 Procedure ...... 102 3.2.1 Study 1 ...... 102 3.2.2 Study 2 ...... 104 3.3 Participants ...... 105 3.3.1 Study 1 ...... 105

vii

3.3.2 Study 2 ...... 106 3.4 Tasks ...... 109 3.4.1 Study 1 (Märchen) ...... 111 3.4.2 Study 2 (Friendship) ...... 112 3.5 Coding ...... 115 3.5.1 General Coding Principles ...... 119 3.5.2 Identifying Coding Units ...... 120 3.5.3 APPRAISAL Categories ...... 121 3.5.4 Resolving Coding Ambiguities ...... 125 3.5.5 Interrater Reliability ...... 136 3.6 Data Analysis ...... 137 3.7 Research Questions for Study 1 ...... 138 3.8 Research Questions for Study 2 ...... 138 CHAPTER 4. RESULTS ...... 139 4.1 Study 1 ...... 140 4.1.1 AFFECT among Study 1 Texts ...... 142 4.1.2 JUDGMENT among Study 1 Texts ...... 147 4.1.3 APPRECIATION among Study 1 Texts ...... 152 4.1.4 Explicitness in ATTITUDE among Study 1 Texts ...... 154 4.1.5 ATTITUDE within Generic Stages ...... 155 4.1.6 GRADUATION among Study 1 Texts ...... 160 4.1.7 Modifications for Study 2 ...... 166 4.2 Study 2 ...... 167 4.2.1 AFFECT among Study 2 Texts ...... 169 4.2.2 JUDGMENT among Study 2 Texts ...... 170 4.2.3 APPRECIATION among Study 2 Texts ...... 171 4.2.4 Inscribed vs. Evoked/Provoked APPRAISAL in Study 2 ...... 172 4.2.5 Polyphony in Study 2 ...... 173 4.2.6 ATTITUDE within Generic Stages ...... 175 4.2.7 GRADUATION among Study 2 Texts ...... 177 4.2.8 Lexical Realizations ...... 179 4.3 Summary ...... 182 CHAPTER 5. ANALYSIS...... 184 5.1 ATTITUDE Trends across Levels...... 184 5.1.1 AFFECT: Emergence of Multiple Perspectives in Study 1...... 185 5.1.2 APPRECIATION in Study 2: From Impact on Self to Value for Others ...... 191 5.1.3 JUDGMENT in Study 2: A Broadening of Targets ...... 198 5.2 ATTITUDE within Narrative Stages...... 211 5.2.1 APPRECIATION in Narrative Stages ...... 211 5.2.2 JUDGMENT in Narrative Stages ...... 215 5.2.3 AFFECT within Narrative Stages ...... 218 5.3 GRADUATION in Study 2 ...... 222 5.3.1 Targets of GRADUATION...... 224 5.3.2 Increased Use of FOCUS ...... 226 5.4 Invoked Evaluation ...... 229

viii

5.4.1 Provoked ATTITUDE in Learner Texts...... 230 5.4.2 Provoked ATTITUDE in L1 Texts...... 231 5.5 Polyphony in Study 2 ...... 234 5.5.1 Polyphony within Level I...... 235 5.5.2 Polyphony within Level II ...... 236 5.5.3 Polyphony within Level III ...... 238 5.5.4 Polyphony within L1 Texts ...... 240 5.6 Evaluative Prosody ...... 246 5.7 Summary ...... 252 CHAPTER 6. DISCUSSION AND CONCLUSION ...... 254 6.1 Summary of Research ...... 254 6.2 Potential Bias of APPRAISAL Categories ...... 256 6.2.1 Ambiguities in Coding APPRECIATION ...... 257 6.2.2 Ambiguities in Coding JUDGMENT ...... 257 6.2.3 Ambiguities in Coding Appraisal Blends ...... 259 6.3 Limitations ...... 262 6.3.1 Small Sample Size ...... 263 6.3.2 Unique Curricular Context ...... 264 6.4 Pedagogical Applications...... 266 6.4.1 Using AFFECT to Garner Sympathy ...... 272 6.4.2 JUDGMENT in Historical Narratives ...... 276 6.4.3 APPRECIATION and Assessment ...... 279 6.4.4 Implicit Evaluation...... 280 6.4.5 Insights Related to Assessment ...... 282 6.5 Future Research ...... 286 6.6 Final Thoughts ...... 288 APPENDIX A: STUDY PROMPTS ...... 290 Study 1 ...... 290 Study 2 ...... 293 APPENDIX B: MÄRCHEN TEXT ...... 295 REFERENCES ...... 299

LIST OF FIGURES

Figure 1. Interpersonal Systems...... 25 Figure 2. Metafunctions of Language and their Realizations in Social Context...... 28 Figure 3. Components of Tenor...... 31 Figure 4. The APPRAISAL System...... 38 Figure 5. Realizations of AFFECT...... 43 Figure 6. Appreciation Categories as Distinguished by Probe Questions...... 46 Figure 7. The APPRECIATION System...... 47 Figure 8. Resources for Expressing JUDGMENT...... 51 Figure 9. Triggers of APPRECIATION and JUDGMENT...... 54 Figure 10. The System Network of GRADUATION via Intensification as Elaborated by Hood...... 57 Figure 11. The GRADUATION System...... 58 Figure 12. Screenshot of a Text as Coded with UAM Corpus Tool 3...... 118 Figure 13. ATTITUDE Scheme for Study 1...... 125 Figure 14. ATTITUDE Scheme for Study 2...... 126 Figure 15. GRADUATION Scheme for Studies 1 and 2...... 127 Figure 16. Nested layers of APPRAISAL in Text xn1...... 134 Figure 17. Criteria for Distinguishing JUDGMENT and APPRECIATION...... 136 Figure 18. Triggers of Blended AFFECT...... 137 Figure 19. UAM CorpusTool Search Function...... 144 Figure 20. Potential Matrix for Instructional Activity on AFFECT...... 280 Figure 21. Potential Matrix for Instructional Activity on JUDGMENT...... 285

LIST OF TABLES

Table 1. Participants in Study ...... 108 Table 2. Participants in Study 2...... 109 Table 3. Frequency Counts of ATTITUDE Types across Curricular Levels in Study 1...... 145 Table 4. JUDGMENT among Study 1 Texts...... 152 Table 5. APPRECIATION in Study 1 Texts...... 156 Table 6. Explicitness in Study 1...... 159 Table 7. Raw Frequencies of ATTITUDE Tokens at each Narrative Stage...... 160 Table 8. GRADUATION among Study 1 Texts...... 165 Table 9. GRADUATION: Force in Study 1 Texts...... 166 Table 10. GRADUATION Targets in Study 1...... 166 Table 11. Means of Expressing GRADUATION: Force in Study 1 Texts...... 170 Table 12. Frequency Counts of ATTITUDE Types across Curricular Levels in Study 2...... 173 Table 13. Frequency of Polyphony among Study 2 Texts...... 178 Table 14. Types of Polyphony among Study 2 Texts...... 179 Table 15. ATTITUDE use within Narrative Stages in Study 2 Texts...... 181 Table 16. GRADUATION among Study 2 Texts...... 182 Table 17. GRADUATION: Force among Study 2 Texts...... 182 Table 18. Means of Realizing GRADUATION: Force in Study 2...... 183 Table 19. Realizations of GRADUATION: Force: Quantification in Study 1...... 184 Table 20. Lexical Realizations within Study 2 Texts...... 186 Table 21. Frequency Counts of AFFECT Types in Study 1...... 190 Table 22. Instances of AFFECT in Text ji1 (Level II)...... 191 Table 23. AFFECT Sources in Study 2...... 195 Table 24. Verbal and Nominal Realizations of APPRECIATION at Level II...... 199 Table 25. Frequency Counts of JUDGMENT Tokens at Level II...... 205 Table 26. Frequency Counts of JUDGMENT Tokens at Level III...... 211 Table 27. Instances of GRADUATION via Infusion in Levels II and III...... 228 Table 28. GRADUATION Targets in Study 2...... 229 Table 29. Probe Questions to Identify Reader Positioning...... 276

CHAPTER 1. INTRODUCTION

Language is a powerful tool, offering a means not only to provide information and construe reality as the language user sees it (e.g., Halliday & Matthiessen, 2004), but also to influence and persuade (e.g., Ryshina–Pankova, 2011; Thompson, 2013), to save face (Hinkel,

2005), to exert power (Fairclough, 2001), and to express truth claims (e.g., Barton, 1993; Chafe,

1986). One of the core performative functions of language, accessible even to learners at the beginning of their language journeys, is the deployment of interpersonal resources to take on conversational roles and position oneself within a larger discourse community. This function may involve the way second language (L2) writers express opinions, address the audience, and define themselves vis-à-vis their own culture or the culture of the L2.

Developing the ability to interact with others and to express a broad range of attitudes and feelings represents a crucial aspect of language learning, and mastering these interpersonal resources requires more than simply acquiring vocabulary and learning grammatical rules.

Successfully aligning oneself with one’s readers and using evaluative language also requires awareness of genre and shared cultural references. Writers1 must not only take account of the levels of formality, modality, and self-reference considered appropriate for the genre in which they are writing; they must also demonstrate awareness of the multifarious potential connotations of the attitudinal language they deploy. At the simplest level of analysis, even the same lexical item can have divergent meanings depending on context; hefty might be an insulting choice when describing a person’s physical appearance but have more positive connotations in the context of a generous donation. Effective use of the interpersonal function of language, then, does not

1 Since the focus of my dissertation is on student writing, I often refer to ‘writers’ for the sake of simplicity, though ‘speakers’ or ‘language users’ would also be appropriate.

consist only of construing dialogic roles (e.g., who is giving or demanding information or goods and services); it also represents a component of literacy (Gee, 1998). Language users who are literate in a particular discourse can use interpersonal resources to successfully engage with members of that discourse community, express their individuality in their writing, and achieve goals such as persuading their readers and reinforcing (or challenging) socially valued concepts through evaluative language. In the case of situations characterized by social and experiential distance, writers must use language to “reconstrue [the social process] interpretatively, re- creating the social context entirely through text” (Ryshina–Pankova, 2016, p. 56). Research on the way language learners develop and employ these interpersonal resources can contribute meaningfully to language pedagogy and curricular development by providing language teachers with a better understanding of students’ meaning-making abilities at various levels of proficiency.

This literature review surveys existing research on textual self-expression through interpersonal resources, which has explored numerous aspects of interpersonal meaning-making in English but has yet to make a thorough study of it in the context of foreign language (FL) instruction. Many constructs that aim to capture various aspects of this textual identity have been studied in detail; the concepts of voice (Matsuda, 2001; Matsuda & Tardy, 2007, 2008), metadiscourse (Hyland, 2005), stance (Gray & Biber, 2012, 2015), evaluation (Hunston &

Thompson, 2000), and APPRAISAL (Martin & White, 2005; White, 2015), for example, have received significant attention in recent decades.2 Such expression can take various forms, many of which can be conceptualized within the systemic functional linguistics (SFL) framework. In

2 It is customary when discussing the APPRAISAL framework to give the names of system networks in uppercase. Among other things, this allows us to distinguish between the ENGAGEMENT framework and engagement as defined by Hyland and others.

SFL, the interpersonal metafunction of language encompasses the goal of the speaker/writer

(e.g., demanding information, making a declarative statement, requesting goods) and the relationship (e.g., formal or informal, frequent or infrequent contact) between a text’s author and its intended audience (Eggins, 2004; Halliday & Matthiessen, 2004). Such interpersonal resources might include references to oneself, the making and evaluation of truth claims, and the appraisal of other writers or phenomena, as well as the GRADUATION resources employed to strengthen or soften that evaluation; we might also define interpersonal resources, as Isaac

(2012) has, as the elements of language through which “the social roles enacted by the writer/speaker in specific situations (. . .) her/his interaction with the reader/hearer, [and] the relations of power and solidarity that s/he establishes” are construed (p. 58).

Note: Since most of the data for this project are drawn from writing produced by L2

German learners, several lexicogrammatical errors are present in the examples presented. To avoid distraction, individual infelicities are not marked as such; except in rare cases, student intent was clear enough to translate into grammatical English. Spelling is as in the original student work.

CHAPTER 2. LITERATURE REVIEW

This review surveys the approaches to interpersonal resources that have developed from various linguistic perspectives and mention notable applications of these approaches to academic writing, pedagogy, media discourse analysis, and other fields. It also addresses the limitations of current research of interpersonal resources. The aspects of language use that play into interpersonal resources have often proven difficult to define, and the topic in general has not been fully explored in a number of important contexts, especially in languages other than

English. While some approaches to language study, particularly systemic functional linguistics

(SFL), certainly emphasize the importance of genre and interpersonal meaning making to language learning, studies conducted so far more often analyze L2 writers’ ability to organize their texts (see, e.g., Ryshina–Pankova (2006) on textual coherence and cohesion) and make abstract arguments. Furthermore, a significant number of the studies that investigate the use of interpersonal resources in student writing focus on the fields of English as a Second Language or

English for Academic Purposes and/or are limited to academic writing, thus belying the frequency and importance of using interpersonal resources for other languages and in other genres. The narrow focus on academic writing is particularly important to dispel. To an extent, the focus on academic texts is only logical: Since the negotiation of multiple viewpoints is an essential skill for academic writers to develop, academic genres do present a natural opportunity to examine the use of, for example, ENGAGEMENT resources (Hood, 2008). However, it may be fruitful to examine the use of interpersonal resources in other genres (e.g., narratives) in which those resources play a key meaning-making role.

2.1 Nature and Function of Interpersonal Resources

Scholars of language have long recognized that the function of language extends far beyond a straightforward means of 1) identifying elements of an external reality and 2) describing that reality to another party. Language users do not merely report on agreed-upon facts of the world; rather, they linguistically construe the reality they perceive by means of linguistic resources, taking advantage of the nature of language as a system in which a choice between many grammatically acceptable options has ramifications for the meaning of every utterance (Eggins, 2004; Halliday, 1993, 2002).

Halliday and other linguists who work within the paradigm of systemic functional linguistics (SFL) have conceptualized the meaning-making potential of language as including three metafunctions: the ideational, the interpersonal, and the textual. Each represents a semantic realization of, respectively, a text’s field (how the text indicates who is doing what to whom), tenor (how the text construes the relationship between speaker/listener or reader/writer), and mode (how the text fits together logically) (Halliday & Matthiessen, 2004). The systemic functional grammar (SFG) conceptualizes linguistic resources not based on formal syntactic categories, but on how they fulfill these three metafunctions. The interpersonal metafunction includes those elements of language that construe “the social roles enacted by the writer/speaker in specific situations (. . .) her/his interaction with the reader/hearer, [and] the relations of power and solidarity that s/he establishes” (2012, p. 58). In even the simplest statements, the author can realize interpersonal meanings at the same time that they comment on the external state of affairs. Regardless of an utterance’s ideational content, its realization carries interpersonal meanings that can comment on, modify, or instantiate relationships between the speaker/writer

and their intended audience; that is, we take on speech roles, such as giving or demanding information or goods and services, whenever we use language (Eggins, 2004).

This use of language to construe particular roles for participants in a conversation has been highlighted by linguists for decades, including those not explicitly identified with the SFL school. Lemke (1989), for example, grounded his discussion of mood and modality and the use of status markers in an exploration of the many potential actions speakers might take in deploying interpersonal resources: “Every clause also defines a putative interaction in a speech situation, a contribution to a dialogue or exchange, a specification of the stance of speakers and addressees toward one another and the thematic content of the clause” (p. 37). Language users select among these kinds of interactions and stances through the informed choices they make among lexicogrammatical options in their repertoire.

Just as speakers and writers can construe reality in a variety of ways through their choice of ideational resources (e.g., conceptualizing the state of having a migraine by assigning the person or the head as the focus of the clause; see Halliday [2002]), they can construe a variety of potential relationships with their audience by deploying interpersonal resources in different ways. Interpersonal meaning-making involves not only the establishment and maintenance of a relationship between author and recipient, but also the expression of evaluations and attitudes, which can foster or threaten alignment between participants in a linguistic exchange (Zappavigna

& Martin, 2018); in effect, interpersonal resources give language users “[t]he ability to evaluate one’s material and establish a connection with readers” (Hyland, 2012, p. 134) or “confirm solidarity with their [readers’] views” (Ngo & Unsworth, 2015, p. 1). This insight is a key underpinning of many of the frameworks described below, which recognize that interpersonal

linguistic resources construe not only conversational roles, but also socially relevant evaluations of what the writer takes to be good, useful, praiseworthy, and true:

We do not use language simply to organize action or to describe (or even create) events and their relations. Language is also a resource for the creation and maintenance of social relations and value systems. Every discourse voice, embodied in text, constructs a stance towards itself and other discourse voices. It evaluates, explicitly or implicitly, what it has to say and the relation of what it has to say to what others do say or may say. Its evaluative orientation includes, but is not limited to, certitude or truth value. It can define any value orientation toward what it says and/or toward what others say: appropriateness, usefulness, morality, pleasurability; all the forms of ‘rightness’ and ‘goodness.’ (Lemke, 1989, p. 39)

As Lemke (1989) and White (2008) have written, we partly owe this insight on the dialogic character of texts to Bakhtin (1981; 1986), who with Valentin Vološinov introduced the idea that any given text or utterance exists within a context of related texts to which it responds.

All writers draw on the ideas and style of voices they have ‘read into’ their own writing (Ivanič

& Simpson, 1992), and an understanding of the ways in which authors both explicitly and implicitly respond to previous utterances can shed further light on the means by which language users create their own meanings. This kind of borrowing can represent a crucial part of securing membership in a discourse community, through demonstrating understanding of, and respect for, that community’s central values and key thinkers (Barletta, Mizuno, & Moss, 2013).

The interpersonal resources of language also provide the opportunity to evaluate not only the value of some external thing, but also the reliability of claims about the world, “enabl[ing] us to establish a stance or orientation toward the probability of a presentational reality or toward the degree of inclination or obligation attaching to a proposal” (Lemke, 1992, p. 83). By taking a stance toward the truth of a proposal or potential reality, writers can establish a sense of solidarity with their audience; illuminating this process is the goal, in particular, of the

ENGAGEMENT system (Martin & White, 2005; White, 2015), which categorizes the ways in which writers can make room in their arguments for the assertions of others. Put more concisely,

“interpersonal language enables speakers to present their opinion about the world and value systems, to engage in dialogue and interact with others, and at the same time, to regulate meanings into different degrees” (Mirallas, 2017, p. 33). Construing social roles, evaluating things and ideas, making and countering arguments: these core building blocks of the interpersonal function provide the foundation for skills that are key to success for language users

— and language learners. By deploying interpersonal resources, writers can create a textual impression of their individuality (Isaac, 2012); endorse, qualify, or challenge an argument using their own and others’ ideas (Aull & Lancaster, 2014; Hyland, 2005c); and express solidarity with their presumed audience (Martin, 2004; Martin & White, 2005). Reference to this suite of skills can help frame productive inquiries into student writing development; asking how well students can make a nuanced argument, or connect with their readers, may offer more insight into their development than tracking their errors in adjective ending usage.

Inquiries along these avenues have already illuminated tantalizing trends; there is evidence, for example, that students’ use of interpersonal resources can index their writing development as they progress from a more personal to a more public voice, characterized by less subjective expressions of stance and an increase in interpersonal grammatical metaphor

(Ryshina–Pankova, 2011). If we define a language learner with advanced literacy in the L2 as someone who can appropriately use the meaning-making system of language as effectively as possible given a certain audience and a particular communicative goal, expressing textual presence through interpersonal resources represents a key stepping stone toward that advanced literacy. Tools advanced language users employ to perform this self-expression and forge connections with their audience has been conceptualized in various linguistic fields as modality

(e.g., Bybee & Fleischman, 1995; Coates, 1987; Stubbs, 1986), voice (e.g., Matsuda, 2001;

Matsuda & Tardy, 2007, 2008; Tardy, 2016), metadiscourse (e.g., Hyland, 2005a, 2005c, 2015), stance (Biber & Zhang, 2018; e.g., Gray & Biber, 2012, 2015), evaluation (Hunston &

Thompson, 2000; Thompson & Alba–Juez, 2014), and APPRAISAL (e.g., Martin & White, 2005).

2.2 Trends in the Understanding of Interpersonal Resources

Linguists have used a variety of approaches over the last thirty years to describe the ways language users engage in dialogue; construe roles for themselves and their interlocutors, evaluate other actors, processes, and propositions; take a stance toward the reliability of information; and align themselves with or against their listeners/readers. In the course of research in several linguistic disciplines, two shifts are evident: movement from a writer-based perspective to one that more explicitly acknowledges the reader, and increased attention given to the function of language resources as well as their form. In the 1980s and 1990s, linguists with an interest in the interaction of language and cognition analyzed the potential of the grammatical system of modality, in order to explore how language users express the idea of possibility, obligation, inclination, and necessity. (Writers who can successfully evoke these abstract concepts can more effectively make nuanced, persuasive arguments.) Studies in this vein tended to focus heavily on formal linguistic categories, as well as on the internal world of the writer, and they typically employed formal modal categories. The cognitive linguist Wallace Chafe (1986), in an influential summary of how “attitudes toward knowledge” (p. 271) are expressed in written

English, confined himself to a description of how, grammatically, writers can indicate degrees of reliability of pieces of information, express subjective belief, and demonstrate logical moves via induction, hearsay, or deduction. Joan Bybee, an expert in morphology and phonology with an interest in psycholinguistics, collaborated with Suzanne Fleischman, a scholar of pragmatics, to further explicate the grammatical tools writers can use to discuss either their graded belief—or

lack thereof—in certain propositions (referred to as epistemic modality) or the obligations of certain actors (deontic modality) (Bybee & Fleischman, 1995). These studies remain writer- centric; in postulating an array of tools writers can use, they seem to presume a given external reality being explained, evaluated, and commented on by the writer. Jay Lemke, a semiotician, was one of the earliest scholars to summarize the move away from this individual-centered view, in which “texts are the products of authors or speakers, who address themselves in an immediate context of situation” and “[l]inguistic resources are then seen as having a primarily communicative function in the sense of the exchange of messages or meanings” (Lemke, 1992, p. 85). In opposition to this view, Lemke asserts that interpersonal meaning-making is dependent on elements external to the writer:

In essence, it is not simply the expression of the attitude of the speaker, it is the construction of the world of social diversity through the text. What matters, then, is not that this speaker ‘has’ this attitude, so much as that the text has meaning in a community where there are specific, divergent attitudes possible, that the text is constructed within that universe of attitudes and also helps in turn to construct it. (p. 85)

Lemke’s insight opens up a new way to interpret writers’ choices in their use of interpersonal resources, allowing researchers to investigate not only how writers construct arguments and express evaluations, but what effect that evaluation is meant to have on the reader and within the discourse community. This understanding of co-construction of meaning is also central to SFL. While not himself a systemic functional linguist, Lemke proposed a model of multimodal social meaning-making through language (which features ‘presentational,’

‘orientational,’ and ‘organizational’) that overlaps with Halliday’s metafunction model (Hong,

2012), and systemic functional linguists share his recognition that written texts do not exist in a vacuum, but reflect an awareness of other writers; they respond both to previous texts and to the predicted response on the part of the reader. The trend toward viewing writing as a collaborative

process, which involves a “de-centering of language and text in favor of a stronger focus on the social patterns of activity revolving around the interpretation and (re)production of texts”

(Lancaster, 2013, p. 202), has expanded to disciplines beyond SFL. In applying the insights of

APPRAISAL to the field of voice studies, for example, Isaac (2012) reports approvingly that

‘voice’ is being conceptualized not merely as an author’s self-presentation; instead, “greater attention is beginning to be paid to the intersubjective or dialogical dimension and the writer’s finely-tuned interaction with the imagined reader” (p. 13). Within the fields of pragmatics and discourse analysis, the view of how language users express evaluation has also shifted from a primarily self-expression-oriented process to one that is “essentially socially constructed and designed primarily, or at least as a major part of its function, to establish solidarity with the addressee” (Thompson & Alba–Juez, 2014, p. 4).

2.3 Overview of Approaches to Interpersonal Resource Analysis

Numerous models have been proposed to account for the expressive potential of interpersonal resources. Through the 1980s, these models typically involved elaborations on the grammatical system of modality, which has to do with expressions of necessity, possibility, certainty, and obligation as construed by the speaker (Boye, 2012). As sketched out above, early research on modality from within the fields of cognitive linguistics and morphology typically focused on grammatical form rather than function and on author rather than audience; it also tended to use words and clauses as the level of analysis, rather than tracing an evaluative thread throughout an entire paragraph, stage, or text. Approaches to modeling interpersonal resources from within the field of second language (L2) writing, such as the taxonomy of stance, engagement, and metadiscourse (Hyland, 2001, 2005a, 2005c, 2010, 2015) developed by English for Academic Purposes (EAP) scholar Ken Hyland, include more explicit attention to the reader

and to the way authors identify their relationship with their readers by reference to commonly held propositions. Hyland’s approach responds to a copious amount of work on the topic of stance and voice from within the field of L2 writing, which overlaps with both our understanding of interpersonal resources and with the multifaceted topic of writer identity/identities (Coffin,

2002; Elbow, 1994; Hirvela & Belcher, 2001; Ivanič & Camps, 2001; Matsuda, 2001, 2015)

Work in the fields of discourse analysis and pragmatics also devotes attention to the co- construction of meanings between writer and reader. The evaluation framework proposed and elaborated by Thompson and his colleagues (Hood, 2008; Hunston & Thompson, 2000;

Thompson, 2001; Thompson & Alba–Juez, 2014; Thompson & Thetela, 1995) focuses not only on how authors assess the veracity or likelihood of propositions, but also how they present their own subjective attitudes and feelings toward the objects of their utterances. Thompson’s work is particularly noteworthy in that it widens the lens of interpersonal resource analysis from the word- and clause-level to encompass patterns of evaluation throughout a text. Recent decades have also seen the development of analytical frameworks within SFL that conceptualize stance, engagement, alignment, and evaluation with reference to the functional categories established by

Halliday. These include the APPRAISAL system first formulated by Martin and White (2005), which includes the subsystems of ATTITUDE (covering much of the same semantic terrain as

Thompson’s evaluation framework) and ENGAGEMENT (which corresponds to aspects of stance and Hyland’s engagement).

Despite their different disciplinary lenses, there is considerable overlap between the phenomena reviewed above (as studied by Hyland, Thompson, and others within the disciplines of pragmatics, L2 writing, corpus linguistics, and rhetoric) and the SFL contributions to interpersonal resources made by Martin and White and others working within APPRAISAL. This

overlap can be seen most strikingly in a synthesis by Isaac (2012), who has argued that voice (as conceptualized within the rhetorical tradition) can profitably be described and defined using an

APPRAISAL-based framework. This and other SFL-based approaches are able to take advantage of the central understandings of APPRAISAL, which may manifest only ad hoc in other theoretical lenses: (a) an emphasis on function (not merely form), (b) the importance of co-construction of meaning with the reader, and (c) the need to examine text-level (not merely word- or sentence- level) meaning making. Indeed, APPRAISAL theory has demonstrated that it can be applied to a wide range of contexts to illuminate meaning making through use of interpersonal resources, including academic writing (Lee, 2007; Swain, 2007), discipline-specific writing (Coffin, 2002), student writing at a variety of grade levels (Derewianka, 2007; Lancaster, 2014; Lee, 2015; Mei,

2007), and L2 writing (Liu, 2013; Ngo & Unsworth, 2015).

The APPRAISAL system subsumes several functions that have a long history in the linguistics literature, such as modality, voice/stance, and evaluation; in the following sections, I explore studies that have contributed to linguists’ understanding of each construct.

2.4 Modality

Modality is a facet of grammar that involves a writer’s projection of their individual viewpoint. It is typically divided into two groupings, epistemic and deontic modality, based on whether the utterance in question concerns subjective beliefs or expressions of obligation and inclination. Epistemic modality involves the subjective experience and beliefs of the writer, who may use epistemic modality to indicate their own certainty or lack thereof regarding an aspect of the world. Early descriptions of modality in the fields of linguistics and corpus linguistics focused heavily on formal grammatical categories—e.g., auxiliary verbs, modal adverbs—that typically encode modality-related meanings. Sociolinguist Jennifer Coates (1987) defines

epistemic modality as “concerned with the speaker’s assumptions, or assessment of possibilities, and, in most cases, (. . .) the speaker’s confidence or lack of confidence in the truth of the proposition expressed” (p. 112). In the context of corpus linguistics, Michael Stubbs (1986) adopted an approach to the ways in which speakers use epistemic modality that draws on ideas from speech act theory (Searle, 1965, 1969) and categorized modality according to the speech roles it allows speakers to enact, “to express personal beliefs and adopt positions, to express agreement and disagreement with others, to make personal and social allegiances, contracts, and commitments, or alternatively to disassociate the speaker from points of view, and to remain vague or uncommitted” (p. 1). Stubbs’s project drew additionally on many older ideas in this area from sociolinguists such as Levinson and Goffman, but he argued that he was the first to conduct a corpus-based linguistic analysis of modality in English, and he made the rather sweeping claim that what he calls commitment and detachment—that is, a writer’s willingness to endorse an idea or truth claim—can serve as an organizing principle for language. Much of what

Stubbs describes as propositional commitment would later be systematized by SFL linguists within the ENGAGEMENT framework. Stubbs’s typology, however, still lies within the field of speech act theory; he categorizes sentences according to whether they represent promises, utterances, invitations, or other kinds of speech acts. It is notable here that, while Stubbs’s approach is functional, interested in how writers do things in the world, it remains focused on the writer and generally ignores the audience except as a passive recipient of invitations, offers, and commands. In contrast, Coates (1987) more explicitly identifies the addressee as an important factor in the use of modality; she points out that making truth claim can have other functions besides simply evaluating the merits of an idea. A speaker may, for example, be interested in constructing solidarity with their audience, in avoiding the giving of offense, or in helping the

addressee to save face; these considerations might argue for the use of difference choices to realize a more tentative stance. These studies of modality tend to offer taxonomies of the various grammatical forms that can be used when employing modality.

Later accounts of modality from grammaticalization theory and usage-based scholars, however, recognize that the meaning of a modality-inflected utterance and the particular modal verb or adjunct chosen may not overlap as closely as an analyst may wish. As Bybee and

Fleischman (1995) point out, there is no one-to-one correspondence between grammatical form and modal function; modal verbs, for example, may express either deontic or epistemic modality, as in Examples (1) through (3), drawn from Harry Potter and the Deathly Hallows (Rowling,

2007):

(1) “Scrimgeour must be surrounded by our people before I act.” (deontic)

(2) “If you have struck any kind of bargain with Griphook, and most particularly if that bargain involves treasure, you must be exceptionally careful.” (deontic)

(3) “There must be a secret message from Dumbledore hidden in the icing!” (epistemic)

In Example (1) above, must notes a necessary precondition before proceeding with an action; in Example (2), must implies an obligation on the listener, and in Example (3), must expresses the speaker’s firm belief about the state of affairs. The difference is not in the modal form but in the context of the conversation. Bybee and Fleischmann (1995) therefore argue that modality can more productively be differentiated based on the role of the person or persons implicated in the modal usage. They prefer to speak of agent-oriented modality, as it refers to the conditions under which an agent is operating—whether the actor under discussion must do, wishes to do, can do, or may do something. Epistemic modality, on the other hand, is oriented toward the speaker and refers to their beliefs about the world. Within systemic functional linguistics, modality is subdivided into modalization and modulation, which maps essentially

onto the distinction between epistemic and deontic modality (Eggins, 2004) but is concerned to a greater extent with the aim of a speaker in making an epistemic or deontic claim. Recent studies of epistemic modality within SFL have followed the trend toward a stronger emphasis on dialogue between two participants in a conversation. Rather than assuming a worldview in which language users were evaluating an external reality common to all observers, systemic functional linguists such as White (2008) recognize that epistemic modality does not, necessarily, have only to do with the truth or falsity of propositions. In the case of propositions whose veracity cannot be ascertained, as in Example (4), the use of epistemic modality is better understood as a means of positioning the writer:

(4) “Well, the sword, Harry! Dumbledore must have known you’d want to go back there, and I mean, Godric’s Hollow is Godric Gryffindor’s birthplace…” (Rowling, 2007)

In this case, the speaker (Hermione Granger) is attempting to persuade the listener (Harry

Potter) of two points: that they should visit the town of Godric’s Hollow, and that they should trust the judgment of the deceased character Dumbledore; she is using the modal verb must here in order to align herself with Dumbledore’s perspective. This understanding of modality implies the existence of many possible alternatives to the proposition being evaluated by the modal construction; epistemic modals “construe the value position currently being referenced in the text as but one of a number of actual or possible dialogic alternatives and thereby (. . .) make space for these alternatives in any ongoing dialogic exchange” (White, 2008, p. 149). The construct of modality has proven fruitful for applications by researchers in academic and argumentative writing, in both L1 and FL contexts. As Gruber (2004) notes, scholars such as Hyland have used modality to differentiate between academic discourses, demonstrating that the usage of each type of modality tends to vary according to discipline. Epistemic modality in particular seems to be difficult for L2 writers to master and may therefore provide a lens through which to evaluate

second language writers in terms of their overall writing proficiency in the L2. Hu and Li (2015), for example, found in an analysis of a corpus of English essays written by both college-level L1

English speakers (200 participants) and Chinese learners (2600 participants representing 10

Asian countries) that the less proficient L2 writers in the sample (who were classified as level A2 of the CEFR (Common European Framework of Reference) based on their TOEFL and TOEIC scores) used fewer and syntactically simpler devices to express epistemic modality, and that L2 writers at all proficiency levels in that study used fewer “booster” devices (see the following section on Hyland) than L1 users.

2.5 Stance: Perspectives from Corpus Linguistics

Corpus linguists such as Biber, Gray, and Hyland have described in great detail the typical grammatical realizations of interpersonal resources, in particular through analysis of the construct of stance, which involves an author’s subjective positioning in relation to other propositions and ideas. Hyland (2012) defines stance as follows, in contrast to the related topic of voice: “[S]tance largely involves the writer’s expression of personal attitudes and assessments of the status of knowledge in a text (. . .) voice, on the other hand, acknowledges the authorized ways of speaking as a community member” (p. 134). Some corpus analyses of stance in recent years have followed a similar trend to that of other work on interpersonal resources. While research in this disciplinary tradition focuses primarily on word- and clause-level markers of evaluation (as one would expect from corpus linguistics), the work of Hyland in particular demonstrates a recognition of the importance of the audience in making meanings jointly with an author, as well as a reference to the context of culture in which an author indicates their stance.

Stance has been defined most recently by Douglas Biber, a leading corpus linguist in the field of stance studies, as “lexico-grammatical devices that function to express an epistemic or

attitudinal meaning” (Biber & Zhang, 2018, p. 98)— that is, markers that encode a writer’s levels of commitment toward a proposition. These markers, which might include adverbials, modals, complement clauses, and evaluative that-constructions (Biber et al., 1999), have been laid out in detailed taxonomies based on groupings with shared lexicogrammatical characteristics. For example, stance is often conveyed via lexis but may also be realized in grammatical stance devices such as complement clauses (that-clauses; e.g., I’m glad that…, it’s possible that…), adverbials (in fact…), modal verbs, and noun + preposition phrases (e.g., the importance of…; Gray & Biber, 2015). In recent research on stance, Biber and Zhang (2018) expanded the scope of their research to include evaluative lexis that positions the writer without the use of explicit grammatical stance markers. Their 2018 study compared two corpora of evaluative texts, one of which (Opinionated, or OP) was characterized by a high frequency of grammatical stance terms, while the other (Informational–Persuasion, or IP) did not boast nearly as many explicit stance terms despite (according to holistic reader judgments) including highly persuasive texts. The authors performed a keyword analysis to identify words in the IP subcorpus that were especially prevalent in comparison with the OP subcorpus. They group the keywords yielded by their analysis as (a) adjectives evaluating other objects, (b) adverbs indicating evaluative manner, and (c) nouns or verbs indicating an indirect value judgment.

These categories bring to mind markers of implicit evaluation found in the work of

Thompson and his colleagues or markers of APPRECIATION and JUDGMENT in the APPRAISAL system by Martin and White (2005); the distinction lies primarily in the corpus-based method

Biber and Zhang used to elicit these keywords. By expanding the scope of corpus-based inquiries into the realizations of interpersonal resources, this study qualifies an argument by Hunston that

the use of implicit evaluative lexis is generally “not amenable to generalisable quantitative corpus analysis” (Biber & Zhang, 2018, p. 112).

The work of Biber and his colleagues offers a concrete illustration of the varied tools a writer can use to index their perspective, but their analyses center around the meaning-making ability of the writer, locating much of the agency with the person of the stance-taker (though this is less pronounced in Biber & Zhang [2018]). Isaac (2012) in particular has criticized this orientation for “foreground[ing] the expressive function of voice over the conative function of relating to and positioning the imagined reader” (p. 53). This limitation, which seems inherent in corpus analysis research, has however been ameliorated in recent work by Hyland (2012), who emphasizes the importance of a broader interactional context against which a writer’s moves must be understood. Hyland focuses in particular on academic genres, in which writers are often expected to project their authorial presence and communicate their affective and epistemic commitments less directly; he argues that academic writers “use language to offer a credible representation of themselves and their work, and to acknowledge and negotiate social relations with readers” (2004, p. 133). Successful academic writers, therefore, need to be able to project an authoritative authorial persona. They do so, in Hyland’s model, by means of stance and engagement markers.

Stance markers (e.g., modal verbs, mental processes, adverbs expressing probability or usuality) indicate a writer’s degree of epistemic certainty (Hyland & Milton, 1997), her/his attitude or emotion toward a concept or argument, and textual presence (this may be encoded, for example, in personal pronouns). Engagement—not to be confused with ENGAGEMENT within the

APPRAISAL framework (Martin & White, 2005)—includes elements that acknowledge and include the assumed reader (e.g., personal pronouns and directives) (Hyland, 2001). Hyland

comes to see readers as decisive in crafting any text, since “writers shape their texts to the expectations of their audiences (. . .) [who] can always refute [their] claims” (Hyland, 2001, p.

549). Hyland emphasizes the need for interaction with an audience in the particular case of academic writing; interpersonal meaning-making in this context “entails projecting a credible authorial voice, making appropriate judgments, establishing their credentials, and opening space for alternative views and evaluations, while claiming solidarity with, responding to, and bringing in their potential readers” (Dueñas, 2012, p. 2). Key lexical markers within Hyland’s stance include hedges and boosters, which realize epistemic stance. By employing hedges, which can be realized lexically by the choice of verb (e.g., it seems…, the authors propose…, x might be …) or adverb (e.g., potentially, possibly, seemingly), writers distance themselves from a proposition they have introduced. Writers may use hedges to inject uncertainty and hesitation into their claims, reducing the assertiveness of their authorial voice; they can be conceptualized as immunization devices to protect writer from critique (Gruber, 2004). Boosters (e.g., adverbs such as surely, definitely, very) perform the converse function, indicating writers’ confidence in their claims. In his work Hyland also uses the overarching concept of ‘metadiscourse’ to illustrate how academic writers comment on their text, using a degree of distance to refer “to the evolving text and to the writer and imagined reader of that text. (…) offer[ing] a way of understanding the interpersonal resources writers use to present propositional material and therefore a means of uncovering something of the rhetorical and social distinctiveness of disciplinary communities”

(Hyland & Tse, 2004, p. 156). Metadiscourse includes any elements that indicate the subjectivity of a text — hedges and boosters and other stance elements, for example, as well as

lexicogrammar that explicitly addresses the reader, such as imperatives and second person pronouns (Hyland, 2010, 2015).

Metadiscourse may also include punctuation or overt references to the organization or layout of the text. More recent corpus analyses by the educational linguists Laura Aull and Zak

Lancaster draw on corpus methodology to examine stance in the context of undergraduate writing in particular; their work provides examples of how research into stance can inform teaching and assessment practice. In their analysis of first-year student essays at the undergraduate level, Aull and Lancaster (2014) found stark differences between more and less advanced writers in their use of hedges and boosters, code glosses, and adversative/contrast connectors, and also reported that less experienced writers also tend to overuse certain stance markers. Aull and Lancaster (2014) argue that these results, taken together, suggest that as writers gain experience in academic discourses, their writing better reflects a greater “awareness of others’ views,” which is characterized by stance markers that “help construct a sufficiently honed and cautious stance in a community of many views” (p. 173). This conclusion reflects

Aull and Lancaster’s own multifaceted conceptualization of stance, which they understand as including not only an author’s subjective position toward propositions, but also the closeness of his/her relationship with the reader, and his/her self-positioning within the discourse community.

Less experienced writers might have the tool to express their stance in the first sense—their presentation of the truth—while more experienced authors can write with nuance that captures elements of the second and third senses as well.

2.6 Voice: Perspectives from Rhetoric and Composition

Within the disciplines of composition, rhetoric, and second language writing, the topic of interpersonal meaning-making is often examined through the lens of a writer’s ‘voice,’ which

can be coarsely described as that quality of writing that evokes the presence, attitudes, and beliefs of the individual writer; the metaphorical presence of the author in a text; or the textual impression of an author’s individuality (Isaac, 2012). Originally described within composition studies as a holistic construct representing the presence of a distinct author in the text (see

Matsuda, 2015; Yoon, 2017 for surveys of the definition of voice), which could include any characteristics of writing that “capture[s] the sound of the individual on the page,” (Elbow, 1981, p. 287) or allow for “feeling-hearing-sensing a person behind the written words” (Bowden, 1999, pp. 97–98, cited in Hirvela & Belcher (2001)), research on voice has expanded to include word- and sentence-level indicators that contribute to a unified authorial presence. More recently,

Hyland (2001, 2005b, 2012) has examined voice from the perspective of English for Academic

Purposes (EAP) by identifying particular words or phrases that invoke a writer’s presence; within the discipline of English as a Foreign Language (EFL), Danbin (2015) aimed to illuminate voice strength via a comparison of essay quality in relation to the frequencies of personal pronouns used by 70 English majors at a Shanghai university.

Though the concept of voice has proven difficult to define (Jeffery, 2011; Stock & Eik–

Nes, 2016), and some outspoken scholars within language acquisition and academic writing have argued that it cannot usefully be defined with reference to specific linguistic and rhetorical devices (Helms–Park & Stapleton, 2003; Stapleton, 2002; Stapleton & Helms–Park, 2008), it would seem to include interpersonal resources such as references to oneself, statements of the beliefs one holds, and moves to distinguish oneself from the audience one addresses. One commonality in definitions of voice is that it is a prosodic quality built up throughout the entirety of a text; in an otherwise vague definition in a canonical publication on voice in L2 writing,

Matsuda (2001) foregrounds its cumulative nature: “Voice is the amalgamative effect of the use

of discursive and non-discursive features that language users choose, deliberately or otherwise, from socially available yet ever-changing repertoires” (p. 40; emphasis mine). This reference to

‘socially available repertoires’ alludes to a trend in the conceptualization of voice over the last three decades in the field of L2 writing, in which descriptions of voice have expanded to recognize the social embeddedness of writing and the importance of the reader’s role in defining authorial voice (Belcher, 1997; Hanauer, 2015; Hirvela & Belcher, 2001; Ivanič & Camps, 2001;

Matsuda, 2001; Matsuda & Tardy, 2007). Matsuda (2001), for example, argues that it is often the audience who reads voice ‘into’ a text in collaboration with the writer. Seen through this joint- construction lens, elements of voice include those strategies writers use to take “social positions

(. . .) as they interact with others within communities by following certain rules and conventions, while at the same time exercising their agency through their individual linguistic choices”

(Flowerdew & Wang, 2015, p. 83).

From the disciplinary perspective of social semiotics and functional linguistics, then, voice can be understood as in effect a construct of negative space, defined in relationship to the constellation of other voices known and accessible to the reader and writer (Lemke, 1989). This shift—from a cognitive, inwardly-focused conceptualization to a more social understanding of voice—gives greater prominence to the related models of stance (reviewed above) and

ENGAGEMENT, the latter an SFL construct which involves the writer’s ability to comment on the truth value of propositions and thereby position her/himself within a community of other voices and comment on the epistemic status of the claims they make (see, e.g., Martin & White, 2005).

Like the research on modality discussed above, research on voice in the context of L2 writing often treats its subject as an amalgamative construct, viewing one author’s voice as something contingent on other’s, a characteristic that incorporates identities belonging to both

the writer and (presumed) reader. In a dissertation which aims to reconceptualize voice within the framework of SFL, Isaac (2012) offers an expanded definition of voice as “the discursive and non-discursive means (. . .) by which a writer construes their presence in a text and their social relations with the reader” (p. 42), arguing implicitly that the expression of textual identity necessarily involves positing a reader with whom the author can align or disalign herself.

2.7 Evaluation: Perspectives from Pragmatics and Discourse Analysis

Interaction with readers can also involve an appeal to shared understandings of the world

— shared assessments of people, behavior, and phenomena. Authors’ subjective assessment of the reality that they construe has been referred to within discourse analysis as evaluation.

Particularly influential research on this topic includes research by scholars who also work within functional linguistics (Thompson), media discourse (Thetela), corpus linguistics (Hunston), and pragmatics (Alba–Juez). Perhaps the most prominent model of evaluation flows from work by

Thompson & Thetela (1995), who emphasize the use of interpersonal resources to construe a

‘reader-in-the-text.’ In work on writer–reader relations on which later evaluation studies drew heavily, they used advertisements to examine the role that grammatical tools such as questions, commands, naming, and referential switches play in signaling the presence of writers- and readers-in-the-text. For them, one key distinction in writer–reader interaction is the deployment of what they call interactive and interactional resources, two types of moves by which authors engage with their presumed audience. Interactive resources presume the reader’s expectations and knowledge and using logical relationships to guide readers through an argument; Thompson and Thetela call these the “‘reader-friendly’ aspects” (p. 104) of the text. Interactional resources, in turn, include markers that evaluate arguments and persuade the reader (or rather reader-in-the- text) to adopt them. This might involve assigning speech roles by means of an interrogative

mood choice, thereby casting the reader in the form of questioner interested in finding an answer

(Thompson, 2001).

Thompson and Thetela further situate evaluation within a system of interpersonal resources (see Figure 1) that include the personal (consisting of modality and evaluation) and interactional (involving speech roles, which are primarily established within the MOOD element of a clause, and projected roles, which are established by the labels ascribed to participants within the text). Projected roles may be instituted, for example, using personal pronouns, implicit addresses to third persons (e.g., all passengers must now fasten their seatbelts), and vocatives.

This early work by Thompson focuses more explicitly on the distinction between what might, in the context of narratives, be called writer–reader and character–character relations — that is, expressions that explicitly involve the reader “without directly setting up interactional expectations” as opposed to moves that position the writer’s argument among other possible points of view, without overtly addressing the audience (Thompson & Thetela, 1995, p. 106).

Figure 1. Interpersonal Systems. (Thompson & Thetela, 1995).

Hunston and Thompson (2000) identify four parameters (good/bad; certain/uncertain; expected/unexpected; important/irrelevant) along which authors can convey their impressions of other agents, things, and processes. Notably, Hunston and Thompson’s framework focuses first

on the meaning conveyed by tokens of evaluation, and not on formal grammatical categories. As might be expected from a model oriented strongly toward meaning making, studies using this framework often have an explicitly pedagogical outlook. Effective expression of evaluation,

Thompson (2001) argues, requires control of writing at multiple levels, going beyond accurate use of a particular grammatical structure. In his 2001 study of interactive and interactional resources, for example, Thompson uses extracts from student writing, as well as reflections from tutoring sessions conducted by the author, to examine the difficulty students often have with deploying resources of evaluation. He argues that explicitly discussing the source of a proposition with novice writers can help them better understand how to communicate with their construed audience. He also recommends this approach for ESL students writing in an academic context, since conventions in academic writing (or in any specialized discourse community) vary from language to language and culture to culture. Thompson & Alba–Juez (2014) have recently revised and extended the original definition of evaluation to emphasize its status as a phenomenon that goes beyond a single linguistic stratum (e.g., only the semantic level or only the lexicogrammatical level):

We now see and thus define evaluation as a dynamical sub-system of language, permeating all linguistic levels and involving the expression of the speaker’s or writer’s attitude or stance towards, viewpoint on, or feelings about the entities or propositions that s/he is talking about, which entails relational work including the (possible and prototypically expected and subsequent) response of the hearer or (potential) audience. This relational work is generally related to the speaker’s and/or the hearer’s personal, group, or cultural set of values. (p. 13)

Applications of Thompson’s evaluation framework often take non-academic genres (e.g., narratives, news media, political discourse) as their focus, aiming to illuminate the means by which authors in politics or journalism can persuade their readers or elicit an affective response to their reporting (Bednarek, 2006).

2.8 Systemic Functional Linguistics

The frameworks sketched out so far generally refer to word- and sentence-level illustrations of interpersonal meaning-making (Mohan & Slater, 2004). In contrast, SFL, the model of language proposed by Michael Halliday and later elaborated by Ruqaiya Hasan,

Christian Matthiessen, Jim Martin, and other linguists of the Sydney school, sees “the complete text, a meaningful stretch of language, as the object of language inquiry” (Gleason, 2014, p.

467). In SFL, language is seen as having three ‘metafunctions’ that a language user employs simultaneously in the course of constructing a text (see Figure 2 below for a visualization of these functions and their realizations).3 The ideational metafunction of language allows a language user to construe reality in terms of the participants and processes occurring and acting upon each other; through the ideational metafunction, a writer expresses their view of who is doing what to whom. The textual metafunction offers language users tools to organize their ideas, lending logical or thematic coherence to a longer string of utterances. And it is the interpersonal metafunction that allows language users to construe their relationships with their audience or interlocutor, either through indicating interpersonal roles, positioning themselves in relation to other voices, or appealing to shared understandings and values (Eggins, 2004;

Halliday & Matthiessen, 2004). Any text or utterance can be analyzed in terms of its meaning- making potential for each of these three metafunctions.

The distinguishing features of SFL can be traced back to its name. First, it sees language as a systemic phenomenon, one in which language users choose between systems of options in order to make meanings. SFL is premised, in other words, on the importance of choice; when a

3 “Text,” in SFL, refers to “any unified piece of language” characterized by the interplay of coherence—“the text’s relationship to its extra-textual context”—and cohesion—“the way the elements within a text bind it together as ‘a unified whole’” (p. 24).

language user chooses to use a particular word, phrase, or linguistic resource, they are simultaneously choosing not to use all of the other options that were available at that juncture

(Barletta, Mizuno, & Moss, 2013; Halliday, 1984).

Figure 2. Metafunctions of Language and their Realizations in Social Context. (Martin, 2007).

An implication of this outlook is that any choice made in the context of potential other choices is a meaningful one, since a language user with knowledge of their alternatives could not avoid making a choice of some kind. This is an insight shared with behavioral economists: There are many situations in which one cannot avoid making choices, situations in which there is no

‘neutral’ option. When a grocery store manager is deciding whether to stack grapefruits, wine, or toilet paper nearest the front entrance, for example, they must fill that space with something, and that choice may have an impact on overall customer behavior (Thaler & Sunstein, 2003). Choice in language can be similarly unavoidable; when directly addressing someone, a speaker of

Korean must choose among an array of personal pronouns, each of which implies a particular kind of interpersonal relationship (Kim & Brown, 2014); choosing one of these pronouns means not choosing any of the others, and avoiding that choice is hardly an option. Similarly, a German

speaker will find it difficult to avoid addressing an interlocutor as either du or Sie, thereby construing their relationship with that interlocutor as more or less formal. Put another way, making interpersonal meanings is not optional (Lemke, 1989).

Returning to the name of the framework: SFL is also a functional theory of language, relying on meaning-making function (rather than formal grammatical categories such as nouns and verbs) as its organizing principle. Crucially, these include the social functions of language.

SFL theorists see language as a social semiotic, and the SFL model of language “is centrally concerned with showing how the organization of language is related to its social use and implies a dialectical relationship between a particular discursive event and the context in which it occurs” (Coffin, 1998, p. 5). This model reflects the understanding, as laid out in the introduction, that language does not merely reflect reality, but “enacts and construes the practices and values of the social context” (Isaac, 2012, p. 27).

A central insight of the SFL approach to linguistic analysis is that it offers a set of key questions to ask when studying language. Just as biologists can help understand the sometimes- counterintuitive process of evolution by continually asking themselves ‘who benefits?’ from a seemingly inscrutable change in an animal’s appearance or behavior (Dennett, 1995), linguists can demystify language by describing not just its structure, but its employment by speakers and writers to achieve particular goals — and how those uses are realized by the systems of discourse semantics, lexicogrammar, and phonetics/phonology. SFL has become the framework of choice for many linguists interested in curriculum development (e.g., Byrnes, 2002, 2006) or assessment

(e.g., Gleason, 2014).

For researchers interested in the interpersonal resources of language in particular, SFL offers a number of advantages. One major contribution of the overarching SFL approach is that it

emphasizes the co-construction of meaning with the reader and examines the buildup of meanings at the textual level (Gales, 2011; Poynton, 1989). An SFL analysis can therefore highlight trends that other approaches might miss (Rothery & Stenglin, 2000), and it is a natural fit for research that requires multiple levels of analysis; as Macken–Horarik et al. (2018) put it, an SFL lens allows linguists to examine not only first-order, but also second-order meanings — that is, not only what patterns of choice exist in a text, but also what those patterns of choice might allow an author to accomplish.

2.9 The Interpersonal Metafunction within Systemic Functional Linguistics

The interpersonal metafunction provides a lens to examine how language users not only establish conversational roles, but also make socially relevant evaluations of what the writer takes to be good, useful, praiseworthy, and true. Halliday (1984) initially divided interpersonal meaning making according to the nature of the interaction (e.g., demanding or proffering) and its object (e.g., information or goods and services), though his research eventually shifted from this emphasis on the nature of exchange to include the negotiation of feelings. These interpersonal resources, when realized in instances of language use, contribute to the tenor of a particular situation. Scholars of tenor analyze the way language users construe the relationships among themselves and their impressions of the social world common to them; Figure 3 displays the various resources that can be used to realize those relationships and impressions (Taylor, 2003).

Within SFL, the interpersonal function of language is realized through the Mood system, which “essentially helps to establish the exchange relations between interlocutors” by establishing the type of clause (e.g., interrogative, imperative, declarative) the writer wishes to use" (Lemke, 1992, p. 85). The Mood system concretely describes how speakers and writers construe dialogic roles for themselves and their audience (in the case of English, through the

arrangement of the subject and finite). The Mood element of a clause consists of (a) an expression of polarity (e.g., yes/no, is/isn’t); (b) the subject, defined as “the thing by reference to which the proposition can be affirmed or denied” (Eggins, 2004, p. 151), and (c) the finite, which

“anchor[s] the proposition in a way that we can argue about it” (Eggins, 2004, p. 152). Mood can be expressed through the lexicogrammatical system of modality, subdivided within SFL into modalization (which relates to information, commenting on the likelihood or usuality of a proposition being true) and modulation (which relates to goods and services, construing an agent’s obligation or inclination to take an action) (Halliday & Matthiessen, 2004; Taverniers,

2008; Thompson & Thetela, 1995).

Figure 3. Components of Tenor. (Taylor, 2003).

SFL theorists have also elaborated other linguistic phenomena that bear on the means by which language users construe their interpersonal reality, such as interpersonal grammatical metaphor (GM). Interpersonal GM represents one example of a use of interpersonal resources that can be fruitfully explained with reference to the SFL grammar. Grammatical metaphor is a key element of the SFL framework that involves the realization of some kind of meaning using

an atypical grammatical form — using a noun, for example, to realize a process (e.g., fight —> the fight). Using GM often allows authors to employ more intricate, crystalline language, packing more information into each clause while reducing its grammatical complexity (Ryshina–

Pankova, 2015; Ryshina–Pankova & Byrnes, 2013). GM also provides resources to construe interpersonal relationships between writer and reader. Interpersonal grammatical metaphor, as described by Halliday (2004) and White (2008), is a subset of GM by which writers express a claim of evidentiality (e.g., it is very likely that the president is a criminal) by means of a mental process, a modal verb, or a modal adjunct. This “extra-propositional element (. . .) has

(rhetorical) scope over a proposition and grounds that proposition” (Taverniers, 2008, p. 19); in other words, using interpersonal GM is a means of adding an additional layer of evaluative meaning to a proposition.

Interpersonal GM often involves reconstruing a mental process to serve as an expression of epistemic or deontic modality. I think, for example, can function to express faith in a truth claim, rather than as mental processes within the experiential metafunction (White, 2008). Under this framework, the statement I believe the president is a criminal is not intended to make a commentary about the writer’s actual mental processes of believing things, but instead to indicate the writer’s commitment to the accuracy of the proposition the president is a criminal.

Interpersonal GM, in other words, provides a means for the writer to indicate “how explicit the writer wants to be about where her assessments are coming from, how subjective or objective she wants them to appear, [and] how definite they are” (Schleppegrell, 2005, p. 183).

Halliday (1985) organizes the use of GM according to explicitness and subjectivity as follows:

• Explicit subjective: I believe that / I think that / I would argue that the president is a

criminal.

• Explicit objective: It is obvious that / it is clear that the president is a criminal. (Here

interpersonal GM construes the belief as fact rather than opinion.)

• Implicit subjective: The president must be a criminal. (The modal verb presents the

evaluation.)

• Implicit objective: Clearly / obviously / surely the president is a criminal. (The modal

adjunct presents the evaluation.)

The effective use of interpersonal GM affects the authority of the writer’s voice. In academic writing in English, for example, beginner writers who overuse implicit subjective modality effect a voice that seems overly tentative, uncommitted to the propositions they are evaluating. Interpersonal GM encompasses resources that some other frameworks discuss separately; for example, some of the markers of evaluation described in Thompson include uses of modalization that can also be discussed using the concept of interpersonal GM.

2.10 APPRAISAL

Also using the overall framework of SFL, Martin and White (2005) developed the

APPRAISAL system to investigate affectively, evaluatively, and epistemically motivated language choices. APPRAISAL is a system network of choices writers can make at the discourse semantic level to engage with their audience through emotional connection, appeal to shared values, and employ persuasive arguments. In terms of the greater SFL framework, APPRAISAL can be seen as a way to categorize discourse–semantic realizations within the metafunction of tenor. It is not integral to the metafunction model, but as the review of APPRAISAL studies will show, it can

offer a particularly applicable description of the semantic work that language does, providing “a complementary view of interpersonal meanings beyond grammar and its clause rank interpersonal systems, such as mood and modality” (Oteíza, 2017, p. 460). APPRAISAL differs from many of the approaches examined so far; it emerged as an elaboration on the SFL grammar and shares SFL’s prioritization of function over form, and it drew on explicitly pedagogical efforts to make resources for expressing attitudes, making evaluations, and creating alignment accessible to second language learners.

One key difference between the APPRAISAL system and other frameworks is that

APPRAISAL looks to the level of discourse semantics, not the level of lexicogrammar, to determine how evaluative meanings are made (Coffin, 2002). As an elaboration of the discourse semantic stratum within the systemic functional view of language, APPRAISAL describes choices that, while they may be realized by a variety of forms at the lexicogrammatical level, share commonalities at the level of semantic meaning. The APPRAISAL approach also facilitates analysis of evaluation patterns that go beyond the level of the sentence or individual word and thus helps capture the overall evaluative prosody (Martin & White, 2005) in texts. The individual meaning of a single token of APPRAISAL may not have significance on its own; however, the intentional accumulation of these tokens throughout a text can “colour the ‘mood’ or ambience of a text and position [the reader] to see, feel, judge and appreciate things in particular ways”

(Macken–Horarik et al., 2018, p. 86). As Coffin and O’Halloran (2005) point out in their analysis of a 2003 story in The Sun newspaper, words or phrases that convey little evaluative meaning out of context may nonetheless contribute to an overall evaluation when used as part of a consistent evaluative prosody. By building up a ‘groove’ of markers that invoke a pattern of positive moral

JUDGMENT, for example, “the text may create the conditions for the reader to make a particular

JUDGEMENT [even when] the writer has left the JUDGEMENT as indirect in the text” (p. 117).

Coffin and O’Halloran are able to use the analysis of a single news item to make a point about the buildup of value-laden meanings. In the article under discussion, the final sentence—

“Mr. Blair will be expected to sign up to the constitution blueprint by the end of June”—while it might seem objective out of context, has the effect of heightening a sense of alarm and opposition when read in the light of previous sentences. Without the benefit of an APPRAISAL of the entire text, analysts might miss an expression of evaluative language that was likely calculated to have a powerful effect on the reader. APPRAISAL, then, offers a way to “trac[e] the build-up of evaluative meaning” (p. 116) throughout the course of an entire text. As Isaac (2012) argues, this represents a departure from previous approaches to interpersonal resources, many of which (e.g., the frameworks of Stubbs, Coates, Hyland, and Biber & Gray reviewed earlier) are characterized by less holistic descriptions and often analyze only explicit word-level lexis and markers of grammar.

This holistic approach to meaning allows linguists to assess the value orientation of an entire text — not merely the assessment a text makes of a certain aspect of reality, but the contribution that the text thereby makes to a social value system recognized by writer and reader.

Lemke (1989) made the case early on for reading entire texts in this light: “Text-meaning realizes social functions (even in texts that may consist of a single word or clause), and among the most important social functions of texts is the maintenance and modification of social value systems” (p. 48). White adopted this reasoning in his elaboration of the ENGAGEMENT framework, which concerns writers’ use of interpersonal resources to evaluate truth claims (this aspect of ENGAGEMENT is closely related to epistemic modality) and thereby create alignment between the writer and the presumed reader. White (2008) argues that while authors can foster

solidarity with their audience by arguing that they hold the same views of the goodness or badness of other things or people, authors can also create that solidarity by appealing to common ideas about what is true and real:

[A]lignment and rapport can just as easily turn upon claims about the material world— versions of events and assertions about what is real—as they can turn upon positive/negative assessments. In the broadest sense, then, alignments are axiological. (p. 146)

Apart from its orientation toward meaning making at the discourse semantic level,

APPRAISAL is also distinguished by its genesis as a tool for writing instruction. The development of the APPRAISAL system was an outgrowth of work on narratives as part of the New South

Wales Disadvantaged Schools Program (Martin, 2014), specifically the ‘Write it Right’ program, which examined 1,000 texts produced by teachers, students and textbook writers within the context of secondary schooling in Australia (Coffin, 1998). From its inception, then, the development of APPRAISAL has been guided by pedagogical considerations, which are reflected in the many recent APPRAISAL studies that draw explicit connections to the best practices of language teaching from their findings (e.g., Coffin, 1998, 2006; Hao & Humphrey, 2012, Liu,

2014; Macken–Horarik, 2003; Macken–Horarik et al., 2018; Mirallas, 2017; Ngo & Unsworth,

2011, 2015; Rothery & Stenglin, 2000).

In its inclusion of a delicate description of APPRECIATION and JUDGMENT—realizations of meaning that reflect evaluations of often-concrete things—APPRAISAL presents a further contrast to models of modality, stance, voice, engagement, and metadiscourse that might seem to more closely track the ways speakers and writers construe and enact the interpersonal relationships between themselves and their audience. Evaluating things and agents in the world might seem to

fall, for example, more clearly within the domain of ideational meanings. White (2008) argues, however, that “all evaluations are interpersonally charged”:

[T]o announce a viewpoint is to affiliate/disaffiliate oneself socially. Evaluations, therefore, have the potential to forge attitudinal communities, bringing together those who share feelings, tastes or judgments and separating those whose feelings, tastes, or judgments are at odds. (p. 146)

In other words: because the act of evaluating something, particularly when communities relevant to the writer are known to have specific views on that thing, has implications for the alignment of the writer with those broader communities (and therefore with the reader).

2.10.1 APPRAISAL Categories

As shown in Figure 4, APPRAISAL includes the categories of ATTITUDE (which in turn includes AFFECT, the expression of emotions, APPRECIATION, the evaluation of things or concepts, and JUDGMENT, the moral evaluation of people or actions), GRADUATION (including

FORCE, a measure of intensification of evaluation, and FOCUS, a measure of the degree to which something is a good representative of its kind), and ENGAGEMENT, which involves the degree of openness to alternative perspectives the writer imputes to themself and the audience.

The discourse semantic meanings described in the APPRAISAL system can be realized using a variety of grammatical categories. Resources for expressing ATTITUDE, for example, might be conveyed via evaluative nominal groups, verbal groups, individual adjectives or adverbs, similes or metaphors, or clauses that reference relevant contextual information

(Macken–Horarik et al., 2018), and each network option has both congruent (explicitly) and incongruent realizations at the lexicogrammatical level (Isaac, 2012). For example, AFFECT is typically realized explicitly within SFL grammar as either a quality (via epithet, attribute, or

circumstance), process (mental or behavioral), or comment (via modal adjunct); Examples (5) and (6) would be congruent realizations of AFFECT:

(5) sie war sehr wütend und traurig ‘she was very furious and sad’ (quality)

(6) Ich habe den ganzen Tag geweint ‘I cried for the whole day’ (process)

Figure 4. The APPRAISAL System. (Martin & White, 2005).

The lexical realizations of APPRAISAL are also situated along a cline of explicitness; some resources automatically carry evaluative weight, while others depend on contextual clues:

APPRAISAL may be “expressed indirectly through contextual and lexicogrammatical features or tokens that imply an evaluation on the part of the writer and trigger an evaluative response on the part of the reader” (Isaac, 2012, p. 165). In some cases, the same word may realize a number of different Appraisal meanings depending on context (Zappavigna & Martin, 2018).

2.10.2 Benefits of the APPRAISAL Approach

APPRAISAL is a fruitful approach for linguists examining “emotional, ethical and aesthetic aspects of evaluation” (Macken–Horarik & Isaac, 2014, p. 68), especially those interested in the means of expressing evaluation in implicit ways. Its emphasis on the social function of language, as well as the accumulation of meanings on the level of generic stages and entire texts, makes

APPRAISAL a useful vehicle for discourse analysis in general (Bednarek, 2008) and the study of texts that draw implicitly on the shared understandings of a given discourse community (Coffin,

1998). For example, linguists with an interest in critical discourse analysis and sociolinguistics were able to use APPRAISAL to shed light on the implications of language used in discussions of socially fraught topics such as the AIDS crisis (Adendorff & de Klerk, 2006) and the sex industry (Adendorff & Pienaar, 2013).

Overall, Oteíza (2017) argues that an APPRAISAL analysis can yield a wealth of insights that other approaches miss “[b]ecause of the possibilities that it offers with regard to the systematisation of interpersonal meanings and the ways in which the negotiation of solidarity between writers and readers legitimate certain positions and social values over others” [p. 469].

The framework’s amenability to additions and modifications also makes it an attractive choice for linguists interested in discourses that have not received as much attention in recent linguistics research (Ngo & Unsworth, 2015). The following sections detail the various systems of

ATTITUDE — AFFECT, APPRECIATION, and JUDGMENT. Unless otherwise indicated, the German examples used below to illustrate each system are drawn from texts elicited during Study 1 of this research project.

2.10.3 ATTITUDE

The system of ATTITUDE refers to the semantic realms of emotion, aesthetics, and ethics

(Isaac, 2012); it includes the subsystems of AFFECT (which evaluates the emotional reactions of human actors), APPRECIATION (which evaluates non-human things and abstract phenomena according to aesthetic norms), and JUDGMENT (which evaluates human individuals or groups according to ethical norms).

While attitudinal resources vary widely in their lexicogrammatical realizations, they share the traits of polarity and gradability. Any instantiation of APPRAISAL can be positively or negatively loaded (Macken–Horarik & Isaac, 2014); e.g., eine großzügige Dame ‘a generous lady’ (+JUDGMENT: propriety), geringe Qualität ‘low quality’ (-APPRECIATION: valuation). The strength of an attitudinal marker can also be increased or decreased by tokens of GRADUATION.

The schema given in Isaac (2012) also adds the possibility of a negated loading, in which the polarity of an APPRAISAL token is reversed by grammatical negation (e.g., gar nicht richtig ‘not at all correct’).

The facets of ATTITUDE can contribute to two of the classic resources of rhetorical argumentation: pathos (appeal to the reader’s emotions) and ethos (appeal to the reader’s moral judgment; Macken–Horarik et al., 2018).

2.10.3.1 AFFECT

The subsystem of AFFECT includes lexicogrammatical resources that construe the author’s emotional states. Based on studies of various written corpora, APPRAISAL scholars have grouped expressions of AFFECT according to the categories listed below (or broadly similar ones).

2.10.3.1.1 Categories within AFFECT

Martin and White (2005) divide AFFECT into the subcategories of Un/happiness,

In/security, and Dis/satisfaction, with Dis/inclination occupying an auxiliary status.

Un/happiness includes expressions of joy, misery, and other generalized “feelings of the heart” (Martin & White, 2005, p. 49), as well as affection (or abhorrence) toward a particular trigger, as in Examples (7)–(8).

(7) Meine Mutter bevorzugte mich und machte mir das Leben so einfach und unanstrengend, wie möglich. ‘My mother preferred me and made my life as easy and comfortable as possible’

(8) Aber wir waren nicht besonders unzufrieden. ‘But we were not especially unhappy.’

In/security includes expressions of fear, unease, and anxiety, or peace, confidence, tranquility, such as in Example (9). Martin & White (2005) identifies these feelings by their shared reference to well-being (or lack thereof).

(9) …nur weil hatte ich Angst ‘only because I was afraid’

Dis/satisfaction expressions of the author’s interest or pleasure in the state of affairs.

Examples of dis/satisfaction such as Example (10) share a concern with goals and their attainability (Martin & White, 2005).

(10) Ich triumphierte, als meine Mutter sprach ‘I rejoiced, as my mother spoke’

Dis/inclination: Bednarek (2008) assigns this category equal rank with in/security, dis/satisfaction, and un/happiness; it involves desire (or lack of desire) for an experience or outcome, as in Example (11).

(11) Ich wollte nur das Gold ‘I only wanted the gold’

Bednarek (2008) also further subdivides each of these AFFECT categories; In/security, for example, can be further divided into the subcategories of disquiet/surprise and confidence/trust.

In my study, I adopt Ngo & Unsworth (2015)’s later revision, which assigns surprise (involving, rather intuitively, the reaction to an unexpected occurrence) its own subcategory, in part because the emotion of surprise has less to do with well-being.

Writers can use expressions of AFFECT to connect with their audiences in two primary ways: through sympathetic connection with the reader, and through an appeal to shared values.

First, by expressing of emotion, the writer makes it easier for the reader to connect with them; in their original synthesis of ATTITUDE, Martin & White (2005) argue that expressions of AFFECT have a persuasive function, “invit[ing] the addressee to respond on a personal level, to empathize, sympathize or at least to see the emotion as warranted or understandable” (Martin &

White, 2005, p. 6). Skilled use of AFFECT can also serve to normalize the views and values construed in a text by indicating the author’s emotional reaction to them (Isaac, 2012).

2.10.3.1.2 Realizations of AFFECT

A realization of AFFECT can reflect either a real occurrence or a human actor’s intentions, desires, or other imaginings; tokens of AFFECT can therefore be coded as either realis or irrealis.

Like other expressions of ATTITUDE, AFFECT is not restricted to a particular lexicogrammatical category. As shown in Figure 5, AFFECT can be expressed through the SFL categories of quality, process, or comment (Macken–Horarik & Isaac, 2014).

Expressions of AFFECT may also take the form of a variety of processes, such as behavioral process (e.g., the boy cried), mental process (e.g., the boy disliked the surprise), or relational process (e.g., the boy felt sad) (Oteíza, 2017).

Figure 5. Realizations of AFFECT. (Macken–Horarik & Isaac, 2014).

Since the kind of interpersonal meaning–making captured by APPRAISAL operates at the level of discourse semantics, we would expect to find similar patterns in a language as closely related to English as German is. Indeed, AFFECT can be expressed via the same processes in

German, as in Examples (12)–(14):

(12) Am nächsten Tag wurde Frau Holle sehr wütend… ‘The next day, Frau Holle was furious…’ (relational process)

(13) …und schimpfte mit mir ‘…and swore at me’ (behavioral process)

(14) Das interessierte sie nicht, weil sie ein Ziel hatte ‘That didn’t interest her, because she had a goal’ (mental process)

2.10.3.1.3 Distinguishing AFFECT from implicit JUDGMENT

While AFFECT was originally described in fairly general terms as construing the emotional states of the author and other human actors, later linguists have cautioned against

applying the AFFECT label too broadly. Thompson (2014), for example, suggests that while

AFFECT applies straightforwardly to the expressed emotional states of the author,

[E]xperiential representations of emotion (. . .) often serve as provoking tokens of judgement. This could turn out to be one of their main, perhaps even overriding, functions in many genres such as narration: other people’s feelings are typically described as a way of depicting what kind of person they are and therefore how the addressee is intended to judge them. (p. 55)

Thompson echoes a note of caution also sounded in Bednarek (2007), who provides the example of the “mother who hates her children” (p. 159) as an expression of an emotion (hatred) that is easily reconstrued as a token of JUDGMENT. In the case of non-authorial AFFECT, then, it is crucial to consider the context; the trigger of an emotion can be just as informative as the nature of that emotion. (In recent years, for example, the phrase he/she loves the president is often used as a token of JUDGMENT rather than AFFECT — depending on the political affiliation of the speaker.) Thompson (2014) therefore limits true AFFECT to instances of “interactant-sourced directed feeling,” relegating other references to emotion to the categories of JUDGMENT or general “textual relief” (p. 56) that is not directly implicated in the APPRAISAL system. In

Example (15), for instance, the narrator describes her mother’s emotions in a way that implies positive JUDGMENT: Propriety.

(15) Schließlich wollte meine Mutter ja nur, dass auch ich endlich etwas Glück im Leben fand ‘In the end, all my mother wanted was for me to finally find some happiness in life’

This could be categorized under AFFECT: happiness: affection, but the effect of the description is to positively construe the mother as someone who quite properly wishes the best for her child.

2.10.3.2 APPRECIATION

If AFFECT represents a personal emotional response, the two further categories of

ATTITUDE—APPRECIATION and JUDGMENT—can be thought of as the reflection and codification

of feelings at the level of entire societies; in Martin and White’s words, these categories represent institutionalized feeling (Macken–Horarik & Isaac, 2014; Martin & White, 2005). In the case of APPRECIATION, these responses are applied to non-human triggers: objects, phenomena, abstract concepts, instances of behavior et cetera (see Chapter 2 for a lengthier discussion of how to distinguish APPRECIATION and JUDGMENT). Humans can also be evaluated via APPRECIATION in terms of their physical characteristics, as in Example (16).

(16) Dazu muss ich anmerken, dass sie — entgegen mir — von besonderer Schönheit war. ‘I must acknowledge that in contrast to me, she was of great beauty.’

2.10.3.2.1 Categories within APPRECIATION

The meaning-making potential of APPRAISAL resources depends a great deal on the context of which things are being evaluated. As scholars of academic writing have noted, studies can be evaluated positively or negatively by commenting on their newness and significance;

APPRECIATION resources that convey these meanings carry greater weight in the context of the academic research article than they might in others. Any given field will involve a range of shared understandings that govern the attributes seen as praiseworthy. In the field of classical music, for example, a review of Bach’s didactic piano fugues might emphasize their adherence to the then-unquestioned rules of counterpoint and harmonic progression; tokens of APPRECIATION such as precise, regular, or exact would be coded as having positive polarity in this context (as they might not if the composer were the more emotion-driven Claude Debussy). It is not surprising, therefore, that a number of subcategories within APPRECIATION have been proposed and modified over the years; many scholars have found it necessary to expand or contract various subsystems depending on the field of the texts they work with. A common scheme can be found in Bednarek (2009a), who subdivides APPRECIATION as shown in Figure 6.

Subcategory (Probe Question) Positive Examples Negative Examples Reaction: impact — arresting, captivating, dull, boring, tedious ‘did it grab me?’ engaging

Reaction: quality — lovely, beautiful, splendid plain, ugly, repulsive ‘did I like it?’

Composition: balance — balanced, harmonious, unbalanced, discordant ‘did it hang together?’ symmetrical

Composition: complexity — simple, elegant, intricate ornamental, extravagant, ‘was it hard to follow?’ simplistic

Valuation: challenging, profound, shallow, insignificant ‘was it worthwhile?’ original

Figure 6. APPRECIATION Categories as Distinguished by Probe Questions. (Bednarek, 2009a, p. 154).

Oteíza (2017) argues that this division reflects the typical realizations of each category;

Reaction is often expressed through the TRANSITIVITY mental process of affection, while

Composition is often expressed through perception and Valuation through cognition. In my study, however, I adopted the changes to this scheme proposed by Ngo and Unsworth (2015)

(see Figure 7), who provide additional criteria for distinguishing between impact (formerly impact: reaction) and quality (formerly impact: quality).

Impact refers to the emotional reaction provoked by the source of the evaluation. As mentioned above, APPRECIATION overlaps with AFFECT when the evaluation of something is based on the emotional response it creates. When the emotional content has been removed from any human source — e.g., eine langweilige Reise ‘a boring trip’ — it can be classified as

APPRECIATION: impact, such as in Example (17).

(17) meine Dienst mit die Frau Holle machte kein Spass ‘My work for Frau Holle was no fun’

Figure 7. The APPRECIATION System. (Ngo & Unsworth, 2015, p. 20).

Following Ngo and Unsworth (2015), quality refers to the ‘goodness of fit’ with a particular standard. This could implicate any token that refers to a previously acknowledged standard (e.g., of beauty, of speed, of music harmony, of scientific rigor), such as schön

‘beautiful’ in Example (18).

(18) Meine Stiefschwester erschien wunderschön und noch schöner als je zuvor ‘My stepsister appeared beautiful, even more beautiful than before’

Composition encompasses the construction of the source of the evaluation. This category of APPRECIATION is particularly applicable to the realm of textual analysis; e.g. ein kohärentes

Argument ‘a coherent argument,’ “[t]he argument was contradictory” (Oteíza, 2017, p. 463).

Valuation, expressed in Examples (19)–(21), is less concerned with aesthetics than quality or

composition; it involves the question of worth and “the application of various normative principles to products and entities” (Lee, 2015, p. 54):

(19) Ich glaubte sie nicht, aber das Gold war echt ‘I didn’t believe her, but the gold was real’

(20) …doch da bekam sie nur noch mehr unerwartete Schwierigkeiten ‘…but there she only encountered even more unexpected difficulties’

(21) Es war das Paradies ‘it was paradise’

APPRECIATION, like other ATTITUDE resources, can be expressed via a variety of lexicogrammatical realizations, as we see in the previous examples.

2.10.3.2.2 Applications of APPRECIATION

APPRECIATION has received particularly detailed attention in the context of academic writing, since this is the primary area in which authorial evaluation is considered acceptable; explicit statements of emotional involvement or moral sanction, by contrast, are less common in scholarly writing (Derewianka, 2007; Isaac, 2012).

2.10.3.2.3 Distinguishing APPRECIATION, AFFECT, and COVERT AFFECT

In their original synthesis of the APPRAISAL system, Martin & White (2005) recognize that a continuum of shared meaning exists between the systems of AFFECT and APPRECIATION.

APPRECIATION, after all, reflects adherence to the standards of aesthetics, and an object’s aesthetic qualities are related in turn to the emotional reaction they spark. In the following example, the source of an emotion has been stripped from the passage, and what might in other contexts be a token of AFFECT is repurposed as negative APPRECIATION: Impact. An object can

be evaluated with emotionally freighted lexis without necessarily implying an agent to feel those emotions, as in Example (22):

(22) The Lauinger Library at Georgetown is a sad, depressing building.

In the absence of a specific emoter, this example would be coded as APPRECIATION: impact (reaction: impact in the original scheme). The distinction between AFFECT and

APPRECIATION, however, is not always so clear-cut. Bednarek (2009b) proposes an intermediate category between the two, COVERT AFFECT, which refers to emotion that is mentioned without reference to a human senser — but in cases in which the human senser is still strongly implied.

Example (23) might be considered an instance of COVERT AFFECT:

(23) ich ignorierte ihre großen, angsteinflößenden Zähne ‘I ignored her large, fear-inducing teeth’

In this case, the narrator mentions an emotion (Angst ‘fear’), and it is clear who is experiencing the emotion (the narrator) even though this is not explicitly stated.

2.10.3.3 JUDGMENT

Like APPRECIATION, JUDGMENT relates to Affect in that it represents the

“institutionalisation of feelings” (Oteíza, 2017, p. 462). In this case, feelings in the realm of ethics relating to morality, legality, and adherence to norms of behavior considered proper by society: “As we are socialized into a culture, the emotional reactions to the world with which we were all born become codified (. . .) emotion is re-socialized as judgements about our character and behavior (. . .) [or] as appreciations of natural and human-made phenomena” (Zappavigna &

Martin, 2018, p. 157). JUDGMENT plays strongly into the ability of a narrative to encode cultural expectations and norms (Isaac, 2012).

2.10.3.3.1 Categories of JUDGMENT

As shown in Figure 8, JUDGMENT is divided in terms of whether a behavior would impinge on legal codes (categorized as social sanction) or is considered socially praiseworthy

(social esteem).

Figure 8. Resources for Expressing JUDGMENT. (Coffin, 2002).

Social esteem includes the subtypes of normality (a person’s comportment with what is considered usual behavior), capacity (involving someone’s physical or mental abilities), and tenacity (concerning someone’s persistence and willpower). Social sanction includes veracity

(the extent to which the character is truthful or trustworthy) and the rather broad category of propriety, which includes a character’s adherence to accepted norms — their degree, in other

words, of “compliance or defiance with the social system” (Rothery & Stenglin, 2000, p. 236).

Instances of JUDGMENT include Examples (24) through (29).

(24) …sie gehörte doch eigentlich gar nicht richtig zur Familie ‘she didn’t really belong (-JUDGMENT: normality) to the family, after all’

(25) …hatte meine Mutter richtig erkannt ‘…my mother had correctly recognized (+JUDGMENT: capacity)’

(26) Endlich passierte ihr etwas Schlechtes und all ihr Fleiß zahlte sich nicht, wie gewohnt, aus ‘finally, something unfortunate happened to her, and all her effort came to nothing (-JUDGMENT: capacity) for a change’

(27) Dabei habe ich mir doch Mühe gegeben ‘I really did give it a shot’ (+JUDGMENT: tenacity)

(28) Was für eine Lügnerin! ‘What a liar! (-JUDGMENT: veracity)’

(29) Gutmütig - oder dumm - wie sie war, teilte sie es natürlich mit uns ‘Naturally, she shared it with us (+JUDGMENT: propriety), good-natured - or foolish - as she was’

2.10.3.4 Realizations of JUDGMENT

As Isaac (2012) points out, JUDGMENT can be realized congruently through both lexis and grammar. JUDGMENT via social sanction is often realized by means of an adjective, nominalization, or verbal process. JUDGMENT via adjective is quite common, as in Examples (30) and (31).

(30) da ich viel klüger und gewitzter als meine Stiefschwester bin… ‘since I am much more intelligent and clever than my sister…’

(31) Meine Mutter nutze diese Gelegenheit, schlau wie sie war aus ‘My mother took full advantage of this opportunity, clever as she was’

In Examples (32) and (33), by contrast, JUDGMENT is realized by means of a nominalized process.

(32) …und müssten nicht dauernd das Kreischen und Weinen der Stieftochter hören ‘…and didn’t have to hear the stepdaughter’s shrieking and crying all the time’

(33) Ich konnte ihren Unmut und ihre Undankbarkeit nicht glauben! ‘I coudn’t believe her cowardice and ingratitude!’

In German, writers have the additional option of using the Konjunktiv I mood to distance themselves from the truth of reported speech. In Example (34), the use of Konjunktiv I implies strongly that the narrator does not believe her sister’s claims; in this case sei acts as a token of negative JUDGMENT: Veracity.

(34) Sie behauptete, es sei aus Versehen gewesen ‘She claimed it was an accident’

Congruent grammatical realizations of JUDGMENT are often achieved via the

MODALITY system, with modalization (probability/usuality) overlapping primarily with social sanction and modulation (obligation/readiness) with social esteem.

2.10.3.5 Distinguishing between JUDGMENT and APPRECIATION

JUDGMENT and aspects of APPRECIATION (particularly the subcategory of valuation) may seem to overlap semantically, and indeed the same tokens may indicate one or the other function when used in different contexts. Bednarek (2007) distinguishes the two in terms of their trigger—that is, the phenomenon being appraised—according to the scheme shown in Figure 9.

Isaac (2012), following Martin (2005), makes a similar distinction; all three categorize evaluations of human beings as JUDGMENT, while labelling outcomes of behavior as

APPRECIATION. Isaac leaves similarly broad room for interpretation; she argues, for instance, that some negatively loaded evaluations of behavior, which would ordinarily be classified as

JUDGMENT, better reflect APPRECIATION when the intent is not to disparage the sources of the behavior in question. She gives the example of children who are described in terms of

‘weakness’ or ‘helplessness’ in the face of governmental interlopers; since the children’s helplessness results from others’ actions, and since the description seems intended to generate

pathos toward the children, it would make little sense to code ‘helplessness’ as an example of the children’s divergence from social norms.

Attitude Trigger (attitudinal target or evaluated Type entity) Example Appreciation abstract things relationships, qualities of life (concrete) material things homes, public buildings, parks, satellites (concrete) semiotic things texts, language, images, painting, music, artifacts, TV shows, films, books, sculptures, plays, recitals, parades, spectacles, performances natural phenomena weather, nature, panoramas, glens, states of affairs happenings and their results

Judgment behavior and actions verbal, mental, physical sentient beings human beings, animals Figure 9. Triggers of APPRECIATION and JUDGMENT. (Bednarek, 2007, p. 128).

The difference in object between JUDGMENT and APPRECIATION resources can have important implications for meaning making. In the field of criminal justice, proponents of restorative justice encourage the use of language that describes an offender’s behavior, rather than the actor itself, as objectionable, the better to avoid labeling the offender as intrinsically criminal. In effect, this makes the case for APPRECIATION over JUDGMENT; correctional officers may find more success construing offenders’ behavior as objectionable, rather than labeling the offenders themselves as deficient in character (Zappavigna & Martin, 2018; Zappavigna et al.,

2010). Example (35), drawn from a text collected in Study 1 represents on such instance, in which the narrator evaluates herself as too sick to work.

(35) Nachdem ich hätte gearbeitet war ich sehr Krank. Die nexte Tage konnte ich nicht arbeiten, weil ich sehr Krank war. ‘After I had worked, I was very sick. The next day I could not work, because I was very sick.’

Here the narrator casts her inability to work as a factor beyond her control; it might therefore make more sense to label this APPRECIATION, or even as AFFECT: in/security, rather than as JUDGMENT: capacity. For further examples of how borders between JUDGMENT and

APPRECIATION have been drawn, see Bednarek (2007, pp. 126–128) and Chapter 3.

2.10.4 GRADUATION

One attribute common to ATTITUDE resources is that they evince positive or negative polarity; most tokens of ATTITUDE (with the possible exception of APPRECIATION: Surprise) evaluate their trigger either positively or negative. This evaluation can also be scaled upwards or downwards, clarified or made more diffuse, through resources of GRADUATION. GRADUATION can be subdivided into FORCE, which involves the upscaling or downscaling of an evaluation, and FOCUS, which includes tokens that increase the specificity or vagueness of the evaluation. As in the case of ATTITUDE resources, GRADUATION can be achieved by means of a variety of lexicogrammatical tools. Among other realizations, these may include comparison particles, exclamations, elaborations, phasing in verbal groups, and intensified verbs or nominal groups

(Macken–Horarik et al., 2018). Crucially, GRADUATION need not take the form of an additional word or phrase; it can also be realized through the use of a verb or nominal group that carries a more intense evaluative meaning (e.g., demolished instead of knocked down, a horde of soldiers instead of an army). In Example (36), wütend connotes a much stronger state of emotion than simply böse or unzufrieden:

(36) Am nächsten Tag wurde Frau Holle sehr wütend und schimpfte mit mir ‘The next day, Frau Holle became very furious (GRADUATION: intensification) and cursed at me’

Hood (2004), in her dissertation, describes potential means of GRADUATION in great detail, sketching out (for example) the various possible means of intensification: via a

premodifier, via infusion in attitudinal attributes, via infusion in a process, and many others. The potential level of delicacy involved in GRADUATION resources is perhaps best realized graphically, as in Figure 10.

Figure 10. The System Network of GRADUATION via Intensification as Elaborated by Hood. (Xu, 2017).

The semantic functions of GRADUATION resources (see Figure 11) overlap with some of the other interpersonal resource frameworks surveyed here. Focus as fulfilment of a process, for example, is typically realized via verbal and mental processes that are described in other frameworks in EAP as encoding stance; e.g., hedging (Hyland, 2005c).

Figure 11. The GRADUATION System. (Read & Carroll, 2012).

2.10.5 ENGAGEMENT

The third component of APPRAISAL, the ENGAGEMENT system, concerns choices writers can make to position themselves toward a given topic and align themselves with their presumed audience, creating solidarity with their addressees (Simon–Vandenbergen, White, & Aijmer,

2007). The ENGAGEMENT system draws on the idea of intertextuality proposed by Bakhtin

(1986), recognizing that any text (or part of a text) exists not in a vacuum, but in dialogue with other texts and voices, taking “up in some way, what has been said/written before, and simultaneously (…) anticipat[ing] the responses of actual, potential or imagined readers/listeners” (White, 2015, n.p.). The Engagement framework meshes particularly well with efforts to understand language at the level of the text.

ENGAGEMENT can be subdivided into moves that expand or contract the range of arguments the author finds it possible to entertain. By taking a dialogically expansive or contractive stance toward propositions referred to in the text, authors can construe those propositions as more or less open to debate, implicitly forging bonds with those members of their audience who hold similar views; this also allows them to “act to construe a readership which is potentially at odds with the writer with respect to the current position” (White, 2008, p. 159) and use interpersonal resources to persuade that readership of worth of their viewpoint.

ENGAGEMENT offers perhaps the most direct conduit toward an understanding of an author’s self-positioning in relation to their audience. In an ENGAGEMENT analysis, each instance of an arguable proposition helps the analyst understand the range of potential ideologies the author is prepared to accept, and therefore the range of potential audience viewpoints with which they claim solidarity (Oteíza, 2017). My analysis does not draw significantly on ENGAGEMENT; for the genres I investigate here, the systems of ATTITUDE and GRADUATION are more informative. A more elaborate explanation of ENGAGEMENT would therefore be somewhat superfluous; interested readers are directed to White (2015) for a current overview of this system.

2.10.6 Explicit and Implicit ATTITUDE

While ATTITUDE can be realized with explicit evaluative tokens, the scope of implicit evaluation is much broader. In her recent summary of APPRAISAL research, Oteíza (2017) describes the difference between explicit and implicit APPRAISAL as follows:

An inscribed appraisal is explicitly expressed in the text, and is associated with specific lexical items and their graduation, whereas an evoked appraisal is manifested in an implicit

manner by reference, for example, to a metaphorical language that can provoke a particular valorative meaning. (p. 462)

Explicit APPRAISAL is denoted as ‘inscribed,’ while implicit APPRAISAL can be subdivided into a number of different types. Martin & White (2005) bundle most implicit

APPRAISAL under the label ‘invoked,’ while others include the categories of provoked (involving the use of metaphor), flagged (involving the use of non-core vocabulary, counter-expectancy, and rhetorical questions), and afforded (evaluation implied through lexis that refers to shared context) APPRAISAL. Bednarek (2007) offers an alternative classification, collapsing provoked and flagged ATTITUDE and reconsidering then as examples of GRADUATION via infusion. The primary distinction, for Bednarek, exists between ATTITUDE that is inscribed (explicitly indicated via evaluative lexis) and that is afforded (understandable with reference to the context of situation or culture). Cases toward the midpoint of this cline of explicitness, involving invoked

ATTITUDE, may be implied through suggestive terms of AFFECT (e.g., ‘Dudley felt guilty,’ which involves an affective response, might also imply that Dudley’s behavior was likely subpar), metonymically implicated by evaluating something adjacent to the object (e.g., remarking on ‘an outstanding performance’ implicitly praises the members of the orchestra who gave that performance), or triggered via meanings from GRADUATION or ENGAGEMENT. Precisely identifying subtypes of implicit APPRAISAL can prove a challenge (see Section 3.5.4), especially since a particular token may fulfill multiple functions at once. Bednarek (2007) argues, for instance, that “inscribed Appraisal of one kind may function to invoke Appraisal of a different kind” [p. 108]. These difficulties are well worth the effort of overcoming, however; mplicit evaluation, Thompson (2014) argues, is

therefore a powerful resource for maintaining values within a culture which gain strength from being so taken for granted that they do not need to be spelt out; and it can, of course, also be deployed manipulatively, since it may be harder for readers and hearers to resist

values which are assumed but not overtly expressed. It can also be used to construct group membership: if you understand the value that is intended to be evoked by the experiential Basis and accept the connection, you thereby display your in-group status. (p. 51)

As Thompson illustrates, successful use of implicit evaluation requires awareness of the shared values that help bind a discourse community together and represents a more demanding means of building solidarity with an audience.

2.10.7 Elaborations to the APPRAISAL Framework

The APPRAISAL framework does not represent a finished product, and recent years have seen both elaborations on, and critiques of, the model. Both the original categorization scheme and its elaborations by Bednarek (2008, 2009b) were based on large numbers of texts, and the network has been further edited and pruned by other authors based on the ideational content of the genres they work with, as well as on the difficulty some authors encountered in distinguishing between certain subtypes.

2.10.7.1 Critiques

Emotions are messy. As Bednarek (2008) points out, it would take an entire book to properly cover the range of disciplinary frameworks that attempt to describe internal attitudes and emotional language. It is no surprise that any system purporting to ‘carve nature at its joints,’ as Plato put it — to impose order on a set of phenomena with naturally fuzzy borders — must contend with a certain level of overlap between its categories.4 The evolutionary biologist

Richard Dawkins (2011) warns against the temptation to cling too dogmatically to artificially imposed borders, arguing that the “tyranny of the discontinuous mind” (p. 54) can lead to pernicious consequences for social and legal systems. In the United States, for example, the legal

4 https://en.wikisource.org/w/index.php?title=Phaedrus&oldid=4812685

categories of ‘adult’ and ‘child’ are highly relevant in criminal justice cases, and the insistence that someone caught up in the legal system belongs unquestionably to either category—or both!—can undermine the ostensible goals of justice (Stillman, 2016).

This is all to say that neither dogmatic adherence to a set of APPRAISAL categories, nor willy-nilly modification of those categories, should be done without a motivation of some kind.

The original framework by Martin & White (2005) was based on a study of a large corpus of data, as was the expansion of AFFECT by Bednarek (2008); this argues for a certain level of deference. Many of the linguists who have proposed amendments to the APPRAISAL framework have also based their critiques in data from naturally occurring language use. One prominent line of criticism, for example, holds that dividing of evaluative markers into the various subsystems of APPRECIATION and JUDGMENT implies a clean divisibility of meaning-making resources that does not exist in actual language use.

It has also been noted that structures often identified with expansive dialogic moves in the ENGAGEMENT system sometimes have a less straightforward effect in actual spoken discourse

(Põldvere, Fuoli, & Paradis, 2016). The problem of fuzzy categories can be addressed in other ways as well. Some scholars interested in APPRAISAL have proposed methods of multiple coding, arguing that many uses of language could be read not as borderline realizations, but as simultaneous enactments of multiple APPRAISAL types, depending on how one reads a given passage (Bednarek, 2007; Fuoli, 2015; Lee, 2015; Page, 2003; Thompson, 2014); in other words,

“an evaluative act can realize more than one parameter” (Thompson & Alba–Juez, 2014, p. 6).

See the following chapter for a lengthier discussion of multiple coding.

2.10.7.1.1 Application to New Fields

Amendments or elaborations to an APPRAISAL system often reflect the context within which a linguist is working. As Ngo and Unsworth (2015) noted, many elements of the

APPRECIATION scheme originally proposed in Martin & White (2005) seem particularly well- tailored to the evaluation of textual composition and rhetoric; for example, the probe question for complexity (‘was it hard to follow?’) lends itself to the complexity of an argument or lecture and less so to a tangible commodity. Ngo and Unsworth (2015), working with narratives, found it useful to subdivide the APPRECIATION category of valuation into significance and benefit/harm.

They further subdivided JUDGMENT: normality into fortune, reputation, and behavior (customary) and JUDGMENT: capacity into the subtypes of mental capacity, material capacity, and social capacity. These subdivisions allowed Ngo and Unsworth to more meaningfully describe the patterns they found in their corpus of data.

Similarly, finding that the scope of APPRECIATION resources varies widely depending on field, Isaac (2012) posited additional subcategories for her analysis of student responses to narratives. Some of her subjects used lexis specific to the field of storytelling in order to construct their appraisals; Isaac therefore instituted the subcategories of composition: narrative device and composition: prominence (foregrounding, backgrounding), as well as a subcategory of APPRECIATION that referred to either the social or symbolic aspects of a text’s sphere of reference. The APPRECIATION subcategory of valuation, which answers the question ‘is it considered worthwhile?’, seems particularly context-dependent; many of the linguists who have proposed revisions to the APPRAISAL scheme include elaborations or clarifications to valuation.

White (2012), whose research focuses on journalism and media discourses, divides valuation in terms of social salience, social authenticity, and social harm; Coffin (2006), researching the

writing of history, according to impact, directness, and validity. Lee (2015), primarily interested in academic discourse, therefore focuses on the concept of worth as typically construed when referring to scholarly articles; within this discourse, a contribution is typically prized if it is new, includes ideas that are challenging and meaningful, is based on reliable evidence, and provides constructive suggestions. In a study of 12 English essays collected from six ESL students from

East Asian countries and six Australian-born L1 English speakers, Lee (2015) found it worthwhile to subdivide valuation into the categories of sociality valuation (‘is it new / socially desirable?’), salience valuation (‘is it significant?’), validity valuation (‘is it reliable?’), and maintenance valuation (‘does it confer advantage?’).

2.10.7.1.2 Implicit Evaluation

As mentioned earlier, Isaac (2012) also increases the level of delicacy in her categorization of implicit evaluation, tracing a cline of explicitness that ranges from provoked evaluation through figurative language (such as euphemism, irony, and understatement) to flagged, afforded, saturated, and evoked evaluation. This level of specificity is motivated by

Isaac’s focus on narratives and student writing exercises that describe them. Effective narrative writers draw on a nuanced repertoire of explicitness in their affective choices, Isaac argues: “the interplay of implied and explicit Affect is crucial to the rhetorical strategy used by the writer to align her readers by constructing solidarity with them and to invoke Judgements of protagonists’ characters” (p. 167). I adopted Isaac’s subdivision of APPRECIATION: valuation into authenticity, consequence, expectedness, possibility, and salience. Authenticity refers to the degree to which something is seen as real or as properly belonging somewhere. As described by Isaac (2012), it parallels the ascription of validity in academic discourse, but it can also indicate evaluations related to identity; when conveying this sense, it “is often expressed implicitly as nominalised

GM describing experiential or conceptual meanings (like ‘identity’, ‘self-discovery’) and lexical metaphors (‘family ties’, ‘ties of blood’, ‘roots’)” (2012, p. 178). I retained this extension to the framework for the analysis of Study 1 texts because of the prevalence of attitudinal makers in the fairy tale texts I collected that commented on the difference between a stepdaughter and her stepmother’s ‘true’ child.

2.10.7.2 Extension to Languages other than English

While the great majority of studies on APPRAISAL have used English as the language of study (with both L1 and L2 language users), some linguists have expanded the framework to other linguistic contexts. Oteíza (2017) and Banks (2017) have identified several such studies in

Spanish and French, respectively. Other APPRAISAL studies in non-English languages include

Vian’s (2008) study of Brazilian Portuguese and Ngo and Unsworth’s (2011) of Vietnamese.

Vian (2008) includes a description of potential APPRAISAL realizations “that work for Portuguese

— and for Spanish, we might add—which are not included in English, such as the nominal group order, the use of several Greek and Latin prefixes, some use of suffixes, use of diminutive suffixes, among others” (Oteíza, 2017, p. 467). Ngo and Unsworth (2011) examine the evaluative meaning-making potential of systems of personal reference in Vietnamese. In that language, the use of an atypical kinship term has important interpersonal implications (e.g., failure to use kinship terms between two people who are in fact related by blood). The use of a person’s given name can also indicate the speaker’s (lowered) attitude toward that person, and the choice of personal pronouns can help to construe the nature of the relationship between

speaker and listener. All of these choices can play into APPRAISAL resources such as AFFECT and

JUDGMENT (especially social esteem or social sanction).

Linguists interested in multimodality have also begun to apply elements of the

APPRAISAL framework to visual and other multimodal media (e.g., Chin, Taib, & Mei, 2017;

Economou, 2012; Macken–Horarik et al., 2018; Ngo & Unsworth, 2015). APPRECIATION and

JUDGMENT, for example, can be realized by “iconic gestures and expressions, stereotypes, dead metaphors, clichés and symbols” (Swain, 2012, p. 93). Political cartoons offer a natural application for multimodal APPRAISAL analysis because of the importance to this genre of persuasion, point of view, and shared understandings. Based on her analysis of political cartoons from a variety of contexts, Swain proposes three evaluative keys, or reporter voices, distinguished by their reliance on particular APPRAISAL resources. Observer voice draws primarily on APPRECIATION; jester voice on negative JUDGMENT: social esteem, and indicter voice on negative JUDGMENT: social sanction.

2.10.8 Theoretical Cross-Pollination: Application of APPRAISAL to other Interpersonal Resource

Frameworks

Some scholars have employed the terminology of APPRAISAL to add explanatory power to other models of interpersonal resources. As mentioned earlier, Coffin (2002) uses APPRAISAL to elaborate a model of voice theory, which she describes as “a descriptive tool for exploring interpersonal styles that have, to a greater or lesser extent, become conventionalized within particular discourse communities” (p. 7). She uses APPRAISAL to examine, in the field of history essays, how writers realize their value judgments when constructing their narratives. In this field, the analysis of APPRECIATION resources is particularly fruitful, since historians often construe historical events as processes that are expressed or commented upon through APPRECIATION

lexis. This type of analysis can also shed light on values common to the discipline at large; “the criteria for valuing a process is shaped by what is regarded as significant or salient within the disciplinary construction of historical knowledge” (p. 12). Coffin (2002) also expands on the

APPRAISAL framework by identifying particular types of voices that can be characterized by patterns of APPRAISAL usage. As they gain proficiency, students tend to move from one voice

(e.g., recorder) to other, more authoritative or argumentative voices (e.g., interpreter, adjudicator). (See also Fløttum et al. (2007) for further illustrations of author roles across three languages.)

One of the most notable applications of APPRAISAL in recent years may be its use in modeling the slippery concept of voice, which as noted above has been defined in numerous ways, often depending on the needs of the author in question. In her dissertation, Isaac (2012) aims to provide an analytical model by which researchers can investigate, explicitly, how writers demonstrate their individuality and enact social identities by making particular interpersonal choices. She finds the APPRAISAL system, along with the complementary Involvement system, particularly useful in identifying ways in which the writer presumes alignment and agreement with their assumed reader and invites the actual reader to adopt certain shared views. APPRAISAL offers a particularly good fit for research on voice, she argues, because it offers means of analysis at both the semantic and lexicogrammatical levels; it also includes multiple subsystems that foreground either the expressive function or the conative function — that is, “the writer’s alignment of the imagined reader to her argument” — of voice (p. 67).

2.10.9 Limitations and Criticisms of the APPRAISAL Model

As Martin and White (2005) made clear from the outset of their proposal of the

APPRAISAL framework, it is not intended to be an infallible, precise description of all

permutations of evaluative language. To the extent that criticisms of the framework address its failure to take account of some element of evaluative language, these can often be ameliorated by increasing the delicacy of the system or rethinking certain categories, as Bednarek (2008) and

Ngo and Unsworth (2015) and others have done. No matter how the APPRAISAL scheme is modified, it will still rely on sorting tokens of evaluation into one category or another. As

Macken–Horarik and Isaac (2014) note, it can be frustrating to try to code borderline cases, both when a realization could belong to one of two categories, or when an instance of evaluation sits along the cline of implicitness in a way that is difficult to pinpoint. They identify three particular challenges facing linguists working within the APPRAISAL framework: Properly identifying implicit cases, describing global as opposed to local patterns of evaluation, and working with evaluation whose meaning is culture- or institution-specific. Thompson (2014) identifies several other difficulties in working with APPRAISAL from the perspective of coding. I have already discussed his argument against classifying all ‘emotion talk’ as instances of AFFECT. He further discusses the difficulty of distinguishing between JUDGMENT and APPRECIATION, as well as what he calls the ‘Russian doll’ dilemma, in which one token of evaluation may be nested within another (and another, and another). Example (37) illustrates an instance of this kind of nesting:

(37) Warum ist meine Stiefschwester nicht nur von besonderer Schönheit, sondern wird auch noch mit Geld überhäuft? ‘Why (-AFFECT: un/happiness) is it that my stepsister is not merely beautiful (+APPRECIATION: quality), but is also bedecked with gold (+APPRECIATION: composition)?’

In this case, evaluations of the stepsister via positive APPRECIATION sit within a clause that implicitly expresses negative AFFECT.

2.11 Applications of Interpersonal Resource Frameworks

The models discussed so far represent the major efforts, throughout the last few decades, to analyze the means by which speakers and writers construe their interpersonal relationships and roles, express a stance in relation to their audience and claims about reality, engage in dialogue and express solidarity with their audience, and perform evaluation of things, phenomena, and people. Scholars in this area have applied these models to a wide array of linguistic goals, from descriptive analyses of academic writing, variations across writing in various disciplines, and narratives, to more pedagogically oriented studies of writing development among students in their first language, as well as writers in EAP, ESL, and L2 learning contexts.

2.11.1 Academic Writing

Numerous studies have applied interpersonal resource frameworks to the study of academic writing, in order to better understand how academic writers learn to negotiate their relationships with their audiences and how they use language to construct their individual academic personae. While academic writing often purports to be an ‘impersonal’ discipline concerned merely with the reporting of information, even the most ‘objective’ academic writing makes use of interpersonal resources, subjective evaluation, and self-mention to some degree, as

Harwood (2005) demonstrated in a corpus analysis focusing on personal pronouns. In the hard sciences, for example, the writer may use personal pronouns to emphasize “the most newsworthy, marketable elements” of their argument (p. 1217) and use self-citations to indicate their self-identified standing in the field, “demonstrating that he is an established player (. . .) with a number of publications already under his belt” (p. 1220). These insights on scientific writing owe a great deal to the comprehensive work of Hood, whose 2004 dissertation provides an overview of objectivity and evaluation in research articles. In this ostensibly objective field,

Hood and Martin (2007) found that evaluation remains prevalent but is expressed in a different form than in other genres: “there is a very strong preference for attitude expressed as

APPRECIATION, rather than AFFECT or JUDGEMENT” (p. 748). In a similar vein, Hood

(2010) found that tokens of ATTITUDE are often nominalized in academic writing. Academic writing also requires the author to engage with numerous additional voices in nuanced ways; this may involve walking quite a dialogic tightrope, including “the expression of ideas in one’s ‘own words’, the authorization of one’s arguments through the inclusion of other sources without over- relying on the quotations from the latter, and the articulation of one’s own critical, endorsing or distancing stance towards the positions taken up by these scholarly voices and in culturally appropriate ways” (Isaac, 2012, p. 8). Writers in a disciplinary community must furthermore demonstrate that they belong to that community and must “have their meanings understood and their identity recognized by other members” (Isaac, 2012, p. 45). These communicative needs are reflected in usages of interpersonal resources that distinguish academic discourses from non- academic ones, as studies from various perspectives have found. Analyses of corpora to elicit patterns of language use have borne fruit for scholars working with many different frameworks.

Even as simple a matter as the selection of interpersonal roles via personal pronouns can be revealing; Kuo (1999) found in one study of 36 scientific journal articles that “writers use strategically exclusive ‘we’ to refer to writers themselves or inclusive ‘we’ to refer to either writers and readers or the discipline as a whole for different communicative purposes” (p. 121).

Working within the framework of stance, engagement, and metadiscourse, Hyland and Tse

(2004), in an analysis of 240 dissertations, found that academic texts tend to be characterized by high use of transitions and hedges because of the need to present arguments clearly and to distinguish between facts (as adjudged by the author) and more debatable claims.

Other scholars prefer a more qualitative approach, using one of the frameworks described here to illustrate the distinctions between a small number of texts, particularly in cases when a literate language user can conclude that the texts are quite different in quality but cannot immediately explain why. Swain (2007), for example, used a comparison of two pieces of academic writing by ESL students to demonstrate that more and less persuasive texts display markedly different patterns of ENGAGEMENT resource usage. APPRAISAL analyses can also be used to evaluate the effectiveness of an EAP course, by assessing students’ ability to write an effective scientific research article before and after taking a genre-based writing course on the subject (Mirallas, 2017), and to illuminate the shared understandings of a discipline (Barletta,

Mizuno, & Moss, 2013).

2.11.2 Variations in APPRAISAL across Disciplinary Communities

In a similar vein, research has shown that different disciplinary communities present differences in the way interpersonal resources are employed. As Hyland and Tse (2004) argue,

[D]isciplines are not only distinguished by their objects of study… [academics’] decisions concerning how propositional information should be presented are crucial. It is these decisions which socially ground their discourses, connecting them to the broad inquiry patterns and knowledge structures of their disciplines and revealing something of the ways academic communities understand the things they investigate. (p. 174)

Hyland and Tse’s (2004) corpus analysis, for example, suggested that more examples of metadiscourse can be found in academic articles in the humanities and social sciences compared to the natural sciences, perhaps reflecting a preference in the natural sciences to avoid overt commentary by, and mentions of, the individual researcher, since these disciplines are supposedly more objective in their research practices. Lancaster (2014) conducted a similar investigation using the APPRAISAL system as his reference. He draws not only on student texts, but also on interviews with instructors, to further elucidate the types of stances that are valued by

the gatekeepers of academic discourses. He finds that in the field of economics, highly rated essays tended to demonstrate what he calls a novice academic stance, one evincing “high commitment, critical distance, and efforts to build a critically discerning reader in the text”

(p. 51); less experienced writers, on the other hand, demonstrated less authority and less critical distance. Lancaster’s triangulated approach also allows him to draw pedagogical implications from his study, whose findings suggest that instructors’ metadiscursive comments might not entirely reflect the kind of writing they actually value highly.

In an exploration of news report genres, White (2003) identifies a subtype of news story that bears similarities to gossip found in spoken conversation. In this Issue Report genre, “the text is organised so that some type of evaluation is strongly foregrounded and by which the text is organised around a patterned alternation between this evaluation and supposedly ‘factual’ information which elaborates or supports that evaluation” (p. 70). Although this analysis predates the full elaboration of the APPRAISAL framework in Martin & White (2005), White relies on the

APPRAISAL categories of AFFECT and JUDGMENT in order to divide his example text into stages containing instantiations of Emotional-Aftermath, Blame, and Praise.

In an analysis of parliamentary discourse, Jakaza (2013) used APPRAISAL to provide a snapshot of evaluative language typical of in Zimbabwe’s governing body from 2009 to 2010, which he then complemented with an analysis of the linguistic resources various newspapers used in order to reframe those words to suit their own ideologies. One advantage of the

APPRAISAL approach in this particular case was its applicability to multiple languages; as Jakaza points out, Zimbabwe is a country with sixteen official languages whose parliamentary debates are characterized by significant code-switching.

In a descriptive analysis of post-match interviews of Australian athletes, Caldwell (2009) found numerous examples of both interviewers and interviewees using GRADUATION and

ENGAGEMENT resources in order to lessen their commitment to assertions made during the interviews. Caldwell argues that journalists follow this strategy in order to maintain a stance of professional neutrality, while players do so in order to conform to an ethos of solidarity characteristic of Australian sports teams.

Swain (2013) analyzed a corpus of articles from Italian newspapers and attempted to group them according to evaluative key, several types of which had been identified for English language journalism. She found that the groupings of APPRAISAL resources common to newspaper articles in English and Italian often did not overlap, making it difficult to categorize articles in Italian based on the existing ‘reporter,’ ‘correspondent,’ and ‘commentator’ voices. As

Swain points out, the less objective character of the Italian press has implications for the country’s political life:

The apparently less clear distinction between ‘factual reporting’ and ‘comment’ articles may be variously interpreted (. . .). It may reflect a common impulse to fill ideational gaps constrained by forms of censorship (. . .) it may respond to the perceived need to bolster the political allegiances of newspapers through moral suasion (. . .). It may also reflect a traditional feature of the Italian journalistic culture which encourages journalists to be opinionisti, and to acquire professional prestige and social status through their comments and opinions. (p. 77)

It is no surprise to scholars of politics or media studies that Italian language newspapers often blur the boundaries between ‘news’ and ‘opinion’ journalism; where this study contributes to existing knowledge is in quantifying those difference, offering an example of how APPRAISAL analysis can complement existing studies in media studies, political science, etc. APPRAISAL studies have also explored community-building facilitated by newer media; Zappavigna (2011)

explored the solidarity-forming function of tweets, in this case post-election tweets in which

“users are affiliating around values relating to the election result” (p. 798).

APPRAISAL can also yield insights in the context of criminal justice. Zappavigna et al.

(2010) used APPRAISAL in conjunction with a multimodal rhetorical analysis to illuminate the way law enforcement agents can assert their affiliation with offenders and appeal to offenders’ positive image of themselves while simultaneously negatively evaluating their behavior.

Zappavigna and Martin (2018) return to this theme with an analysis of evaluative coupling in a set of interviews drawn from the same corpus. Coupling occurs when a particular participant is repeatedly paired with a particular attitudinal value; in the case of this interview, a juvenile offender’s mother, construed as the emoter, is paired with negative affect as a means of encouraging the offender to empathize with her suffering.

2.11.3 Student Writing (L1)

Both emerging L1 writers and less experienced academic writers face the challenge of adapting the language resources they can control to new contexts with requirements that differ from the discourses they had previously mastered. Some students, even in their L1, may struggle with the expectations codified in their instructors’ assessment practices, which in a study by

Coffin (2002) “frequently turned on the ability of students to a) unpick the often highly indirect and implicit evaluative meanings of texts and b) infuse their own texts with the appropriate interpersonal colouring, to adopt, as it were, the right ‘voice’ and thus an effective intersubjective positioning strategy” (p. 9). Comparative studies have consistently found that student writers display less practiced command of interpersonal resources than mature writers;

Gruber (2004), for example, found that professionals tend to cluster deontic modality in their conclusion sections, while student writers tend to use deontic modality more intermittently. This

finding seems to have pedagogical implications for L2 writing instruction, of the kind that

APPRAISAL scholars often make explicit.

Tracing the use of interpersonal resources can provide a means of differentiating between high- and low-rated student essays in a variety of fields. Among undergraduate essays in geography, for example, Mei (2007) used an ENGAGEMENT analysis to show that low-rated essays contained more bare assertions (monoglossic clauses), while higher-rated essays tended toward heteroglossic and dialogically expansive clauses. In a similar vein, Lee (2015) found in an analysis of essays in social science that successful writers showed a preference for the maintenance valuation and salience valuation subtypes of APPRECIATION (where valuation is

“concerned with writers’ assessment of the social significance of the text/process” [p. 51]) when making an evaluation, whereas less successful writers are more vague in their use of

APPRECIATION.

The ENGAGEMENT system seems particularly useful for illustrating many of the difficulties students often have with crafting pieces of persuasive writing. This is a difficult task whether in a writer’s first or second language; at the undergraduate level, students are often expected to construct nuanced arguments that do not merely recapitulate, but also evaluate the proposals of other authors: “Particularly as students progress into upper-level writing courses in the disciplines, they are expected to construct stances that are at once critical, authoritative, and dialogically expansive” (Lancaster, 2013, p. 212). The difficulty of writing in this way can be compounded by miscommunications between professors and students about what kind of writing is expected and how it will be evaluated. Furthermore, as Lancaster and others have demonstrated, different disciplines may value different kinds of argumentative styles; that is, a highly graded essay in economics may exhibit different APPRAISAL patterns than a highly graded

essay in history. Implicitly, this study touches on the larger question of how many disciplinary ways of writing students should be taught. While Lancaster refrains from making explicit judgments on this topic, APPRAISAL studies such as these can still inform discussion on this question, by clearly illustrating the characteristic linguistic differences between various discourses and hinting at the aspects of those disciplinary shifts that students typically struggle to master.

APPRAISAL can also offer a framework to analyze work from much younger students, who are often called on to produce persuasive writing in demanding contexts. Derewianka

(2007) conducted a detailed analysis of texts from students in secondary school to explore the ways in which more experienced writers tend to recontextualize personal emotions as moral

JUDGMENT and APPRECIATION of broader situations. Thomas, Thomas, and Moltow (2015), in an analysis of essays from a national writing test for Australian students in years 3, 5, 7 and 9, found that the APPRAISAL framework allowed them to describe trends among the essays. The highest-scoring students, for example, deployed a greater proportion of invoked ATTITUDE markers than inscribed ATTITUDE.

Similarly, research by Macken–Horarik et al. (2018) on English narratives written by students of similar ages has suggested that “one hallmark for development (. . .) is a growing capacity to deploy metaphor, Elaboration, and other resources associated with Invoked

Appraisal” (p. 97). Their work suggests that teachers should also be attentive to their students’ developing abilities to indicate evaluation prosodically — that is, by conveying evaluation via a series of tokens, so that the accumulation of meanings makes it clear that the target of appraisal is meant to be seen in a certain light. The contrastive analyses reported in Macken–Horarik et al.

(2018) suggest further benchmarks for writing development among students in their first language:

With the move to adolescence, there is an expectation that a writer’s evaluations will display more awareness of the social values of the community, drawing on a wider range of attitudinal resources from the sub-systems of Judgement and Appreciation to evaluate, critique, and challenge. (p. 192)

More advanced writers among the texts surveyed by Macken–Horarik et al. (2018) also tended to construct an ‘expert’ stance for themselves by means of impersonal evaluations and appeals to logos, and their writing evinced more indirect forms of GRADUATION (p. 202). Many of these applications to student writing aim “to demystify ‘being critical’ from a linguistic perspective” (Lee, 2015, p. 73) — that is, to apply linguistic analysis to feedback on student writing that can often seem frustratingly vague (e.g., be more concise, adopt a more critical stance, express these ideas more clearly). Studies such as those cited above would fit well into a cycle of program evaluation (Davis et al., 2016), providing a common rubric for teachers to use to assess the writing they expect to see against the writing their students have mastered.

2.11.4 Second Language Writing

Several studies in recent decades have used interpersonal resources as one lens to examine the ways in which L2 writers develop their abilities to construct an authoritative authorial persona and incorporate multiple perspectives into their writing. For L2 writers, both the language and the social context of the discourses in which they write may be less familiar to them than to writers using their L1. An analysis focused on the interpersonal function, and on

APPRAISAL in particular, can yield clarity on specific elements that characterize effective L2 writing, which in turn can guide instructors and curricular developers in their establishment of learning goals.

When analyzing L2 writing, it is important to keep in mind that students writing in a second language do not comprise a monolithic bloc; such writers may include heritage language learners, ESL students, students in EAP or ESP programs, and students developing competence in a foreign language that they do not encounter outside of school. All of these groups of students would be expected to have different needs and display different levels of command of interpersonal resources. Even students who have had a good deal of exposure to the target language, whether at home or in a study abroad context, may struggle to make appropriate meanings in their writing (Schleppegrell, 2005). Needs of writers in second language environments may vary only according to level of language exposure, but also according to the students’ age and overall experience with writing. Hirvela and Belcher (2001), in a case study of professional writers whose life situations compelled them to begin writing in a second language, found that their subjects had starkly different needs than novice writers at similar levels of language proficiency, since their subjects

… already possess voices and established identities as professional writers in their native language. Having achieved some measure of success and recognition as first language (L1) writers, these students come to our classrooms with an already existing self-representation of themselves as writers, indeed as good writers. They have, by virtue of their experiences and achievements, already grappled with many of the issues we address in L2 writing instruction, albeit in an L1 context. (p. 84)

Hirvela and Belcher (2001) therefore see the project of teaching second language writers as one that should be embarked on with a recognition of those writers’ existing tools, with a view toward enhancing what they are already able to do.

Studies of writing in ESL and English for Academic Purposes contexts often include comparisons between L1 and L2 writers, as well as between highly and less highly rated essays by L2 writers. Studies by Lee (2006) and Liu (2013) offer illustrations of how an APPRAISAL analysis can contextualize the gaps between such groups of essays. In a comparison of two

essays by Chinese learners of English, Liu (2013) found that the more successful essays involved resonances between the tokens of APPRECIATION and AFFECT that were most often emphasized by GRADUATION; higher-rated essays included more consistent attitudinal prosody involving surges of emotion attributed to the author, which also contributed to a clearer authorial stance.

Lee’s dissertation analyzed argumentative/persuasive essays written by L1 English speakers and undergraduate students of East Asian backgrounds. Among other results, she found that the

APPRAISAL usage of high-graded essays (both between L1 and L2 writers, and among L2 writers) included significantly more examples of multiple coding (that is, when a word or phrase can fulfill more than one APPRAISAL function at once; see Chapter 2 and, e.g., Barletta, Mizuno, &

Moss, 2013, Bednarek, 2007; Fuoli, 2015; Lee, 2007; Page, 2003). This finding might suggest that an APPRAISAL approach is particularly well suited to analyzing higher-quality writing, as other models of interpersonal resources seem less amenable to the use of multiple codings.

Studies on stance (in Hyland’s sense) have suggested correlations between L2 writers’ use of hedges and boosters and their ability to exploit the full range of authoritative potential in academic writing. L2 writers tend to use a narrower selection of hedges and boosters compared to L1 writers (Flowerdew & Wang, 2015; Hinkel, 2005). In many academic disciplines, “caution, possibility, and delimited claims [are privileged] over certainty, while incoming [first-year] writers tend to use stance features that achieve the opposite kind of stance” (Aull & Lancaster,

2014, p. 164). Working within the ENGAGEMENT system of APPRAISAL, Aull and Lancaster found similar results to Flowerdew and Wang as well as Hinkel; the less experienced L2 writers in their study tended to use fewer unique markers of epistemic stance, repeatedly using the tokens with which they were most familiar. Aull and Lancaster argue that as writers become

more adept at deploying ENGAGEMENT resources, they learn to express stance through less direct means, reflecting a nuanced awareness of the views that inform their own argument.

In a classroom context, research on interpersonal resources has sometimes examined L2 writers’ ability to engage with their audience in genres involving greater interpersonal distance, such as the academic article and the book review. Just as research on interpersonal resources has evinced a trend toward recognition of the dialogic, reader-centered role of that linguistic function, research on L2 writing suggests that one indication of writing development may be an increase in intersubjectivity and orientation toward the reader in their texts. Ryshina–Pankova

(2011), for example, conducted an analysis of L2 writers’ book reviews based in part on the writers’ deployment of theme. Writers can employ various strategies in their use of theme; for example, they can thematize references to the audience (or the author) in order to explicitly construe the reader–writer relationship, or they can choose to emphasize ideational meanings, therefore downplaying that interpersonal relationship. Ryshina–Pankova used a quantitative analysis of theme type and reader/writer mentions, as well as a qualitative discussion of the function of various themes, to trace a developmental trajectory of L2 writers in terms of their ability to engage and persuade the reader, and the resources they are able to use in order to do so.

This study found a trend in L2 writers’ book reviews toward less overt expressions of stance and more indirect mentions of the reader. Most starkly, the more advanced L2 writers included more thematizations of the reader and fewer thematizations of the writer in their texts.

Ryshina–Pankova (2011) also sees the writers as making progress in their construal of the reader-in-the-text; in order to persuade the reader of the book review to read the book in question, “the writer strategically construes the reader’s potential reaction to the text and enacts him as someone who is interested in the book and puzzled by its plot” (p. 251). Some of the

writers analyzed in this study also demonstrated awareness of the particular affordance of

German in their use of man, the impersonal/generic first-person singular pronoun, which can be literally translated as one but is more widely used than its English counterpart in order to

“construe [. . .] an argument that displays a common judgment and enjoys a common support” (p.

252).

Writing within the context of a professional discourse community and presenting a consistent authorial identity, one that reflects an ongoing dialogue with other voices and is clearly positioned vis-à-vis a presumed reader, presents a challenge to both L1 and L2 writers; particular discourse communities are characterized by often-implicit understandings about the level of certainty, authorial presence, and explicit reference to other voices it is appropriate to project.

Regardless of their L1, developing writers must typically be ‘socialized into’ discourse communities in which they interact professionally. Overt instruction can support this learning process; Ryshina–Pankova’s (2011) study indicated that L2 writers can, over time, make significant progress in their ability to construe an ideal reader and thereby set up that reader as open to persuasion. Isaac (2012) contributes to this goal by offering six criteria of an APPRAISAL- based model of voice which teachers and students could use to analyze and improve L2 writers’ work; such a model, she argues, should (a) be context-based, (b) define the relationship between the actual writer and the identities projected by that writer, (c) reflect the aspects of voice that contribute to both self-portrayal and dialogue with the reader, (d) account for meaning-making at multiple levels beyond a selection of lexicogrammatical features, (e) account for variations in voice not only between texts, but also between stages of the same text, and (f) be able to describe

a writer’s choices both to adopt conventionalized forms appropriate to a particular context and to

“fus[e] these choices in spontaneous and idiosyncratic ways” (Isaac, 2012, p. 18).

Isaac’s (2012) model represents a step forward in our understanding of voice and interpersonal resources generally in that it illustrates a link between a writer’s linguistic choices and their projected persona, “relat[ing] syndromes of interpersonal choices to the enactment and projection of the writer’s identity” both at the level of words and clauses and at the level of prosodic, cumulative expression; it also provides a more detailed model of how writers both align with their readers, “foreground[ing] the writer’s creation of bonds of affinity with an ideal reader through the projection of their reciprocal identities” (p. 35) and reflect understandings common to the disciplinary and cultural context.

A consistent recommendation from these researchers is that means of deploying interpersonal resources should be explicitly taught (Lancaster, 2011; Lee & Deakin, 2016). For

L2 writers who have experience crafting a clear authorial voice in their L1, overt instruction in the use of ENGAGEMENT moves such as those described by Aull and Lancaster, with reference to the conventions of the register appropriate to the students’ task, can be a useful awareness- raising tool.

2.11.5 Narratives

The work of evaluative language is particularly overt, and essential for meaning making, in narrative genres. Zappavigna and Martin (2018) go so far as to call it “an essential part of all storytelling” (p. 181). Of the frameworks outlined in this literature review, APPRAISAL seems to offer a particularly rich means of analyzing the instructive power of narratives — among other things, by examining the process of how narratives “‘go to work’ on readers” (Macken–Horarik,

2003, p. 485) by constructing axiological (value-oriented) relationships with their presumed

audience. An APPRAISAL analysis also offers a means of categorizing various types of stories; by mapping patterns of APPRAISAL types across a variety of story texts, narratologists can create system networks that describe these story genres (Zappavigna & Martin, 2018). AFFECT seems to be a particularly meaning-laden resource for narratives. Isaac (2012) found that this APPRAISAL category “serves three principal purposes: characterisation, identification and bonding, and the development of ethical discernment” (p. 161). AFFECT offers a natural means of evaluation for narrative genres, which permit the explicit expression of emotional states (through the voices of the narrator, other actors, or the characters themselves). Characters in narratives are often

“associated with a preferred mode of expression or emotional style” (Macken–Horarik & Isaac,

2014, p. 81); an analysis of AFFECT across multiple characters may reveal patterns in the way different voices express themselves and can provide grist for richer interpretations of the narrative.

As Thompson (2014) points out, however, emotional expressions should not always be taken at face value and may in fact constitute instances of JUDGMENT: “[O]ther people’s feelings are typically described as a way of depicting what kind of person they are and therefore how the addressee is intended to judge them” (p. 55). The interweaving of both AFFECT and JUDGMENT allows a skillful author to spark affinities between their characters and their presumed audience, inviting the reader “to a position of APPRAISAL and narrative empathy—emotional solidarity with or, at least, understanding of the motives of a given character. Secondly, the reader is expected to take up a position of discernment—adjudication of the ethical values adopted by a given character” (Macken–Horarik, 2003, pp. 286–287).

AFFECT, APPRECIATION, and JUDGMENT provide a lens through which to trace differences between storytelling styles. In a study of 23 oral narratives on the topic of childbirth, which

included 14 women and 9 men, Page (2003) found that men’s narratives tended to include more instances of JUDGMENT in which they appraised their own participation negatively in terms of capacity and tenacity, “construct[ing] roles for themselves that are peripheral to the experience of childbirth” (p. 230). In narratives by women who assisted in childbirth, by contrast, instances of

JUDGMENT tended to be interwoven with expressions of AFFECT that emphasized their empathy with and closeness to the birth mother. These patterns of APPRAISAL, Page argues, strengthen a culturally embedded understanding of mothers as primary caregivers and men as secondary helpers that goes beyond whatever gendered differences might exist in nature. APPRAISAL analysis has also been used to offer insight into writing about narratives, such as student evaluations of school texts. Rothery and Stenglin (2000) used APPRAISAL to analyze a highly rated student essay written in response to a prompt on the final public examination for English in secondary schools in New South Wales, Australia. Their analysis revealed a notably high proportion of JUDGMENT markers, perhaps surprising given that the students were ostensibly writing about the effectiveness of a literary text in exploring a particular theme. The prevalence of JUDGMENT becomes less surprising, Rothery and Stenglin argue, if narratives are viewed as vehicles for communicating social norms. Students heavy use of JUDGMENT here, they suggest, dovetails with the convention that “English as a school subject is primarily about apprenticing students into cultural values and a socioculturally determined ethical system of behaviour”

(p. 242).

The analysis of narrative stages in this dissertation relies on the formative work of Labov and Waletzky (1967), which was extended with a linguistic analysis by Rothery and Stenglin

(1997). Narratives as described here involve “individuals struggling against adversity as represented by a disrupted activity sequence that leads to a crisis point” (Rothery & Stenglin,

1997, p. 240) and are differentiated from other story genres by their three- or four-stage structure.

Narratives begin with an Orientation, which apart from introducing the main characters and setting creates a “context for understanding what is to follow in the subsequent stages of the genre” (Rothery & Stenglin, 1997, p. 236). Narratives are distinguished from recounts by the inclusion of a Complication stage (and its attendant Resolution) and may also include a Coda, which typically involves a shift in spatiotemporal perspective as the narrator comments on the present state of affairs in contrast to the past (Toolan, 2001). The Complication/Resolution pair, which represents the crux of a narrative, typically involves “disruptions that constitute a crisis whose outcome is problematic for those involved in them” (Rothery and Stenglin, 1997, p. 239).

Interpersonal resources are crucial to the construction of an effective narrative; they allow the author to “build[] up tension and suspense about the outcome of events and highlight[] their danger for the narrator (. . .). [I]nterpersonal meanings serve to create a sense of responsibility for the narrator to counteract (. . .) the disruption and overcome it” (Rothery and Stenglin, 1997, p. 253). Narratives provide an especially rich avenue for the exploration of evaluative language, as “[e]valuation is more often interspersed throughout a narrative than limited to one stage”

(Macken–Horarik & Morgan, 2011, p. 289).

2.11.6 Pedagogical Applications

Many of the approaches to voice, situated as they are within the disciplines of rhetoric and composition, naturally explore the concept as a means of explaining it to developing writers.

Likewise, it is not surprising that the APPRAISAL framework, which grew out of efforts to improve secondary school students’ understanding of media discourse, contains pedagogical applications within its DNA. Numerous APPRAISAL studies conclude their analyses with

recommendations for the classroom based on their findings (e.g., Chang & Schleppegrell, 2016;

Lancaster, 2011). In a recent comprehensive proposal for rethinking the teaching of English,

Macken–Horarik et al. (2018) incorporate APPRAISAL as a key tool in their model of functional grammatics, which aims to make the meaning-making potential of grammar more accessible to both students and teachers. These efforts can be applicable to both L1 and L2 instruction; Ngo and Unsworth (2015) list pedagogical considerations among the primary justifications for their revision of the APPRECIATION system:

A more delicate system as proposed will better facilitate the work of language teachers by providing them with a toolkit to support learners to develop a repertoire of diverse attitudinal meanings and linguistic resources in realising these attitudinal meanings, thus, enabling language learners to achieve more evaluative precision in communication in contexts that they will certainly encounter as an international tertiary student in an English speaking country. (p. 22)

Scholars with an interest in pedagogy have found it fruitful to apply these insights to written genres that novices find particularly demanding, such as the scientific research article.

One approach to development of scientific writing skill involves the analysis of ideational– interpersonal couplings, which “allows us to propose a pathway to support students in their undergraduate biology courses to develop a scientific argument” (Hao & Humphrey, 2012, p.

174). In an analysis of published research articles in the field of biology, Hao and Humphrey

(2012) illustrated a generic progression for the Research Warrant stage based on recurring couplings of attitudinal resources with the ideational fields or subfields they appraise. Research

Warrants typically begin by discussing objects of study appraised via positive valuation, move on to negatively appraise the research of other scholars by means of APPRECIATION: composition, and set up the current study by assessing it with positive APPRECIATION: composition resources.

Hao and Humphrey (2012) argue that their findings have immediate pedagogical relevance: “the

multi-functional linguistic analysis on which [the identified moves] are based makes visible to students the tools for achieving the goal of this important stage of research reports” (p. 188).

Educational linguists have also used evidence from APPRAISAL analyses to buttress calls for reform in the way texts are taught as socially valued constructs. Rothery and Stenglin (2000) concluded that then-current teaching practices did not provide students with the tools they needed for a critical reading of texts whose values conformed to the dominant paradigm; instead, literature was “taught in school contexts in such a way that it ‘naturalises’ conventional sociocultural practices and values” (p. 243). Like Coffin, they argued for teaching students to use

SFL resources to understand how texts implicitly convey social values. This is a particularly relevant concern in the field of history, where the preferred narratives of the dominant social group may be taught as though they were uncontroversial. These educators see APPRAISAL as a tool that helps teachers ensure that students do not come to view history as an “objective, value free discipline” (Coffin, 1998, p. 17).

Successful assessment is another key aspect of any comprehensive pedagogical effort, and research based on interpersonal resource analysis has contributed insights in this vein as well. Macken–Horarik (2003) draws on interpersonal resources in order to sketch out a means of assessing successful responses to narrative; at a further level of remove, Mirallas (2017) used

APPRAISAL as a tool with which to assess the success of a pedagogical intervention.

Shifting the lens from students to teachers, APPRAISAL also represents a powerful tool for

SFL researchers interested in teacher training and teacher cognition. Goetze (2018) used

APPRAISAL as one theoretical framework in an interdisciplinary dissertation that bridged insights from SFL and psychology to better understand the roles that FL/L2 teachers’ emotions and anxiety play in the language classroom.

2.12 Limitations and Avenues for Further Study

In this review, I have described various means of studying how language users construe dialogic roles; perform evaluation; take a stance toward other voices and ideas; and create alignment and solidarity with their ideal audience, as well as some applications to which those frameworks have been put. While they differ in theoretical underpinnings, the conceptualization of most of these models has tended toward emphasizing function over form and taking the reader’s perspective into account alongside that of the writer. Current research on interpersonal resources, while capacious, is hampered by a vagueness of definitions and is characterized by a lack of studies examining (a) genres other than academic writing and (b) languages other than

English. Some interpersonal resources have proved slippery to define, with scholars from different disciplinary backgrounds often using the same term to capture slightly different aspects of language, as in the case of ‘voice’ (Hyland, 2005c; Matsuda & Tardy, 2007; Stapleton &

Helms–Park, 2008; White, 2015). This difference is somewhat to be expected, since most of these terms are employed in the context of different disciplines; their use will naturally reflect the understanding of theories in which they are mentioned. Still, it makes reviewing the literature on voice and engagement somewhat cumbersome.

The scope of investigations into the use of interpersonal resources could also be broadened. This criticism is a qualified one, for there are in fact numerous studies within SFL that explore contexts as wide-ranging as spoken conversation (Yang, 2013), music (Caldwell,

2014), journalism (Achugar & Oteíza, 2009; Bednarek, 2006; Huan, 2015; Liu, 2017; Martin,

2004; Swain, 2013; White, 2003), politics (Jakaza, 2013; Miller, 2004), sports (Caldwell, 2009), and narrative (Macken–Horarik, 2003; Page, 2003; Rothery & Stenglin, 2000). However, within research on interpersonal resources in general, the bulk of research on authentic texts has focused

on academic writing (see, e.g., Flowerdew & Wang, 2015; Hyland, 2001; Lancaster, 2013, 2014,

2016; and Liu & Zhou, 2014, among many others). To the extent that interpersonal resources are examined in an L2 writing context, that second language is very often English (Belcher, 2013;

Isaac, 2012), with some exceptions previously mentioned in the Section 2.10.8. To the best of my knowledge so far, only a few studies (Ryshina–Pankova, 2011; Crane, 2016) explicitly examine interpersonal resources in German learner texts from the SFL perspective.

One consequence of this lack of research is that students and teachers of German may not be fully aware of the linguistic resources language learners at various curricular levels may be able to use in order to successfully make evaluative meanings. This kind of meaning making goes beyond the acquisition of appropriate vocabulary; different types of interpersonal resources may be involved in evaluative moves, depending on the genre of the text in question. (For example, effective irony in fairy tale parodies is often accomplished by means of GRADUATION,5 while evaluation of other genres might be characterized by a greater variety of inscribed or invoked ATTITUDE markers.) Understanding these patterns is an important step toward being able to effectively teach language learners to engage their audience in the ways they intend.

Opportunities for further research present themselves at the intersection of these applications.

While research on student writing assignments through the lens of interpersonal resources is hardly wanting, the scope of those assignments seems to be constrained to academic discourses; the texts available for analysis are often ones that students would write in response to a specific prompt or to satisfy a singular audience (the instructor). Relatively little work seems to have been done on other genres of writing in which L2 writers might engage — narratives, for example, or other instances of less public-facing writing.

5 I am indebted to Marianna Ryshina–Pankova for this insight.

This project aims to expand the knowledge already gained through the study of interpersonal resources by applying the APPRAISAL lens to L2 German writing in narrative genres. This research would also involve applying the APPRAISAL model to another language, shedding light on its applicability in a non-English context. Hopefully this work can not only provide additional insight into the development of L2 learners’ ability to use interpersonal resources in a wide set of contexts, but also offer pedagogical implications for the second language classroom.

2.13 Reasons for Undertaking this Project

As I outlined in Chapter 1, this project responded in part to a call for additional studies of non-academic writing in languages other than English, as well as to the suggestion by Liu and

McCabe (2017) and Myskow and Ono (2018) that patterns of use among APPRAISAL subtypes, as well as sources and targets of APPRAISAL and implicit evaluation, could bear additional attention.

The choice of the narrative genre in particular was further motivated by my experiences as a language teacher. Through my training as a graduate instructor in the GUGD and as an immersion language teacher at the Concordia Language Villages (CLV), I have worked with many students whose desire to learn sometimes outstripped my ability to offer scaffolding for their learning. Students who invest years, curiosity, and diligence into learning an additional language should have the opportunity to use that language appropriately and effectively, in a variety of contexts, to achieve their aims. I believed, for reasons fleshed out in the next section, that command of narratives would offer my potential future students a particularly powerful tool with which to pursue those goals.

In their respective mission statements and curricula, both the GUGD and CLV center a vision of language advancedness that goes beyond communicative competence. The GUGD curriculum, as discussed earlier, draws on insights from curricular reformers in the 1990s and

2000s who oriented their efforts around the goal of guiding students toward multicompetence

(Hall, Cheng, & Carlson, 2006) or (multi)literacy (Byrnes, 2001, 2002; Byrnes & Kord, 2001;

Cope & Kalantzis, 2009; Gee, 1998; Kern, 1995, 2002; Paesani, 2016; The New London Group,

1996). The title of the current curriculum—Developing Multiple Literacies—refers to an ongoing project of curricular renewal (now entering its third decade) that aims to educate

“competent and culturally literate users of German” who “critically explore their own assumptions in terms of [the German-speaking] world” (Georgetown University, 2020).

As an instructor at Georgetown, I worked with students who would go on to work and study in a variety of communities outside the English-speaking bubble; the curriculum, with its focus on genre, helped students practice habits of reading and inquiry that should fit them to communicate with situational and cultural awareness and pursue their goals across various contexts. This image of our future students also characterizes CLV, a summer learning institute offering immersion instruction in 15 languages primarily to middle and high school students, even while it differs in many respects (learner demographics, duration, prevalence of written language) from a college language department. CLV’s goals also reflect an understanding of language instruction as a springboard toward engagement with others across multiple languages and professional contexts.

CLV’s current mission statement expresses the primary goal of “inspir[ing] courageous global citizens”; the long form of the mission statement includes five goals—only one of which explicitly mentions language—that gesture toward the roles CLV students may one day play in

the global community. Instruction at CLV’s German language village, Waldsee, does include a number of playful activities, games, and songs designed to spark students’ love of language learning, support them in learning ‘survial German,’ and encourage them to return in future summers (Hamilton, Crane, & Bartoshesky, 2005). However, programming for the community as a whole foregrounds the development of language users who can modulate their language according to the context of situation. The immersive environment allows instructors to simulate a

‘playworld’ in which students and teachers can take on and trade personal and professional roles, adapting their language to suit (Hamilton & Cohen, 2004); physical aspects of the site from its architecture to its signage contribute to encourage cultural practices that would be familiar in

German-speaking countries (Quam & Hamilton, 2020). The entire student body also participates in historical simulations designed to familiarize students with cultural referents—e.g., national socialism and memorials to the Shoah, the Wende and reunification, guest worker programs and current discourse on immigration—that a literate user of German would recognize.

All this is to underline the central role of increasing students’ learning opportunities in the motivation to pursue this research. I chose to examine student writing in the hope of adding granular detail to the profile of language learners at various proficiency levels, the better to meet students ‘where they are’ with scaffolding that challenges them appropriately. Hopefully I have already justified the focus on L2 German and genres outside of typical academic writing; the specific focus on narratives, however, deserves further explanation.

2.14 Affordances of Narratives

I selected narratives as the topic of this project because of their enormous potential as vehicles of persuasion, cooptation, and contribution to the cohesion (or dissension) of a group

identity. Writing an effective narrative requires understanding the perspectives and feelings of one’s audience. Interpersonal negotiation is key to the effectiveness of narratives; skilled narrators create a plausible possible world “through exploiting interpersonal resources that prompt interest, empathy, awe, suspense and ethical discernment in an audience” (Macken–

Horarik et al., 2018, p. 44). In some contexts, a text may have a presumed audience of one — in an academic setting, for instance, in which the audience/instructor’s role is to evaluate the writer’s persuasive abilities (Liu, 2014). In most cases, however, a text represents one instance of an ongoing dialogue between writer and readers (Hunston, 1993) and reflects an awareness of the context available to those readers. Indeed, a text may only make sense given shared knowledge assumed on the part of the readers (Tilakaratna & Mahboob, 2013). Appealing to those readers, using among others the strategies discussed in Chapter 4, allows a writer of narratives to enlist their participation in their goals.

Narrators experienced in adopting their presumed readers’ stances can exploit those perspectives for a wide range of purposes. Political scientists have long recognized, for instance, that members of a political community generally interpret new events or ideas based on their congruence with narratives that those members hold to be true (Polletta & Callahan, 2017;

Shenhav, 2006). The same events can have widely varying consequences depending on how they are interpreted by the public. Some political analysts argue, for instance, that American voters interpreted the release of the Report on the Investigation into Russian Interference in the 2016

Presidential Election (the “Mueller Report”) as either affirmatively exculpating the Trump campaign from having collaborated with foreign actors, or offering evidence that the Trump administration had engaged in obstruction of justice. Support for either narrative can be found in the 768-page report, and more passive news consumers may be unaware of elements of the report

that contradict their preferred narrative (Patja Howell, 2019). This helps explain the decision by

Attorney General William Barr to release a summary of the report rapidly after the announcement of the report’s completion; to the extent that Barr prompted journalists and pundits to frame discussion of the report within the exculpation narrative, his expertise as a narrator aided his political allies.

Expanding the lens from specific political events or parties, telling and retelling narratives can strengthen or weaken the cohesion of groups characterized by a political or ethnic identity. Critical Discouse Analysis (CDA) studies, for instance, have found that “the conversational stories involving minority ethnic groups told by majority-group members characterized (. . .) ‘foreigners’ as culturally and even physically threatening, alien and strange, incomprehensible, irrational, untrustworthy, irreligious and prone to criminality” (Toolan, 2001, p. 234). Such narratives about in-group and out-group distinctions can ramify throughout a discourse community, emphasizing traits as lines of demarcation between socially constructed groups. In short, narrators skilled at identifying and exploiting their audience’s knowledge and attitudes can use narratives not only to inform but also to evoke sympathy, guide the processing of new information, and strengthen or weaken the bonds between groups in a society. By conveying socially accepted moral values and judgments, they offer a means of socializing listeners/readers into a culture. For all these readers, narratives represent a powerful linguistic tool that advanced language learners must understand and use. While the narratives these language learners were asked to write belonged to the fields of fairytale morality and personal experience with friendship, the evaluative language many of them used allowed them to engage in moves of affiliation and persuasion that are often employed on a larger scale.

2.15 Focus on Evaluation

This project focused in particular on the use of evaluation within narratives because of its central role in contributing to the meaning of those texts. Evaluation is a key meaning–making resource for narratives in particular, perhaps “the most important element in addition to the basic narrative clause” (Labov, 1972, p. 366). The parts of a narrative that convey evaluation assert the narrative’s relevance to the reader; evaluation “provides a response to the potential question, ‘So what?’” (Labov, 2008, p. 4) and “creates coherence by providing commentary that makes the text meaningful” (Warren & Winkler, 2016, p. 35). Evaluative language also serves as an indicator that correlates with other measures of advancedness in written texts (Warren &

Winkler, 2016). Evaluative language suffuses even ostensibly neutral texts, allowing (for example) newspaper headline writers to frame a story in a way that either endorses or condemns its participants (Toolan, 2001), and mastery of the type of evaluation appropriate to a particular genre is an important step toward adopting the appropriate authorial voice (Morton & Llinares,

2018).

In the context of narratives, writers can use evaluative language to strengthen the affiliation between themselves and their audiences by endorsing a value system. Even single instances of an evaluation can refer back to a broader ethos (O’Donnell, 2014), and “[t]his value- system in turn is a component of the ideology which lies behind every text” (Hunston &

Thompson, 2000, p. 6–7). These callbacks allow the author to assert values that they and their presumed audience hold in common, invoking solidarity between writer and reader. Indeed,

Oteíza (2017) argues that “appraisal always involves the negotiation of solidarity” (p. 457; emphasis mine). Appeals to solidarity allow authors to drum up support for controversial ideas without necessarily making an explicit case for those ideas, instead asserting commonalities

between the reader and those who support the idea. Zappavigna et al. (2010), writing on the field of restorative justice, illustrate how repeated couplings of ideational and evaluative meanings further the process of affiliation. In this study, the authors describe a prosody of evaluation in which the speaker repeatedly links language involving respect (with either positive or negative polarity) with people and concepts associated with the listener’s community. By linking the speaker’s preferred traits with the listener’s allies, the speaker can assert that they hold these values in common.

As we have seen in Chapter 4, authors can evoke shared values with any of the three

ATTITUDE types, including AFFECT. AFFECT plays a particularly powerful role in political narrative, where readers’ emotional responses to an appeal are often more relevant than logical argumentation (Westen, 2008). Politicians sometimes try to persuade by appealing to what

Martin and White (2005) characterize as antipathy, recognizing that reaction to certain topics with disgust often correlates with conservative leanings; activating these emotions can strengthen a reader’s agreement with socially conservative propositions (Lakoff, 2002). Expressing moral revulsion “is linked to our evaluation of a person as ‘degraded, base, or subhuman’ (2016: 821), based on their socially improper actions” (Benítez–Castro & Hidalgo–Tenorio, 2019, p. 325).

Evaluation also comes into play in the genre of courtroom discourse, where both counsel and judges leaven their remarks with JUDGMENT (Heffer, 2007).

Several of the studies mentioned here used APPRAISAL as an analytical framework in the same way I have used it in this dissertation; while the contexts and stakes of these interactions are different, the evaluative strategies are often similar to those employed in personal narratives.

Mastering evaluative language allows students to navigate these consequential interactions with greater confidence.

In short, learning to use evaluation effectively within narrative genres can prepare students to move their audiences toward socially relevant ends. Prompted by calls for further research into the use of APPRAISAL types and relevance of APPRAISAL sources and targets, I chose to explore what learners of German typically knew how to do when writing in narrative genres, in the hope of better supporting language learners in future years.

CHAPTER 3. METHODOLOGY

The present study aims to elicit patterns of APPRAISAL use among student narrative compositions at three curricular levels. The project consisted of two phases, with data from the first phase informing the development of the second.

3.1 Research Design

This research project consisted of two studies, one an exploratory study that aimed to expose patterns and elicit research questions, one a cross-sectional study of learner texts that aimed to answer research questions sparked by the first study. Data from the two studies were analyzed both quantitatively and qualitatively.

Before going into further detail on those measures, it is important to justify the choice to employ a mix of both quantitative and qualitative methods. While mixed-methods research has emerged as a field of rigorous inquiry and insight, the label ‘mixed-methods’ should not be applied willy-nilly to studies that blend both strands of data without good reason. Mixed- methods practitioners have raised concerns that the label has already been too broadly applied;

Hashemi and Babaii (2013), for example, found in a review of 205 research articles that relatively few recent mixed-methods studies truly integrate these two types of data collection in a well-grounded way.

While definitions of mixed-methods research vary in their emphases and specificity, one criterion of valuable mixed methods research seems to be an a priori justification for mixing the two approaches. Qualitative and quantitative data should be intertwined not only in the outcome of a study, but from its inception onward: “[H]igh-quality mixed methods research requires mixing throughout an entire study: from forming research questions, to sampling, to data

collection, to analysis, and, finally, to interpretation” (Hashemi & Babaii, 2013, p. 829).

Tashakkori and Teddlie (2008) further identify a range of reasons for embarking on a mixed- methods study which a researcher should identify in advance and address with the type of data collected; these motivating factors include completeness, compensation, and diversity. A developmental mixed methods study, for example, is one in which the results of one segment of a research project (e.g., a qualitative interview) give rise to new questions to be probed in the next phase, perhaps with a larger data set (e.g., a quantitative survey).

The present project is perhaps best described in the terms offered by Tashakkori and

Teddlie as a developmental study, in which quantitative results (occurrences of APPRAISAL tokens) from the initial stage (the exploratory study) informed the research questions to be investigated in the second stage (the cross-sectional study), guiding the researcher toward potentially interesting patterns that might otherwise be submerged in an intimidating ream of texts. Other studies to employ a developmental mixed-methods design include Lancaster (2012) and Barletta, Mizuno, & Moss (2013).

Qualitative analysis is also used to contextualize findings of the second part of this research project in order to address completeness, another of the goals set out by Tashakkori and

Teddlie (2008). The provision of completeness offers a particularly compelling reason for integrating qualitative measures in a quantitative study. As Liu (2017) points out, a too-narrow focus on quantitative analysis can blind researchers to the cumulative effect of patterns of evaluative language in context:

To date, much of the literature that explores the language of evaluation and persuasion in media texts takes a quantitative approach. Studies often deploy corpus-based methods to identify one or a limited number of lexical terms or grammatical features seen as indicative of the register, or subcategorisations of types of texts. Quantitative studies of this kind may or may not be accompanied by additional qualitative aspects in which the researcher returns to source texts to explore local co-textual associations. Where corpus-based analyses are restricted to limited features in very local contexts, the scope of vision is significantly

constrained as to what can be revealed about discourse patterns in general, and affiliation strategies in particular (Bednarek 2006, 2010a; Sanz 2011). (p. 204)

In other words, a purely quantitative study can mislead researchers who only dive deeper into their data at points where isolated phenomena ‘jump out’ at them, ignoring the broader sweep of context. Indeed, as I found when analyzing data from the exploratory study, raw frequency counts related to GRADUATION (e.g., the occurrence of GRADUATION via Isolation vs. GRADUATION via Infusion) offered a somewhat misleading impression without the context offered by a qualitative analysis.

The choice of APPRAISAL as an analytical construct also recommends a qualitative element in any analysis of research data. As attested to by the frequent use of APPRAISAL to examine politically or socially fraught questions—e.g., Adendorff & de Klerk (2006), in which the authors analyze, and find morally wanting, the community response to an AIDS outbreak— the APPRAISAL framework aims to categorize linguistic expressions of evaluation that may be deeply subjective, culturally mediated, and/or embedded in prosodic chains of meaning stretching throughout an entire text. The framework’s “focus is on language in its social function

(. . .) and it is based on the analysis of naturally occurring discourse” (Bednarek, 2008, p. 13), and it is therefore not surprising that APPRAISAL categories have blurry boundaries by nature

(Thompson & Alba–Juez, 2014). A quantitative summation of types of APPRAISAL tokens might present a misleadingly cut-and-dried picture of an author’s use of interpersonal resources. Some lexical items may represent instances of APPRECIATION in one context and JUDGMENT in another, and readers may differ on the appropriate APPRAISAL category to use in some cases (as would be expected from a model that encourages the development and refinement of new subcategories).

For quantitative approaches, the indeterminacy of APPRAISAL categories can be ameliorated by allowing for multiple coding of APPRAISAL tokens, which involves labelling a

token as expressing, e.g., both AFFECT and JUDGMENT to some degree. This approach recognizes advanced writers’ ability to “exploit maximally the linguistic resources in response to the demands of a complicated context” (Adendorff & Pienaar, 2013, p. 53). However, a qualitative analysis will likely shed further light on a skilled author’s use of interpersonal resources to convey nuanced evaluations and imply shared understandings. These considerations argue for a mixed-methods approach (or at least not a purely quantitative orientation) when dealing with

APPRAISAL, and the following sections outline how this approach was conceived and carried out in this particular project.

3.1.1 Curricular Context

These learner texts were collected from students in the context of an articulated, genre- based curriculum in which genre is used “as a building block for the construction of curricular trajectories toward advanced literacy” (Ryshina–Pankova, 2016, p. 70). In a curriculum designed around genre-based instruction, teachers and curriculum developers outline a sequence of units based around genres that progress from discourses among intimates (e.g., diary entries, letters to friends and family) to public-facing discourse involving greater social distance (in an interpersonal and not an epidemiological sense; e.g., argumentative speeches, academic summaries). As part of an ongoing process of feedback, reflection, and (re)assessment, instructors select authentic texts representative of particular genres—based on their coherence with the topic of the unit and the presence of lexicogrammatical features that achieve communicative purposes characteristic of those genres—and develop assessment tools such as speaking and writing tasks that allow learners to demonstrate their mastery of content and language knowledge that allow them to communicate effectively within the constraints of those genres. For further detail on this process as it continues to play out in the Georgetown University

German Department, see Section 6.2 and, e.g., Byrnes, 2001; Byrnes, Maxim, & Norris, 2010;

Crane, 2006; Ryshina–Pankova, 2010, 2013; Cunningham, 2018.

At the level of the unit and individual lesson, genre-based pedagogy involves helping students understand the purpose that underlies an author’s choice (from a repertoire of options) of a particular linguistic token; in other words, “promoting literacy development [through] the explicit teaching of the organizing features, functions, and choices of grammar and vocabulary that are available to interpret and produce a variety of specific genres” (Troyan, 2016, pp. 317–

38). This requires an understanding of the patterns common to particular genres; a central task for teachers in genre-based instruction involves “translating the knowledge gained as a result of the genre analysis into pedagogy as a process of creating scaffolds for learners to discover the form-meaning connections and use them in their own production” (Ryshina–Pankova, 2016, p.

71). Rather than describing language (exclusively) in terms of formal categories and rules, instructors can direct students’ attention to lexicogrammatical resources that authors use to achieve particular communicative goals (Pessoa, 2017), organize these linguistic building blocks by field and function in a teaching construct such as a Wortfeld (Crane, 2016), and offer feedback on drafts of student work based its appropriateness to the genre (Byrnes, Maxim, &

Norris, 2010).

3.1.2 Study 1 (Exploratory)

In the first study, written narrative texts by L2 learners of German were collected and coded for APPRAISAL in order to identify patterns of evaluative language use that could be further investigated in a study of a larger set of texts. Participants, who were drawn from second-, third-, and fourth-level undergraduate German courses at Georgetown University, were given 60 minutes to compose a narrative in response to a fairy tale (“Frau Holle”) they had recently read.

100

In order to control for as many confounding factors as possible, subjects were given the same task regardless of language level and given the same amount of time to write their texts (though some did not use the entire allotted time). The narrative task was piloted with a small group of

German-speaking undergraduate students, who provided feedback on the wording of the prompt and its suitability for learners at various levels of language proficiency.

Once collected, these tasks were anonymized and coded for APPRAISAL categories using the modified version of Martin and White’s (2005) model proposed by Bednarek (2008).

Instances of evaluative language that did not seem to fit within this scheme, instances that seemed to require double coding, and otherwise ambiguous cases were flagged to aid in updating the APPRAISAL scheme and coding guidelines to be used in Study 2. Frequency counts for each category and subcategory of ATTITUDE were then tabulated, both within each text and within the narrative stages of each text. Cases in which particular categories seemed unusually frequent were noted, and the concordancer feature of UAM CorpusTool (O’Donnell, 2008) was used to examine these instances in context.

3.1.3 Study 2 (Cross-Sectional)

The second study consisted of an analysis of narrative texts collected from a cross- sectional sample of learners of German. This study employed convenience sampling; learners were graduate or undergraduate students enrolled in first-, second-, and third-year language courses in the GUGD and were contacted via their instructors to request their participation in the study.6 Subjects were given 60 minutes to compose a narrative in which they described a situation when a friend at Georgetown had helped them solve a problem; this narrative task was

6 Results from the two studies would have been more easily comparable if writers had been drawn from the same set of curricular levels; however, I was constrained by the number of instructors and students who volunteered to participate in the studies. In particular, it proved difficult to locate volunteers for Study 1 in Level I courses.

101

selected to draw on the learners’ existing knowledge of content, genre expectations, and lexicogrammar. The major aims of this study included (a) exploring specific research questions raised in the first study; (b) taking advantage of a larger data set in order to elicit patterns of evaluation characteristic of introductory, intermediate, and advanced language learners of

German; and (c) probing the validity and ease of use of an APPRAISAL schema edited for use with

German learner narratives.

The learner texts were quantitatively analyzed as follows:

1. Each text was coded for APPRAISAL tokens according to the sequence outlined in the Procedure section. (Two additional coders coded a total of 20% of the data to assess interrater reliability.)

2. Patterns of APPRAISAL resource occurrence (e.g., instances of AFFECT vs. APPRECIATION

vs. JUDGMENT), in particular those flagged in the initial study, were checked for statistical significance using the chi-square test. 3. A concordancer was used to facilitate the identification of commonly occurring instances of

particular APPRAISAL categories.

I then engaged in a close reading of the learner texts, focusing on patterns suggested by both studies, to explore the various means these learners used to express their attitudes and evaluations, and to draw tentative conclusions about what these texts might be able to teach language instructors.

3.2 Procedure

3.2.1 Study 1

Volunteers for the initial study were solicited via an e-mail forwarded by their instructors.

Participants who indicated an interest in the study were e-mailed the text of the fairy tale and

102

asked to read it independently before meeting for a joint writing session. The fairy tale in question, “Frau Holle,” was modified from a text taken from the Goethe Institut’s “Grimms

Märchen zum Lesen,” a selection of twelve fairy tales expressly chosen and simplified for instructional purposes.7 Two graduate student instructors of German provided input on the text to be used for the pilot study. In order to make the text comprehensible for beginner-level students, translations were provided for words that are less commonly found outside of fairy tales (e.g., die

Spule ‘spool of thread,’ der Kessel ‘cauldron’).

Having read the fairy tale, participants met with the researcher for a writing session lasting 60 minutes, during which they composed a narrative in which they retell the fairy tale from the perspective of “Pechmarie” (the lazy sister), one of the two central characters in Frau

Holle. Participants met as a group in order to keep writing conditions consistent but wrote their narratives individually.

The task prompt (see Appendices) explicitly asked participants to explore their character’s relationships with other characters in the story, to express how their character feels about the events that take place, and to address their audience either explicitly or implicitly.

Participants were encouraged to use their knowledge of the German language to the best of their ability in order to compose an effective narrative. In the hope of forestalling a situation in which students might spend so much time checking their texts for grammatical mistakes that they would not compose a text of meaningful length, participants were also told that grammatical accuracy is not a major focus of the task. Writing sessions were divided into timed “planning,”

“writing,” and “re-reading for final check” phases. The writers were encouraged to use ten

7 http://www.goethe.de/lrn/prj/mlg/mad/gri/de9114028.htm

103

minutes to plan their narrative and ten minutes to review, and the researcher announced when these intervals began and ended.

3.2.2 Study 2

Results from the first study elicited several apparent patterns of APPRAISAL usage that suggested avenues for further inquiry; however, the small sample size of the initial study (seven subjects) made it difficult to draw any conclusions with confidence. This cross-sectional study aimed to draw on a larger number of students in order to explore these patterns.

Data collected from the first study also informed the following aspects of the second study:

APPRAISAL Scheme. In narrative texts characterized by a profusion of APPRECIATION, it was often difficult to objectively differentiate between instances of APPRECIATION: reaction: impact and APPRECIATION: reaction: quality. In the interest of construct validity, I adopted Ngo and Unsworth’s (2015) modification to the schema in which Quality is removed to its own subcategory and Reaction is reconceptualized as simply Impact.

Coding Methodology. Several other ambiguous cases of APPRAISAL use proved difficult to categorize using the initial schema. For these cases, I added additional coding guidelines

(based largely on Thompson & Alba–Juez [2014]) to decrease the subjectivity of categorization in borderline cases.

In Study 2, students were asked to write a personal narrative describing a real-life situation in which they needed assistance and a friend at their university helped them (for reasons behind this choice of field, see the section on Tasks below). The task prompt was piloted with the help of undergraduate students in the GUGD and one native speaker of German (a GUGD graduate student). Feedback from these participants led to minor edits of the language of the

104

prompt. As in Study 1, participants met as a class in order to keep writing conditions consistent

(but wrote their narratives individually).

3.3 Participants

Participants in both studies were students enrolled in an undergraduate German course at

Georgetown University. Curricular levels in the German department comprise either two semesters of a regular course or one semester of an intensive course. In order to control for the level of German experience among each group of participants, students were approached for participation in the study at the end of either the second semester of each level (e.g., semester 2 of Level I) or at the end of a semester of intensive German (which covers an equivalent amount of material during a single semester).

Students were offered a small number of extra credit points in recognition for their assistance with this research project. Details of participants’ previous experience with German and other languages were gathered with a biodata form (see Appendices) similar to the form used to collect student information at the beginning of a semester. The dissertation research was approved by Georgetown’s Institutional Review Board, and all participants signed informed consent forms.

3.3.1 Study 1

Seven L2 learners of German (five males and two females) took part in the first study.

Table 1 shows key demographic information. All were undergraduate students with an age range of 18–22 (average = 19.7). Subjects were enrolled in GUGD courses in Levels II (intermediate),

III (advanced), and IV (Text in Context, the final course in the required sequence). Placement in these courses is determined by the results of a placement exam administered at the start of each

105

semester; students with no prior German experience, or who do not place into Level II, enroll in

Level I (Mozgalina & Ryshina–Pankova, 2015). I assigned participants ID numbers based on the instructor’s name and their curricular level.

Table 1. Participants in Study 1. Participant ID Curricular Level Age L2 ji1 II 21 French (fluent) si1 II 19 French (basic), Hindi (basic) si2 II 22 French (proficient), Turkish (basic) da1 III 19 none da2 III 19 none mt1 IV 20 none mt3 IV 18 Spanish (AP 5) xn1 L1 31 English xn2 L1 n.a. English Note. n.a. = not available.

All were native speakers of English, though three were proficient in at least one Romance language, and two of them also had some basic knowledge of a non-Romance language. The subjects varied widely in their previous experience with the German language and in their level of previous study at Georgetown. One subject had worked in a German-speaking country and had German-speaking relatives; the others’ experience with German-speaking environments was limited to brief leisure or study trips abroad. Participants in the first study also included two native speakers of German (for comparison purposes). The L1 speakers of German were fellow graduate students in the German department. The L1 German speakers were both female, with ages of 31 and 25.

3.3.2 Study 2

Twenty-three L2 learners of German took part in the second study. All were Georgetown undergraduate or graduate students, or students enrolled in Georgetown summer courses.

Subjects had an age range of 18–61 (average = 21.5). Table 2 shows key demographic

106

information. Many subjects were native speakers of English, though several were speakers of a different L1 (e.g., Portuguese, Spanish, Hungarian, Greek, Bengali, Japanese). Several had advanced proficiency (based on self-reporting) in at least one Romance language. The subjects varied widely in their previous experience with the German language and in their level of previous study at Georgetown.

Participants in the second study also included five native speakers of German (for comparison purposes). The L1 German speakers included three males and two females, with an age range of 21–30 (average: 23.8). These participants were recruited from among the teaching staff at Waldsee, the German program of the Concordia Language Villages summer immersion programs.

The sample size of this study remains relatively small compared to many investigations in the field of corpus linguistics. This is, however, not uncommon in APPRAISAL studies, which often involve close analysis of texts and have small sample sizes (Bednarek, 2008).

Table 2. Participants in Study 2. Participant ID Curricular Level Age L1 L2 kb1 I 19 English French kb2 I 20 English/Portuguese kb3 I 18 English kb4 I 18 Spanish English,, French kb6 I 18 English Polish fi1 II 18 English fi2 II 19 English fi3 II 19 Spanish, English fi4 II 27 Hungarian, Greek English, Spanish, Catalan, Sanskrit, Latin, Ancient Greek fi5 II 21 Italian English, French fi6 II 18 English Spanish

107

Participant ID Curricular Level Age L1 L2 fi7 II 18 English fi8 II 20 English Spanish fi9 II 20 English, Bengali fi10 II 19 English fi11 II 20 Spanish, English French, Italian si6 II 19 English, Japanese si7 II 20 English si10 II 20 English si11 II 21 English Spanish ma1 III 19 English Chinese ma2 III 18 Portuguese English, French ma3 III 24 English French ma4 III 19 English French, Spanish wa1 III 18 English wa2 III 61 Spanish English, French wa3 III 21 English Spanish wa4 III 20 Chinese English wa7 III 18 English wa9 III 19 English wa10 III 20 Chinese (Mandarin) Latin wa11 III 19 Chinese (Mandarin) fsa1 III (Summer) 23 English Spanish fsa2 III (Summer) 20 English xn4 L1 23 German English, French, Spanish xn5 L1 21 German xn6 L1 24 German xn7 L1 30 German xn8 L1 21 German English, Italian

108

3.4 Tasks

The genre of narrative was chosen because it was a genre assumed to be familiar to all or almost all of the students, both from their previous schooling and from their study of German at

Georgetown University. The narrative genre is explicitly taught in Levels I and II of the GUGD curriculum in a Level I unit on new year’s celebrations and a Level II unit on fairy tales, so all of the participants in this study should be generally familiar with the typical narrative stages of

Orientation, Complicating Action, Evaluation, and Resolution (Georgetown University, 2020).

The fairy tale in particular was chosen as a common point of reference because it typically contains repeated actions and other features that may be familiar even to beginner students.

Other studies of learner texts have adopted a similar strategy of controlling for experience by choosing a genre familiar to all or almost all students (see, e.g., Myskow & Ono, 2018).

Students in GUGD courses were also expected to be familiar with the nature of genres in general as communicative processes with particular stages and aims (Martin & Rose, 2003).

German courses at Georgetown include explicit instruction on the typical characteristics of several genres; indeed, “each unit in the GUGD is thematically oriented and culminates in a speaking and/or writing task that represents a certain genre—a speech, a fairytale, a recipe, etc.”

(Warren & Winkler, 2015, p. 31). It was therefore expected that students would understand the typical outline of a narrative, and that unfamiliarity with the nature of narratives could be excluded as a confounding factor. It was also assumed that most students would be familiar with the genre of fairy tale, reformulations of which are nearly always ubiquitous in Western mass media (Zipes, 1979). (This could be interpreted as a limitation of this study; it could be argued that a study focusing on texts written within the particular environment of Georgetown would only have implications for instructors at Georgetown. I will address this concern in Chapter 6.)

109

Furthermore, working within the narrative genre offers students an opportunity to deploy a broad array of the meaning-making resources to which they have access. The purpose of this study is to elicit the range of evaluative meaning making of which students at various levels are capable; it was therefore important not to place extraneous constraints on that expression.

Assessment rubrics for language tasks are sometimes constructed in a way that offers information about language learners that is essentially negative in character. If an essay is graded by subtracting points for incorrectly formed lexicogrammatical items, the instructor may learn that certain students have yet to master the rules of word order or noun–adjective agreement specific to German, but they may lack insight into those students’ abilities to tell a compelling story and connect with their audience. In a related way, certain genres demand the use of a constrained set of generic stages and discourse markers; if students were tasked with writing an entry in a technical manual, they would likely restrict their use of interpersonal resources to implicit ones. The narrative genre permits a comparatively wide array of interpersonal meaning- making options, offering an environment within which language learners at a variety of levels can demonstrate what they are able to do.8

The question of which genre to assign is far from an idle one. The success of ongoing curricular renewal in the GUGD is based in part on understanding the importance of a sequenced curriculum in which texts move from primarily personal to primarily public-facing genres.

Students in Level I of the GUGD curriculum compose writing tasks that involve communication with family, friends, and other intimates who share common understandings; at Level II, students explore aspects of German-speaking culture through personal, public, and literary stories; in

Level III, students make the shift to genres such as political speech that involve a more distant

8 I am indebted to Heidi Byrnes for this insight.

110

audience (Georgetown University, 2020; Warren & Winkler, 2015). In each study, I chose a genre that should be accessible to most if not all participants.

3.4.1 Study 1 (Märchen)

In the first study, participants were asked to read the Märchen (fairy tale) “Frau Holle” and re-tell the narrative from the perspective of one of the characters. (The prompt for this task can be found in the Appendices.) The activity of having students respond by assuming the role of a character has been proposed by previous studies as an effective means of stimulating students’ creativity (Urlaub, 2011).

The specific fairy tale “Frau Holle” was chosen for the following reasons:

1. It includes a progression of narrative stages characteristic of the genre and should therefore present few surprises to learners familiar with narratives. 2. The story itself is less well-known than other Grimm fairy tales such as Dornröschen (Sleeping Beauty) or der Froschkönig (the Frog Prince), and therefore students would likely be equally unfamiliar with the specific details. 3. A simplified text of the fairy tale specifically oriented toward German language learners was available from the Goethe Institut’s website.

Another way of putting this is that I intended to control for as many factors as possible and keep the ‘playing field’ as level as feasible, in assigning participants this writing task. By assigning the Frau Holle text, I hoped to offer a text that was comprehensible (both in terms of the predictability of its generic structure and the difficulty of its lexicogrammar) to learners at many levels and also equally unfamiliar to those learners, so that no participant’s text would be impacted by their previous knowledge of the story.

111

3.4.2 Study 2 (Friendship)

In the second study, participants were given a somewhat broader prompt and asked to write a narrative involving a time when a friend helped them solve a problem or recover from a difficult time.

As in Study 1, a guiding consideration in writing the task prompt was to control for as many variables as possible while presenting the participants an opportunity to demonstrate a broad range of their meaning-making abilities. I was therefore interested in assigning a writing task that required participants to draw on knowledge of content and genre that they would already possess, while keeping the topic broad enough that both beginning and advanced students would be able to compose a meaningful text.

In selecting a task for participants of various language levels, it made sense to choose a genre with which most students would already be familiar. A genre such as the academic précis or political appeal would clearly not be appropriate for all participants; beginners would be unfamiliar with the generic stages of these tasks and could not be expected to produce an appropriate response. This consideration prompted me to maintain the narrative genre as the focus of my inquiry.

Students at every level of the GUGD language sequence often return to similar fields

(subject areas) as they move through a sequence of genres. One ideational thread running through many of these genres is the field of friendship, which might be addressed in the context of a personal letter intended for a single recipient, a fairy tale, or an historical narrative meant for a wide audience. From beginning to advanced levels, a student’s ‘need to mean’ may center around evaluating the character and behavior of their friends, either to evoke an emotional

112

response in their reader, convey (dis)approval of an (im)moral action, or make a larger point about political realities.

The actions of a friend might be expressed in a variety of different ways depending on the genre. Students in the GUGD read and analyze model texts before composing their own texts in the same genre; two such model texts in the GUGD curriculum, one at Level I and one in Level

III, illustrate this point. Example (1) is drawn from a Level I text in which a university student deals with an injury.

(1) Meine Kommilitonen aus Münster kümmern sich aber sehr nett um mich. Sie besuchen mich und haben mir auch ganz viele Bücher ins Krankenhaus gebracht. Morgen darf ich auch schon das Krankenhaus verlassen und wieder zu den Vorlesungen gehen – leider nur nicht mehr mit dem Fahrrad… Ich bin sehr froh, dass nichts Schlimmeres passiert ist.

‘But my fellow students in Münster are taking care of me very kindly (+JUDGMENT: propriety). They visit me and also brought me a lot of books in the hospital (+JUDGMENT: propriety). Tomorrow I can leave the hospital and go to lecture again — but unfortunately (- AFFECT: happiness) not by bike yet… I’m very glad (+AFFECT: happiness) that nothing worse happened (+APPRECIATION: impact).’

The description of the author’s classmates’ actions combines both Processes and

Circumstances to provide a consistent positive evaluation of these friends. The author describes the classmates’ behavior explicitly as nett ‘nice/kind’ and describes them engaging in the sorts of acts a reader would be expected to associate with generosity and friendship. This is not an explicit evaluation of the classmates’ moral conduct, but the implication of positive JUDGMENT is fairly clear. The author also uses various explicit tokens of APPRAISAL to indicate her mental state: leider ‘unfortunately’ to express unhappiness at her continued injuries and froh

‘happy/glad’ to counter this sentiment with relief at her relative good fortune.

In this model text for Level I learners, then, the author expresses evaluation of her friends in fairly clear ways. Instances of JUDGMENT and AFFECT are realized congruently, and even

113

when JUDGMENT of the author’s friends is not explicitly stated, the reader is not required to make any large deductive leaps.

A model text at a later curricular level (Level III) offers an example of evaluations on a similar topic realized through linguistically and semantically different means. While this text is also concerned with the semantic field of friendship—even more so than the previous text—the author’s evaluations of his friends are considerably more oblique. This text, Drei Freunde by

Joachim Klecker (Humann, 1991), concerns the reactions of three childhood friends to the construction of the Berlin Wall. One of the three, Eberhardt, imbibes the patriotic fervor of the

East German regime and proudly serves as a border guard; another, Wolfgang, subtly pumps

Eberhardt for information on how to escape to the West. The narrator, while ostensibly less partisan than either, makes his opinion of Eberhardt clear in passages such as Example (2):

(2) Eberhardt, wenn er auf Urlaub war und wir ein Glas Bier zusammen tranken, erzählte gern davon, was er bei der Armee für Heldentaten vollbrachte. Er versäumte keine Gelegenheit darauf zu verweisen, daß er Geheimnisträger sei, prahlte aber andererseits damit, daß Eingeweihte wie er wissen, wie man die Sicherheitsanlagen überwinden könnte.

‘Eberhardt, whenever he was on leave and we drank a glass of beer together, would gladly talk about the heroic deeds (-JUDGMENT: propriety) he accomplished in the army. He lost no opportunity to point out that he was (subj.) (-JUDGMENT: propriety) a keeper of secrets, but boasted (-JUDGMENT: veracity) on the other hand that insiders like him knew how to bypass the security checkpoints.’

Earlier in the text, the author has indicated his dislike for overly nationalistic pride in one’s military service, and so the description of Eberhardt’s actions as Heldentaten ‘heroic deeds’ can be read as expressing sarcasm. We might therefore code this entire description of

Eberhardt as an instance of JUDGMENT: propriety; in the narrator’s eyes, he is not acting as an appropriate public servant should.

The use of the subjunctive form sei casts further doubt on Eberhardt’s status as a dutiful officer. In German, the subjunctive mood is often used when quoting reported speech to distance

114

the narrator from the content of that speech (the subjunctive in English has no equivalent function). The subjunctive does not always indicate doubt on the part of the narrator (its use in newspaper reporting, for example, is a standard feature of the genre and does not necessarily indicate skepticism), but its appearance in this context strengthens the impression of the relationship between the author and Eberhardt as inconsistent or problematic.

The narrator’s choice of Process confirms the impression he has already seeded. The verb prahlen ‘to boast’ typically carries negative connotations; it is braggarts and blowhards speak in this way. One can speak proudly of one’s work without boasting, and the narrator’s choice of this particular verb reinforces the negative appraisal of Eberhardt’s moral character. This phrase further acts as an instance of negative JUDGMENT: veracity; Eberhardt has just insisted that he is a Geheimnisträger ‘keeper of secrets,’ yet in the next breath he shows himself willing to betray those secrets.

In these excerpts, we see that texts appropriate to different curricular levels can evaluate friends’ behavior and the nature of friendship through a variety of linguistic means. I wrote the prompt for Study 2 with this consideration in mind, expecting that students would find the field of friendship comprehensible enough to discuss even at Level I, but broad enough to explore using nuanced language.

3.5 Coding

Texts in both studies were coded for APPRAISAL markers using UAM CorpusTool

(O’Donnell, 2008, 2012, 2014), a piece of software specifically designed to facilitate APPRAISAL analysis. UAM CorpusTool allows for the coding of tokens of any length (e.g., word, sentence, paragraph, entire text) and for both multiple and nested codings (see Figure 12 for a screenshot

115

displaying these functions). The application also includes a concordancer that facilitates viewing each instance of APPRAISAL in context, and it permits searches for one token in the presence of another particular token (e.g., one can find all instances of APPRECIATION: impact that co-occur with positive GRADUATION, or within the Orientation narrative stage).

Figure 12. Screenshot of a Text as Coded with UAM Corpus Tool 3.

Coding APPRAISAL data presents an interesting challenge because of the indeterminacy involved in the entire endeavor of expressing evaluation. As Bednarek (2007) points out, coding

APPRAISAL involves marking both typological distinctions, based on differences in kind (e.g., between ATTITUDE and GRADUATION), and topological distinctions, which reflect differences in degree among instances of the same phenomenon. Coding can be particularly difficult when working with invoked ATTITUDE (Tilakaratna & Mahboob, 2013); appraisal may be expressed explicitly, or it may rely on contextual information available only to a small range of readers. In their proposed update to the APPRAISAL framework, Macken–Horarik and Isaac (2014)

116

emphasize the importance of accounting for the cline of implicitness and recommend that

APPRAISAL studies, arguing that analysts should

make this [indeterminacy] a feature of the account rather than something to be pushed to the margins (as an embarrassment in the analysis). This means making space for the dance of implicit and explicit evaluation, for the different co-textual and contextual frames within which evaluation occurs and for the situation — and culture-specific nature of the analytical enterprise. (p. 79)

Rigorous APPRAISAL coding can also be complicated by the presence of evaluative prosody; coding schemes must “take account of both local and global patterns of meaning in texts” (Macken–Horarik & Isaac, 2014, p. 83). These global patterns include what Macken–

Horarik (2003) calls metarelations; throughout an entire text, patterns of evaluation may shift from stage to stage, resulting in shifts in meaning. A text may also contain multiple voices; in narratives, for example, different characters may be distinguished by differing patterns of evaluation common to each. A diligent APPRAISAL analysis should take these semantic patterns into account.

One means of doing so is to code not only the type of evaluation made via a given

APPRAISAL token, but also its source and/or target. It is useful to know whether the narrator is describing a state of affairs as desirable, or whether that evaluation is put in the mouth of another character (who may or may not be viewed as a reliable source). Determining the source of an evaluation is crucial (especially in the context of academic writing) in order to determine the evaluative key of a passage (Barletta, Mizuno, & Moss, 2013; Swain, 2013), allowing the analyst to assess the narrator’s desire to convey subjectivity or impartiality and to identify interest groups with whom the narrator might be trying to align. Novice academic writers, for example, may attribute evaluative language primarily to recognized authorities, in order to add weight to their arguments.

117

Keeping track of the targets of evaluative language can also be valuable in order to make a principled distinction between ATTITUDE subtypes (Coffin, 2006). Distinguishing between

APPRECIATION and JUDGMENT, for example, generally requires identifying the target of the

APPRAISAL; JUDGMENT is applied human targets, while APPRECIATION assesses nonhuman objects, phenomena, and behaviors.

Based on some of these considerations, Macken–Horarik & Isaac (2014) proposed a detailed model (pp. 87–88) for coding linguistic data based on APPRAISAL categories in a way that accounts for implicitness and context. I expanded on this sequence as follows. based on additional recommendations from Thompson (2014) and Lee (2015):

(1) Divide each text into narrative stages (orientation, complicating action, evaluation, resolution, coda). (2) Divide each text into clauses with reference to Eggins (2004). (3) Identify tokens of APPRAISAL, beginning with explicit choices and moving toward more implicit tokens. Note the targets and sources of each APPRAISAL token. Tokens can be individual words or whole phrases and may be embedded within other tokens (Thompson, 2014a). (4) Assess the overall impact of the passage. (5) Map patterns of APPRAISAL items from one stage of the text compared to other other stages. (6) Code each token as an instance of ATTITUDE and/or GRADUATION. (7) Identify instances when double coding would be useful and add that second coding as described in, e.g., Thompson (2014a). (In borderline cases of APPRECIATION vs. JUDGMENT, ‘trust the text’ based on whether a person or a thing is evaluated) (8) Assess the degree to which particular characters, figures or viewpoints map onto patterns of APPRAISAL choices. (9) Compare the text with others at the same curricular level, asking questions such as the following: (a) Which APPRAISAL choices are most common at each level? (b) Which APPRAISAL choices are most common at each narrative stage? (c) Do certain characters’ voices display particular patterns of APPRAISAL? (10) Compare the text with others at other curricular levels, asking questions such as the following: (a) Which APPRAISAL resources are most commonly used at each level? (b) Does the balance of explicit vs. implicit APPRAISAL demonstrate a statistically significant trend from level to level? (c) Do the lexicogrammatical realizations of APPRAISAL choices demonstrate a statistically significant trend?

The order of these steps is not random; in this sequence, instances of ATTITUDE are identified early on but not categorized until the overall impact of the passage has been taken into

118

account. Other scholars of APPRAISAL might recommend a modified order of operations;

Bednarek (2009a), for example, suggests

classifying attitudinal lexis first, and then looking at the pattern in which it is used, and what the effect of this usage is. This is not to deny the importance of the pattern as meaning- making (Hunston 2003); in fact it actually helps us to explore this aspect of meaning creation in terms of appraisal further. That is, rather than giving precedence to either lexis or pattern, it might sometimes be necessary to discuss both aspects of attitudinal appraisal. (p. 184)

The order of operations I have chosen, however, should help the reader make coding choices in borderline cases. In situations when a linguistic item could be coded as instantiating one APPRAISAL subcategory or the other, it often makes sense to consider the cumulative effect of nearby APPRAISAL tokens. See, for instance, the JUDGMENT of Senator Amy Klobuchar in the

New York Times excerpt in Example (3):

(3) …outwardly, Ms. Klobuchar is the embodiment of “Minnesota nice” — polite and intent on being able to “disagree without being disagreeable,” as she wrote in her 2015 memoir, “The Senator Next Door.” In an era of Twitter rants and senatorial showboats, she is the worker bee in the background, tallying up how many of her bills get signed into law: 24, she said, since Mr. Trump became president.9

The descriptor worker bee could be interpreted in at least two ways: as an indicator that the subject lacks creativity or interest, or that the person described is industrious and deserving. The context of the passage makes it clear that the journalist intends the latter meaning.

3.5.1 General Coding Principles

In addition to the specific coding disambiguation rules I outline below, I adopted the general coding principle that misspelled words and infelicitous forms in general should be coded as though they were spelled and formed correctly, as long as the intent of the writer is clear, following the example of Myskow and Ono (2018).

9 https://www.nytimes.com/2018/11/26/us/politics/amy-klobuchar-2020-election.html

119

3.5.2 Identifying Coding Units

Within applied linguistics, the choice of a unit of analysis is often crucial. If one is interested in the occurrence of errors in students’ verbal contributions to class discussion, should one focus on the errors per word, per clause, or per T-unit (among many other options)? If the researcher counts only the number of units containing errors, the choice of measurement unit is enormously important; most L2 learners will commit some kind of error in the course of an entire T-unit, but this does not mean they are as error-prone as the novice learner who has not yet learned the rules of subject–verb agreement. Along similar lines, Polio (2017) argues that the use of clauses per T-unit as a measure of complexity often led researchers to a mistaken impression of complexity in written work—which, as noted by Ortega (2015) and Ryshina–Pankova (2015), often repackages the complexity of dependent clauses within a nominalization or prenominal adjectives. Within APPRAISAL studies, however, there seems to be a less strict emphasis on selecting tokens of uniform length. In Examples (4) through (8) selected from Martin & White,

2005, APPRAISAL tokens may be realized by single adjectives (4), an adjective with a modifier

(5), a verbal process that describes emotion-laden behavior (6), similes that evoke implicit understandings (7), or potentially entire chains of clauses (8).

(4) “word about these remarkable paintings began to spread” (p. 64)

(5) “He came running fighting mad” (p. 61)

(6) “Mother’s tears were falling down” (p. 61)

(7) “they fenced us in like sheep” (p. 65)

(8) “For they are feeling as vulnerable as a man who has already had his arm torn off by a lion, and sits in the corner holding his stump and waiting for the lion to finish eating and come

120

for him again.” (p. 65)

The guiding principle here (if one can be said to exist) seems to be that APPRAISAL tokens can be of any length but should be indivisible; that is, if the token is expanded or shortened, its meaning would change or become incoherent. That said, nested APPRAISAL certainly exists (see section 3.5.4.1).

3.5.3 APPRAISAL Categories

I used the schemes displayed in Figures 13 through 15 to sort evaluative tokens into

APPRAISAL categories. These figures, generated by the UAM CorpusTool software, display the system networks I used to code texts for ATTITUDE and GRADUATION. Each category branching diagonally from a node indicates that a coding decision will be made for that category. For example, for a token coded as AFFECT will also be coded for AFFECT-TYPE (choosing among the options ‘authorial’ and ‘non-authorial’) as well as AFFECT-TYPE-2 (un/happiness, dis/satisfaction, in/security, dis/inclination, or surprise) and TRIGGER (realis or irrealis). In

Chapter 5, I describe the updates I made to the ATTITUDE scheme for Study 2.

121

Figure 13. ATTITUDE Scheme for Study 1. (Generated by UAM CorpusTool).

122

Figure 14. ATTITUDE Scheme for Study 2. (Generated by UAM CorpusTool).

123

126

Figure 15. GRADUATION Scheme for Studies 1 and 2. (Generated by UAM CorpusTool).

3.5.4 Resolving Coding Ambiguities

3.5.4.1 Multiple Coding and Multiple Layering

When tokens of APPRAISAL have been properly isolated, it can still be difficult to categorize them in a principled way.

In ambivalent cases, a means of double-coding APPRAISAL tokens may be necessary

(Adendorff & de Klerk, 2006). Some tokens may be difficult to sort into one category or another, making it justifiable to count them as evoking both. In a slightly different case, one token may be read equally reasonably as expressing the meanings of two APPRAISAL categories (or subcategories) simultaneously. Bednarek (2007), who has outlined these ambiguities in perhaps the greatest detail, refers to them collectively as examples of polyphony. Polyphony occurs when one lexical item “either blends or fuses different kinds of Appraisals” (p. 111). Bednarek’s categories of polyphony, which can be placed on a cline of greater to lesser ambiguity, include

Invoked Appraisals and Border Phenomena (in which a token expresses perhaps one, perhaps the other category), as well as Fused Appraisals and Appraisal Blends (in which a token expresses both).

Invoked Appraisals. In Invoked Appraisals, lexical items “‘inscribe’ meaning from one

Appraisal system (e.g. Judgement) and ‘invoke’ meaning from another Appraisal system

(e.g. Appreciation)” (Bednarek, 2007, p. 111). One or the other of these meanings may ‘pop out’ from the text depending on the semantic lens with which it is being read. Barletta, Mizuno, and

Moss (2013), for example, cite an excerpt describing students engaged in cooperative, praiseworthy learning behavior. Due to their laudable actions, students are appraised in terms of positive JUDGMENT; at the same time, the teaching method that fostered this behavior is implicitly evaluated via positive APPRECIATION. Depending on the thematic emphasis within the

125

rest of the passage (is the text primarily concerned with describing student behavior, or with modelling effective pedagogy?), one or the other of these voices may receive a preferential reading.

Border Phenomena. In borderline cases, the categorization of an APPRAISAL token becomes subjective; an analyst’s coding choice could plausibly go either way. In example (7) below, Bednarek (2007) illustrates ways in which the description of an emotion can fall more or less clearly into the categories of AFFECT and APPRECIATION: reaction: impact (or

APPRECIATION: impact in my taxonomy, as adapted from Ngo & Unsworth [2015]) . The last example can safely be labeled as AFFECT, with the experiencer of the emotion so clearly foregrounded; the first is an example of APPRECIATION, since the object of the evaluative phrase is a phenomenon (‘that’) and not a person (whoever may be feeling the surprise, or lack thereof, now goes unmentioned). The intermediate cases, however, are more difficult to pin down, as

Bednarek (2007) illustrates in Example (7):

(7) that’s not surprising I don’t find that surprising that is not a surprise that is not a surprise to me that doesn’t surprise me I’m not surprised (pp. 121–122)

For this particular type of border phenomenon, Bednarek (2007) proposed the new APPRAISAL category of COVERT AFFECT (in which an emotional reaction is emphasized and the subject of that reaction is clearly present but is either not mentioned or is deemphasized). I coded examples such as the following as expressing COVERT AFFECT, since they included, but did not foreground, an identifiable source of an emotion. Examples (8) through (13) display instances of

COVERT AFFECT drawn from Level III and L1 texts in Study 2.

126

(8) Leider musste ich “Ja” sagen, weil ich einen Computer brauche ‘Unfortunately I had to say “yes,” because I need a computer’

(9) Diese Semester ist fuer mich schwerer als der erste ‘This semester is a lot more difficult for me than the first’

(10) Für mich war diese Situation sehr schwer ‘This situation was very hard for me’

(11) Mein zweites Fach Englisch macht mir sehr viel Spaß ‘My second subject, English, is a lot of fun for me’

(12) Das macht mich sehr traurig ‘That makes me very sad’

(13) Das hat mich sehr beruhigt, denn in Mathe hatte ich immer das Gefühl, dass alle anderen Studenten schon viel mehr wissen als ich ‘That calmed me down a lot, because in math I always have the feeling that all other students already know more than I do’

Border Phenomena present particular challenges for the APPRAISAL coder, who must find a way to sort them into one category or another without making arbitrary, case-by-case decisions, which might be tempting when coding a sentence such as Example (14):

(14) Sixty-four years ago, the U.S. Senate censured the bullying demagogue Sen. Joseph R. McCarthy of Wisconsin for conduct that ‘tended to bring the Senate into dishonor and disrepute.’10

In the first part of this sentence, Sen. McCarthy is clearly the object of negative

JUDGMENT (bullying demagogue), but the second half of the sentence is ambiguous. Should the

Post be seen as assessing a negative APPRECIATION of the senator’s conduct (conduct being a phenomenon, not a person) or as providing a further example of its negative JUDGMENT of

McCarthy the public figure?

At some point, categorization of Border Phenomena inevitably becomes a judgment call

(pun not intended) on the part of the reader. The more obvious coding choice may depend on the

10 https://www.washingtonpost.com/opinions/the-case-for-censuring-trump/2018/07/18/13f226ce-8acc-11e8-a345- a1bf7847b375_story.html?utm_term=.012d2e220105.

127

position of the reader; e.g., how they interpret ‘old’ (positive JUDGMENT, indicating wisdom, or negative APPRECIATION, indicating a state of decay?) may depend on their own cultural understandings (Rothery & Stenglin, 2000). For the purposes of my study, I adopted a protocol that hopefully leads to high interrater reliability, even if an outside rater might reasonably quibble with the choices. Example 10 above, for instance, would be coded as including one instance of JUDGMENT (of McCarthy the person) and one of APPRECIATION (since his conduct is a nonhuman phenomenon).

Whatever specific coding choice is made in this situation, it is worth noting that the description of McCarthy’s actions, in which the author nominalizes his actions to be appreciated or implicitly judged, opens up a broader range of meaning-making possibilities and is characteristic of greater advancedness (Derewianka, 2007). By framing McCarthy’s behavior in this way, the author makes an assessment not only of this individual senator, but also of the conduct he exemplifies, and takes an implicit stance on how a good senator ‘should’ behave.

Coding for Border Phenomena and Invoked Appraisal can also serve to highlight examples like this, in which the author makes a statement on behavior in terms of morality or ethics generally

(O’Donnell, 2014).

Fused Appraisals. Fused Appraisals are less ambiguous but may be just as difficult to code. In these cases, “a lexical item can ‘infuse’ meanings from simultaneous typological systems such as Attitude, Graduation and Engagement” (Bednarek, 2007). It is not difficult to call examples of Fused Appraisals to mind; these essentially include what Martin and White

(2005) refer to under the category of GRADUATION via Infusion, in which an evaluation is

128

intensified by means of a more heavily weighted lexical item; as in the example of a brilliant player rather than a good one (Martin & White, 2005) or as in Example (15):

(15) Am nächsten Tag wurde Frau Holle sehr wütend und schimpfte mit mir ‘The next day, Frau Holle became quite furious and cursed at me’

Identifying Fused Appraisals appears significantly easier than accounting for them in a principled, systematic way in a coding scheme. Properly coding for instances of Fused

Appraisals would require, according to Bednarek (2007), a “topological classification of

Appraisal items, showing how they differ in intensity with respect to each other” (p. 114).

Bednarek’s 2008 book Emotion Talk Across Corpora essentially takes on this classification project for the case of AFFECT in English.

Appraisal Blends. At a further layer of delicacy, Bednarek (2007) labels the conflation of meanings within a (sub)system as Appraisal Blends. She identifies several cases in which

Appraisal tokens clearly belong within either APPRECIATION, AFFECT, or JUDGMENT but call up multiple subtypes of meaning within those categories. In other words, Fused Appraisals have to do with polyphony in terms of APPRAISAL subtypes (ATTITUDE, GRADUATION, and

ENGAGEMENT), while Appraisal Blends include conflations within an APPRAISAL type (or subtype).

The students described in Example (16) below, for example, are simultaneously evaluated as being skilled in reaching their goals (+JUDGMENT: capacity) and as being less than honest (-JUDGMENT: veracity):

(16) Or perhaps in Slytherin You’ll make your real friends, Those cunning folk use any means To achieve their ends. (Rowling, 1998)

129

The previous examples are relatively difficult to identify but relatively easy to code in

UAM Corpus Tool once categorized. An additional wrinkle of complexity emerges when a text contains instances of not only multiple coding, but also multi-layering, in which one APPRAISAL token is nested within a longer coding unit that expresses a meaning belonging to another

APPRAISAL category. This point is illustrated in Examples (17) through (19):

(17) Das Arbeit war so schwer, dass meine Dienst mit die Frau Holle machte kein Spass ‘The work was so difficult (-APPRECIATION: impact) that my service with Frau Holle was no fun (-AFFECT: un/happiness)’

(18) Schließlich wollte meine Mutter ja nur, dass auch ich endlich etwas Glück im Leben fand ‘In the end, my mother only wished (+AFFECT: dis/inclination; invoked +JUDGMENT: propriety) for me to find a little happiness (+AFFECT: un/happiness) in life as well’

(19) Mein Verdacht, dass sich Aschenputtel nur vor der Arbeit drücken wollte und hoffte, dass das, was sie Schönheit nannte, ihr den Weg im Leben erleichtern würde, wurde bestätigt, als das Mädchen am Abend nicht nach Hause kam. ‘My suspicion that the Cinderella just wanted to dodge work (-JUDGMENT tenacity), and hoped (+AFFECT: dis/inclination) that what she called beauty (-APPRECIATION: quality) would ease her life, was confirmed (+JUDGMENT capacity) that evening when the girl didn’t come home.’

As sketched out in Figure 16, a lengthy token of ATTITUDE (in this case, JUDGMENT) may contain nested within it evocations of other ATTITUDE categories and subcategories. In this case, the narrator includes evaluations of her stepsister’s character and emotional state in order to make a positive assessment of her own capacity as a judge of character. Each of these APPRAISAL tokens should be coded separately. (As mentioned earlier, the ability to do this is one of the major selling points of the UAM Corpus Tool software.) In order to more easily identify instances of this kind, in which multiple evaluations interweave to create a more complex harmony of evaluative language, I also added a ‘nested’ code to mark as ‘yes’ or ‘no’ for each

APPRAISAL token (e.g., ‘is this APPRAISAL token nested within another such token?’). This made

130

the overall coding process somewhat more arduous, but I suspected it would help me flag informative patterns of language use.

Figure 16. Nested layers of APPRAISAL in Text xn1.

3.5.4.2 Border Phenomena

Taxidermizing every conceivable expression of evaluative language is neither feasible, nor is it the goal of this coding scheme. As Bednarek (2009b) points out, some level of ambiguity is inevitable:

Ambiguities may still arise in the classification of attitudinal lexis, since “categories of semantic analysis cannot be specified in such clear-cut terms as can grammatical categories” (Eggins & Slade 1997: 139–140). Perhaps the very fine distinctions that are made by appraisal theory (e.g. affect: un/happiness, in/security etc; appreciation: reaction: impact, reaction: quality, composition: balance, composition: complexity etc; judgement: normality, capacity, tenacity etc) can be included in less specific attitudinal meanings in a first step, following the advice given by Macken-Horarik: “If our analysis is to be sensitive to the shaping force of logogenesis, it cannot (should not) be too dense or overdetermining” (Macken-Horarik 2003: 318). Similarly, types of attitudinal targets or attitudinal assessments

131

might not be clearly distinguished from each other and a taxonomy of such attitude types is clearly needed. (pp. 184–185)

Keeping these limitations in mind, I adopted guidelines for resolving ambiguities in the hopes of reducing subjectivity on the part of the rater.

For Study 1, I adopted a coding approach from Thompson and Alba–Juez’s (2014) overview of work on evaluation; when difficulties arose because of ambiguities that went unaddressed by this approach, or because of phenomena particular to the German language, I added additional coding guidelines for Study 2.

Thompson (2014a) takes the principle of “trust the text” as his lodestar. In borderline cases between APPRECIATION and JUDGMENT, for example, Thompson categorizes an APPRAISAL token as one or the other based on whether a person (APPRECIATION) or a thing (JUDGMENT) is evaluated: “[T]he wording should be taken as the basis for the initial assignment of categories.

This frequently comes down to taking the Target at face value” (Thompson, 2014b, p. 58).

Coffin (2006) relies on a similar principle when assessing whether an evaluative token counts as an instance of APPRECIATION: social valuation; what matters is “Who is appraised?” Thompson’s

(2014) criteria are organized visually in Figure 17.

Figure 17. Criteria for Distinguishing JUDGMENT and APPRECIATION. (Thompson, 2014).

132

Therefore, to distinguish between APPRECIATION and JUDGMENT, I considered the Trigger of the evaluation. If it is clearly a person being appraised, it should be coded as JUDGMENT. If it is clearly a non-human being appraised: APPRECIATION.

The potentially confusing Examples (20)–(22) illustrate the rule. In each, because the target being appraised is an event or behavior, the author is coded as expressing APPRECIATION: social valuation (closely related to JUDGMENT).

(20) Es war nicht recht (-APPRECIATION: social valuation)! Meine gemeine, schlechte Schwester bekam Geld und ich, ein nett, intelligent, Jungfrau, bekam Pech! ‘It was not right! My mean, bad sister received money and I, a kind, intelligent maiden, received pitch!’

(21) Das heißt, dass es nicht sicher (-APPRECIATION) mit einem *Radfahren zu fahren ist, wenn es regnet. ‘That means it isn’t safe to bike when it rains.’

(22) Aber das war falsch (-APPRECIATION: social valuation), weil jede andere Studenten viele über Islam und den Arabisch-welt gekennt haben ‘But that was incorrect, because all the other students knew a lot about Islam and the Arab world!’

In this case, ‘das’ refers to a previously made statement rather than a person. An exception to the personhood rule would be a purely aesthetic description of a person, though even physical descriptions can often be laden with implicit ethical evaluation, as in Example

(23):

(23) Wie ich letztes Feiern gesagt habe, hatte ich eine sehr, sehr dreckig Mitbewohnerin ‘As I said during the last vacation, I had a very, very dirty roommate’

In this case, the narrator expressed negative JUDGMENT: capacity of their former roommate as someone who could not keep her space clean, emphasized by GRADUATION via isolation and repetition (sehr, sehr ‘very, very’).

133

3.5.4.3 Appraisal Blends: Pure and Blended AFFECT

Love is many-splendored — not only in song, but also in its evaluative manifestations.

Bednarek (2007) points out that the verb to love can evoke a range of APPRAISAL-related meanings, not all of which reflect only the emotional state of the one who loves. She identifies two principal meanings, “distinguished by the appraised entity or attitudinal target (Trigger)”: (a)

“to have very strong feelings of affection for sb” or (b) “to like or enjoy sth very much”

(pp. 128–129). Bednarek argues for a double coding of such instances of AFFECT (as blended

AFFECT / APPRECIATION or blended AFFECT / JUDGMENT) based on the trigger of the emotion in question (see Figure 18 for a list of potential linguistic realizations of these Fused Appraisals).

Figure 18. Triggers of Blended AFFECT. (Bednarek, 2007).

However, Bednarek (2007) glosses somewhat over potential differences in meaning between authorial and non-authorial AFFECT in blended situations — that is, whether the emotion is attributed to a text’s narrator or to another human actor. Putting an evaluation in another character’s voice does open up additional meaning-making potential for the author, as Bednarek notes in her discussion of Fused Appraisals. For example, attributing an emotion of love to an acknowledged authority might involve ENGAGEMENT as much as AFFECT; if an advertisement

134

proclaims that a rock star loves a new beverage, the point is to convey approval of the beverage, not merely to comment on the rock star’s emotional state.

Sometimes the expression of AFFECT can turn the lens of evaluation back on the emoter as well. If the emoter is described as loving something that is universally admired or despised, the emoter is often judged based on their identification with that thing. Judgments on capacity or tenacity can also be implied through an affective process like love, as in Example (24):

(24) Where is Tim right now, as his body lies on the couch? Are his soul and self still here, in the room, or have they already slipped away? He has put up a valiant, almost incredible fight for his life, but by now he has lost that fight. When he was a little boy, he used to make people laugh because he got so frustrated with board games; he didn’t like playing those games, with their rules and tricks. He loved sports, and running, and playing with his friends at the beach.11

In this example, potential tokens of Affect (he didn’t like playing those games; he loved sports) function less to describe the subject’s emotional state than to give the reader a sense of his character; Tim, the author argues, was a determined, fair-minded, possibly stolid young man whose death the reader is meant to grieve.

To sort between these situations when coding, I tentatively adopted the following coding guidelines to be combined with a contextual reading:

• Emotion on the part of the subject; Trigger refers to an aesthetic quality: blended AFFECT

/ APPRECIATION of Trigger

• Emotion on the part of the subject; Trigger refers to a person or moral behavior: blended

AFFECT / JUDGMENT of Trigger

11 https://www.theatlantic.com/magazine/archive/2017/11/a-death-at-penn-state/540657/.

135

• Emotion on the part of another character; Trigger refers to an aesthetic quality:

blended AFFECT / APPRECIATION of Trigger

• Emotion on the part of another character; Trigger refers to a widely

praised/condemned moral behavior: blended AFFECT / JUDGMENT of Emoter

Examples (25) through (27) demonstrate various cases of these Appraisal Blends found in learner texts in Levels II and III:

(25) Diese Treffen mit die NAIMUN gruppe war sehr laut und schwer. Ich mag diese schwierig Diskutieren nicht und ich war müde und gestresst. ‘This meeting with the NAIMUN group was very loud (-APPRECIATION: valuation) and difficult (-APPRECIATION: impact). I don’t like (Blended -AFFECT: dis/satisfaction / -APPRECIATION: impact) these difficult (-APPRECIATION: impact) discussions, and I was tired (-AFFECT: un/happiness) and stressed (-AFFECT: in/security).’

(26) Sie würden das Stadt liebe, es ist aufgeregt, und es hat vielen Aktivitäten ‘You would love (Blended +AFFECT: un/happiness / +APPRECIATION: quality) the city; it is exciting (+APPRECIATION: impact) and has many activities (+APPRECIATION: valuation)’

(27) Jetzt bin ich sehr traurig weil ich vermisse John ‘Now I am very sad (-AFFECT: un/happiness), because I miss (Blended -AFFECT: un/happiness / +JUDGMENT: propriety)’

3.5.5 Interrater Reliability

The challenge of ensuring the validity of categorizing linguistic data with such blurry boundaries has sparked a number of different approaches to coding. In Liu (2014), the author coded their data twice at different times, which allowed for refinements in the author’s coding criteria. I followed a similar approach in editing the APPRAISAL scheme I used between Study 1 and Study 2 in light of the difficulties I had in coding some categories.

Coding all the data alone, however, leaves any analysis open to the charge of subjectivity.

In order to establish interrater reliability, two additional coders familiar with the APPRAISAL framework coded a total of 20% of the texts as well. I first discussed with the coders the specific

136

APPRAISAL scheme to be used and changes that had been made from Martin and White’s (2005) original model. Texts were given to the coders with tokens already selected, and the coders were invited to categorize each token, with the option of selecting multiple categories for any given token. This approach was chosen in order to minimize the ambiguity of what constituted agreement; in other words, to avoid a situation in which the two coders agreed that a certain phrase should be coded as APPRECIATION but disagreed on the precise length of that phrase.

3.6 Data Analysis

Once the writing tasks had been coded according to the guidelines sketched out above, I used UAM CorpusTool to calculate frequency counts for instances of each ATTITUDE category and subcategory (per 1000 words) for each curricular level and each curricular text, as well as the distribution of inscribed (explicit) and invoked (implicit) APPRAISAL within each level and text.

Results from Study 1 suggested potentially significant differences in the occurrence of certain ATTITUDE subcategories, including the following potential trends: (a) greater reliance on

APPRECIATION or JUDGMENT within the Orientation and Complicating Action narrative stages;

(b) greater frequency of non-authorial AFFECT in higher-level texts; and (c) greater frequency of

GRADUATION: intensification by means of infusion, rather than isolation, in higher-level texts.

The sample size of Study 1 was too small to draw any firm conclusions from these patterns.

I therefore used the data set from Study 2 to check whether the pattern found in Study 1 of relative frequency distributions of these APPRAISAL tokens was substantiated in Study 2, with a larger sample size, and statistically robust. To this effect, I submitted the frequencies to chi-square test analyses and checked for statistical significance in any differences across distributions. (Chi-

137

square tests were not run in every instance, as they cannot be used given zero instances of a certain phenomenon.) A concordancer was used to facilitate the identification of commonly occurring instances of particular APPRAISAL categories. I then used these instances, as well as statistically significant differences in APPRAISAL use, as a starting point for a qualitative analysis of the different ways in which students at different levels used these resources.

3.7 Research Questions for Study 1

The following research questions guided the initial exploratory study.

RQ1. What types of APPRAISAL resources did the L2 German writers use, and how frequent was each type?

RQ2. What affordances does the UAM CorpusTool software offer for the coding of

APPRAISAL in L2 German writing?

3.8 Research Questions for Study 2

Patterns of APPRAISAL use were further analyzed in Study 2 to address the following research questions:

RQ1. What types of APPRAISAL resources (e.g., types of AFFECT, APPRECIATION, and

JUDGMENT) are most commonly used at each curricular level?

RQ2. What types of APPRAISAL resources are most common in each narrative stage?

RQ3. How does the prevalence of inscribed (or explicit) ATTITUDE, as compared to invoked (or inscribed) ATTITUDE, differ across texts and curricular levels?

RQ4. Do certain voices within each text display particular patterns of APPRAISAL?

138

CHAPTER 4. RESULTS

In this chapter, I report descriptive statistics for both studies and instances of differences across curricular levels in the frequency of usage of particular ATTITUDE types, as well as patterns of ATTITUDE use across narrative stages. I also flag patterns to be explored in greater detail in Chapter 5.

The UAM Corpus Tool (UAMCT), which I used to code texts from both studies, allows users to generate tables of frequency counts for each ATTITUDE type, both in terms of raw frequencies, mean frequencies, and frequency per 1,000 words (O’Donnell, 2014). O’Donnell

(2014) recommends avoiding the frequency per 1,000 words statistic “except as a general measure of the evaluative ‘warmth’ of the writer” (p. 103), since a writer who uses, e.g.,

JUDGMENT relatively often but uses less evaluation generally might have a similar JUDGMENT frequency per 1,000 words to a writer who uses more evaluative language overall but uses each

ATTITUDE type equally often.

I used five layers to code these texts within UAMCT: (1) Appraisal, to code individual tokens for ATTITUDE types and subtypes, as well as explicitness, polarity, polyphony, and nesting; (2) Graduation, to code individual tokens for GRADUATION; (3) Stage, to code larger sections of text as belonging to particular narrative stages (i.e., Orientation, Complicating

Action, Evaluation, and Resolution); (4) Curr_Level, to code entire texts based on the curricular level from which they were solicited; and (5) POS, an automated layer that labeled each word based on its part of speech.12 I used separate layers in order to more easily compare ATTITUDE

12 I am indebted to Michael O’Donnell for his comprehensive, patient e-mail guidance in setting up these layers.

139

and GRADUATION usage within stages and across curricular levels, which is possible through the

UAMCT search feature using queries such as the following:

in in

This query yields a list of instances in which Level II writers express AFFECT in an orientation stage, as shown in Figure 19.

Figure 19. UAM CorpusTool Search Function.

The analyst can then click on individual tokens to examine them in context.

4.1 Study 1

Study 1 contained nine texts, including three from Level II (Intermediate) of the GUGD curriculum, two from Level III (Advanced), and two from Level IV (Text and Context), as well as two texts written by L1 speakers for comparison purposes.

The three Level II texts contained a total of 887 words (for an average of 295.67 per text), which included 120 instances of ATTITUDE. These included 32 instances of AFFECT, 68 of

JUDGMENT, and 20 of APPRECIATION. JUDGMENT was the Attitude type most often used in Level

II texts, followed by AFFECT and APPRECIATION. Table 3 provides the raw frequencies of words and ATTITUDE tokens, as well as tokens per 1000 words, at each curricular level.

140

The two Level III texts averaged 560.5 words in length (for a total of 1121 words). These texts contained 33 AFFECT tokens, 51 JUDGMENT tokens, and 61 APPRECIATION tokens.

APPRECIATION tokens were relatively more prominent among these texts than in Level II, comprising 42.07% of the total.

Level IV texts contained 683 total words, for an average of 341.5 over the two texts.

These texts included 93 instances of ATTITUDE, comprising 12 tokens of AFFECT, 46 of

JUDGMENT, and 35 of APPRECIATION. Level IV writers expressed relatively little AFFECT

(12.90% of ATTITUDE instances) than those in Levels II and III; one Level IV text expressed just a single instance of AFFECT.

L1 speakers wrote a total of 1973 words (averaging 986.5 per text), including 245 tokens of ATTITUDE. Their texts contained 55 instances of AFFECT, 94 of JUDGMENT, and 96 of

APPRECIATION. The two texts contained strikingly different distributions of ATTITUDE types. In one text, JUDGMENT accounted for 51.59% of total ATTITUDE tokens, with APPRECIATION at

23.02%; in the other, APPRECIATION accounted for the lion’s share at 56.30% of ATTITUDE instances, with JUDGMENT at 24.37%.

Table 3. Frequency Counts of ATTITUDE Types across Curricular Levels in Study 1. Attitude Type Level II Level III Level IV L1 N % N % N % N % Affect 32 27.1% 32 22.2% 12 12.9% 54 22.1% Judgment 67 56.8% 51 35.4% 46 49.5% 94 38.5% Appreciation 19 16.1% 61 42.4% 35 37.6% 96 39.3% Total 118 144 93 244

Comparing global patterns of ATTITUDE across curricular levels suggests that Level II texts contain a lower proportion of APPRECIATION tokens than would be expected from a random

141

sample. Comparing Levels II and III, a chi-square test of independence indeed showed a significant difference by curricular level for ATTITUDE, χ2(2, N = 264) = 21.17, p < .01. Other divergences in overall ATTITUDE frequency, however, did not show statistical significance; a chi- square test showed no significant difference between curricular Levels III and IV in their frequency of ATTITUDE, χ2(2, N = 237) = 5.68, p = .058. Given the small sample size, the possibility cannot be discarded that the lack of significance is simply due to statistical power.

Several of the learner texts drew relatively little from a particular APPRAISAL category.

One Level II text, for instance, contained only 3 instances of APPRECIATION (out of 33 total

ATTITUDE instances); one Level IV text contained just one instance of AFFECT (out of 35 total).

As has been noted, the sample size of this group of texts is too small to draw sweeping conclusions on this point; in general, however, it seems on closer inspection that many of the learner writers selected a particular attitudinal axis along which to describe the characters in their fable, and drew much of their evaluative language from that field. In Text si1, for example, the author reimagines the hardworking sister as someone too unintelligent to get out of doing work; accordingly, the author relies to a significant extent on the resource of JUDGMENT (and specifically JUDGMENT: capacity). I speculated that a larger sample of texts might allow me to assess whether texts at higher curricular levels would contain a more balanced distribution of

ATTITUDE resources.

4.1.1 AFFECT among Study 1 Texts

The texts were coded for AFFECT in three ways. I recorded the subtype of each instance

(un/happiness, dis/satisfaction, in/security, dis/inclination, or surprise), then noted whether the emotion belonged to the author or another party (authorial or non-authorial evaluation) and

142

whether the trigger of the emotion corresponded to a real event or a possible happening (realis or irrealis).

Non-authorial evaluation ranged from 8.33% of AFFECT instances (in Level IV) to

27.27% (in L1 texts). The proportion of non-authorial evaluation did not differ significantly by level, χ2(3, N = 132) = 2.12, p = .55. Variance based on the realis/irrealis dichotomy was likewise not statistically significant, χ2(3, N = 132) = .69, p = .87. Examining each text individually, I noted that one L1 speaker text contained a higher proportion of non-authorial evaluation (12 instances out of 32 AFFECT tokens, or 37.5%) than any other single text or group of texts at a curricular level. (The other L1 speaker text included just three such instances.) I wondered whether, in a larger sample, texts written by L1 speakers and more advanced language learners would include a higher proportion of non-authorial evaluation, which might reflect their ability to include multiple perspectives in their narrative. This question prompted me to track both the trigger and source of AFFECT in more granular detail in Study 2.

Relative frequency of AFFECT types varied between learner and L1 speaker texts. Among the learner texts, AFFECT: un/happiness and AFFECT: in/security were the most common types of

AFFECT used. L1 speaker texts included relatively more instances of dis/inclination than the learner texts. A chi-square test comparing Levels II and III, as well as L1 speaker texts, showed a statistically significant difference in the relative frequency of use of AFFECT type by curricular levels II and III, χ2(6, N = 118) = 20.64, p < .01. (Level IV could not be included in this comparison because these texts included zero instances of AFFECT: dis/satisfaction.) In general, writers at lower proficiency levels seemed to rely more heavily on the happiness/unhappiness distinction in order to convey the narrator’s impression of events, while L1 writers seemed

143

comfortable drawing on a wider variety of emotions. I will explore the effect of these differing approaches to AFFECT when I return to this topic in the Analysis chapter.

AFFECT was typically realized at lower levels by either verbs or adjectives. Within Level

II texts, adjectives typically dominated the tokens of AFFECT: un/happiness in particular, as in

Examples (1) and (2).

(1) sie war sehr wütend und traurig ‘she was very furious and sad’

(2) Ich war so glücklich; ich wurde Gold bekommen ‘I was so happy; I would receive gold’

However, there were two instances of AFFECT: un/happiness realized via verbal processes, shown in Examples (3) and (4).

(3) Ich habe den ganzen Tag geweint ‘I cried for the whole day’

(4) Ich lebte meine Mutter ‘I [loved] my mother’

Verbal phrases were more common for realizations of AFFECT: dis/inclination or in/security, such as in Examples (5)–(7).

(5) Ich wollte nur das Gold ‘I only wanted the gold’

(6) Meine Mutter und ich, ohne sie, hatten nicht so viel für essen ‘My mother and I did not have so much to eat without her’

Level II texts included only one instance of AFFECT realized by a noun:

(7) Wenn meine Schwester sagte, dass sie traurig und müde war und Heimweh hatte… ‘[When] my sister said she was sad and tired and homesick…’

Within the next-highest proficiency level, writers relied much more heavily on verbs in order to realize expressions of AFFECT. Level III texts included relatively fewer AFFECT tokens

144

realized by adjectives or adverbs. Examples (8) through (11) display realizations of un/happiness and dis/satisfaction characteristic of the level.

(8) Wir waren ein bisschen traurig aber ‘We were a bit sad (-AFFECT: un/happiness) but’

(9) …die immer ein traurig und langweiliges Geschwister war ‘who was always a sad (-AFFECT: un/happiness) and boring sibling’

(10) weil ich schon wusste, dass dieser Unfall Mutti zornig machen würde ‘because I already knew that this accident would make Mother angry’ (-AFFECT: dis/satisfaction)

(11) Sie ist böse über die Arbeit. ‘She is angry (-AFFECT: dis/satisfaction) about the work.’

Nouns also occurred less often as realizations of AFFECT at Level III; Examples (12)–(14) represent the only such instances.

(12) und hatte allgemein keine Sorgen ‘and in general had no worries (+AFFECT: in/security)’

(13) nur weil hatte ich Angst von die Alte ‘I was afraid (-AFFECT: in/security) of the old one’

(14) sagte dass sie Heimweh hatte ‘said she was homesick (-AFFECT: un/happiness)’

Rather, Level III writers used a wide array of verbs to reflect their emotions. Most

AFFECT tokens—especially tokens of AFFECT: un/happiness—were realized in single verbs, such as in Examples (15)–(18).

(15) Dieses Spiel gefiel mir ‘I liked (+AFFECT: un/happiness) this game’

(16) Ich dachte nur an das Gold und hörte sie nicht ‘I thought only of the gold and did not hear (-AFFECT: dis/inclination) her’

(17) ich mag ihr nicht ‘I don’t like (-AFFECT: un/happiness) her’

145

(18) Deshalb lächelte ich ‘For that reason I smiled (+AFFECT: un/happiness)’

A few, however, consisted of longer phrases that referenced other evaluatively laden terms. These phrases employ nesting, in which tokens of one ATTITUDE system are included within a longer token realizing another instance ATTITUDE. In Example (19) below, the narrator’s

AFFECT (in this case, surprise) is sparked by behavior that is negatively appraised via JUDGMENT.

(19) Ich konnte ihren Unmut und ihre Undankbarkeit nicht glauben ‘I could not believe (-AFFECT: surprise) her cowardice (-JUDGMENT: tenacity) and thanklessness (-JUDGMENT: propriety)’

Finally, Level III texts included a few instances in which instances of AFFECT were realized by entire subordinate clauses, such as in Examples (20)–(22). These typically involved the use of a subordinating conjunction (dass ‘that’ or bis ‘until’), which requires greater control over word order on the part of the writer.

(20) Ich dachte dass ich in eine schlechte Situation war ‘I thought I was in a bad situation (-AFFECT: in/security)’

(21) Ich putzte mich und putze mich, bis Blut aus meinen Fingern spritzte. ‘I washed and washed, until blood sprayed out of my fingers (-AFFECT: in/security).’

(22) Endlich kann sie ihre Arbeit wieder machen! dachte ich. ‘Finally she can do her work again! (+AFFECT: un/happiness) I thought.’

Level IV texts included fewer instances of AFFECT overall, but they seemed to continue a trend in which AFFECT was instantiated less often by adjectives or nouns and more often by verbs or entire clauses. This set included just two instances each of realization by adjectives and nouns. Realizations such as those in Examples (23)–(24) were more common.

(23) …hatte die Tochter viele Hoffnung, das sie was gutes für ihre Familie tun konnte

‘…the daughter had much hope (+AFFECT: dis/inclination), that she could do something good for the family’

146

(24) So hatte die Familie mehr Platz, mehr Essen, und müssten nicht dauernd das Kreischen und Weinen der Stieftochter hören

‘So the family had more space, more food, and did not constantly have to hear the stepdaughter’s screeching and crying (+AFFECT: in/security)’

Here again, nesting allows the writer to connect her internal state directly to the behavior of others.

4.1.2 JUDGMENT among Study 1 Texts

Texts were coded for JUDGMENT type (normality, capacity, tenacity, propriety, or veracity) and also for the target of each JUDGMENT. The narratives used JUDGMENT to evaluate either the narrator herself, her sister, their mother, or Frau Holle (the other central character in the narrative). There were instances of each JUDGMENT type at each curricular level, with

JUDGMENT: propriety the most common overall (as might be expected in a fairy tale that stakes out a position on the appropriate way to behave in situations involving diligence and respect).

Texts at higher levels included a relatively higher incidence of JUDGMENT: normality. Overall, the L1 speaker texts seemed to distinguish themselves from the learner texts in that they employed a more balanced use of JUDGMENT resources. Across the L1 speaker texts, four of the five JUDGMENT types account for between 20.2% and 25.5% of total JUDGMENT instances (the exception is JUDGMENT: veracity, which is seldom used in these texts). By contrast, some learners seem to rely on one or two particular types to the exclusion of others; JUDGMENT: tenacity comprises only 7.8% of JUDGMENT tokens among Level III texts, and JUDGMENT: normality registers low percentages at Levels II (13.4%) and III (9.8%). Table 4 displays frequency counts of each type of JUDGMENT in Study 1 texts.

JUDGMENT, like AFFECT, was typically realized at lower levels by either verbs or adjectives. Level II writers used a number of pejorative adjectives such as böse ‘mean,’ dumm

147

‘dumb,’ faul ‘lazy,’ or töricht ‘foolish’ to condemn either of the sisters, and less often adjectives of praise such as fleißig ‘hard-working’ or freundlich ‘friendly.’ Examples (25) through (27) below provide a characteristic sample:

(25) Aber mit der Familie die sie schützte war sie sehr böse ‘but with the family that protected her she was mean (-JUDGMENT: propriety)’

(26) Weil sie so schmutzig und töricht war… Because she was so dirty and foolish (-JUDGMENT: propriety)…

(27) Alle im Dorf haben gedenkt, dass meine Schwester Perfekt war ‘Everyone in the village thought that my sister was perfect (+JUDGMENT: propriety)’

Writers at this level also describe a number of praise- or condemnation-worthy actions, such as in Examples (28) through (30).

(28) Obwohl wir haben nicht so viel für essen, hatte meine Mutter meine Schwester im Haus begrüßte ‘Although we did not have so much to eat, my mother welcomed (+JUDGMENT: propriety) my sister in the house’

(29) Diese Frau benutzte meine Schwester wie ein Sklave! ‘This woman (-JUDGMENT: propriety) used my sister like a slave!’

(30) Ich versuchte und versuchte zu arbeiten ‘I tried and tried (+JUDGMENT: tenacity) to work‘

Table 4. JUDGMENT among Study 1 Texts. Judgment Level II Level III Level IV L1 Type N % N % N % N % normality 9 13.4% 5 9.8% 13 28.3% 19 20.2% capacity 20 29.9 11 21.6 4 8.7% 23 24.5% tenacity 18 26.9 4 7.8 10 21.7% 22 23.4% propriety 19 28.4 20 39.2 16 34.8% 24 25.5% veracity 1 1.5 4 7.8 3 6.5% 6 6.4% unclear 0 0.0 7 13.7 0 0.0% 0 0.0%

148

Modal verbs were more common among tokens of JUDGMENT than among tokens of

AFFECT. Level II writers often used können ‘to be able to’ to assess characters’ tenacity or capacity; in a few cases, they also used mögen ‘to like’ or wollen ‘to want’ in order to implicitly praise or condemn a character, as in Examples (31)–(33). In Example (32), for instance, the narrator accuses her sister of not wanting to share her bounty, and of therefore displaying greed and ungratefulness.

(31) Weil ich immer sehr krank war, konnte ich nicht wie sie arbeiten. ‘Because I was always very sick, I could not work (-JUDGMENT: capacity) like she did.’

(32) …meine Schwester mochte nicht das Geld mit uns teilen ‘…my sister did not want to share (-JUDGMENT: propriety) the money with us’

(33) Meine Mutter wollte, dass ich auch Gold bekomme ‘My mother wanted me to get gold too (+JUDGMENT: propriety)’

JUDGMENT was realized by nouns relatively less often at this level; the only clear instance of JUDGMENT expressed by a noun is Aschenputtel, the demeaning equivalent of Cinderella, in

Example (34).

(34) Meine Familie brachte ein bisschen mehr Geld, weil wir unserer Aschenputtel für vielen Tagen nicht hatten. ‘My family needed a bit more money, because we did not have our Cinderella (-JUDGMENT: normality) for many days.’

There are also several references to Goldmarie as a Stiefschwester ‘stepsister.’ This could be an instance of negative Judgment: normality (e.g., that she is only a stepsister and does not belong to the ethical family), but in the cases at this Level, it is used as a label without any implication that it is meant to be derogatory.

Level II included a few instances in which an adverbial phrase realizes JUDGMENT; in

Example (35), the frequency of Goldmarie’s efforts indicates her solid work ethic.

149

(35) setzte jeden Tag in der Nähe des Brunnens und spann ‘sat every day near the fountain and spun (+JUDGMENT: tenacity)’

Level III texts included a wide array of adjectives as realizations of JUDGMENT; however, compared to the previous level there were relatively more realizations in the form of nouns.

Adjectival realizations of JUDGMENT included instances in Examples (36) and (37).

(36) Sie ist gemein und dumm. ‘Sie is mean (-JUDGMENT: propriety) and stupid (JUDGMENT: capacity)’

(37) aber sie war eine schlechte Schwester ‘but she was a bad (-JUDGMENT: propriety) sister’

Verbal realizations in this level typically appraised not the person performing an action, but that action’s object; in Example (38), the narrator casts doubt on her sister’s veracity by describing her mother’s disbelief of her extraordinary tale.

(38) Meine Mutter hat es auch nicht geglaubt. ‘My mother didn’t believe it either’ (-JUDGMENT: veracity)

Nominalizations such as Faulheit ‘laziness’ in Example (39) began to appear at this level, with Level III writers describing a character’s trait as a concept that could be explicitly commented upon.

(39) ich konnte nur traurig an meiner Faulheit denken ‘I could only think sadly about my laziness’ (-JUDGMENT: tenacity)

Level III narrators also used evaluatively laden words for kinds of people in order to convey their disdain for Frau Holle, such as in Examples (40) and (41).

(40) “Boese Hexe!” schrie ich ‘ “Evil Witch!” (-JUDGMENT: normality) I screamed’

(41) Dann hatte ich die Alte gesehen. Sie war hässlich und ich mag ihr nicht. ‘Then I saw the old one (-JUDGMENT: normality). She was ugly (-JUDGMENT: normality), and I don‘t like her.’

150

This level included one instance in which JUDGMENT was realized by a verbal phrase with subordinate clause. In Example (42), the narrator implicitly asserts her own capacity by describing a situation she was able to predict.

(42) Ich lächelte, weil ich schon wusste, dass dieser Unfall Mutti zornig machen würde ‘I smiled, because I already knew (+JUDGMENT: capacity) that this accident would make Mother angry’

The clearest pattern to emerge from JUDGMENT data involved the target of JUDGMENT tokens. I initially opted to code for JUDGMENT target based on the recommendation of Don

(2016), who argues that “a writer’s choice as to whom or what s/he targets for attitudinal assessment can be just as significant regarding the nature of the textual persona being constructed as the type/sub-type of ATTITUDE being advanced” (p. 6). The texts in this study bore out that observation. The narrator in each of the texts evaluated the narrator herself relatively often and the characters’ mother least often. At higher language levels, the narrator seemed to bring the resources of JUDGMENT to bear more often on the narrator’s sister (the ‘good’ sister, commonly referred to as ‘Goldmarie’). All texts except those from Level II evaluated the ‘good’ sister using JUDGMENT more often than any other character. A chi-square test comparing all four levels showed a statistically significant difference by curricular level in JUDGMENT target, χ2(9, N

= 257) = 20.69, p < .05. Level II writers did evaluate the sister with explicit JUDGMENT, as in

Examples (43) and (44), but they evaluated the narrator herself nearly as often.

(43) Sie war dumm ‘She was stupid’ (-JUDGMENT: capacity)

(44) Mein Stiefschwester ist nicht schlau! ‘My stepsister is not clever!’ (-JUDGMENT: capacity)

The pattern of JUDGMENT targets suggests that texts in Level II generally set up an opposition between the two sisters, excusing or qualifying the behavior of one while criticizing

151

the conduct of the other. In texts at higher curricular levels, and in L1 texts, other actors in the drama come in for more censure or praise, allowing the narrator to critique the overall situation in which she lives, and which incentivizes her lazy and greedy conduct.

4.1.3 APPRECIATION among Study 1 Texts

Each group of texts employed a wide array of APPRECIATION resources. Across each level, APPRECIATION: social valuation was the most frequently used type. This is unsurprising given the genre; we would expect a fairy tale to contain expressions that evaluate an occurrence according to its reflection of social norms. Table 5 shows frequency counts of APPRECIATION types in Study 1 texts.

Table 5. APPRECIATION in Study 1 Texts. Appreciation Level II Level III Level IV L1 Type N % N % N % N % reaction 1 5.3% 11 18.0% 7 20.0% 7 7.3% composition 7 36.8% 17 27.9% 11 31.4% 22 22.9% social 11 57.9% 33 54.1% 17 48.6% 67 69.8% valuation Total 19 61 35 96

Most texts also used JUDGMENT in order to evaluate characters as more or less ethical;

APPRECIATION: social valuation, by contrast, is used to evaluate situations, occurrences, or behaviors for which the human source is not explicit. In Example (45), for instance, the author evaluates the treatment of the narrator as unfair without specifying what person treated her that way (though we know it to be Frau Holle). Based on the coding procedure described in the

152

previous chapter, this would be categorized as APPRECIATION because of the lack of a human target.

(45) Ich bin nämlich die ungerecht-behandelte Tochter ‘I am namely the unfairly treated daughter’

APPRECIATION was realized lexically in a variety of ways, to a greater degree than

AFFECT or JUDGMENT. The examples below were characteristic of APPRECIATION: social valuation within these texts. Realization by adjectives with positive or negative polarity was relatively common, as in Examples (46)–(48).

(46) einem Abend allerdings passierte ihr entgegen ihrer Perfektion etwas ungeschicktes ‘one evening, something awkward happened to her despite her perfection’

(47) Es ist ein sehr nett, glücklich, und froh leben ‘It is a very nice, happy, and happy life’

(48) Das war auch keine so schlechte Sache ‘That was also no very bad thing’

Within L1 texts, APPRECIATION: social valuation was realized in a greater variety of ways, such as idiomatic phrases—Example (49)—or rhetorical questions — Examples (50)–(51).

(49) …hätte früh aufstehen müssen. Das geht natürlich nicht! ‘…would have had to get up early. That doesn’t work, of course!’

(50) …an das Gold, aber war es das wirklich wert, so viel dafür zu schuften? Immerhin bin ich viel zu privilegiert, um so harte Arbeite zu tun! ‘…about the money, but was it really worth it to slave away so much for it? Anyway, I am much too privileged to do such hard work!’

(51) Warum bekommen manche alles und die meisten nichts? ‘Why do some get everything while most get nothing?’

The frequency of APPRECIATION types did not differ significantly by level, χ2(6, N = 212)

= 10.09, p = .12.

In coding these texts, I found it challenging to distinguish between APPRECIATION: reaction: impact and reaction: quality; I encountered numerous border cases in which the

153

distinction between the two seemed arbitrary. For this reason, I modified the coding scheme for

Study 2, incorporating the differentiation between impact and quality proposed by Ngo and

Unsworth (2015).

In general, the patterns of ATTITUDE usage within this exploratory study can be described in terms of breadth, with texts at higher levels exploring a greater range of ATTITUDE resources, both in terms of the number of different subtypes used (particularly in the case of AFFECT), the range of targets evaluated (in the case of JUDGMENT), and the lexicogrammatical realizations employed (in the case of APPRECIATION). In the Discussion and Implications chapters, I explore a few of these texts in greater detail and speculate on how some of them could be used in instructional contexts to guide L2 writers to develop their mastery of resources for evaluation.

4.1.4 Explicitness in ATTITUDE among Study 1 Texts

Texts across each curricular level demonstrate similar patterns of inscribed as opposed to invoked ATTITUDE. At each level, inscribed (or explicit) ATTITUDE comprises between 56.99%–

60.69% of all ATTITUDE, with invoked ATTITUDE comprising 39.31%–43.01%. Only texts from

L1 speakers contain any instances of provoked ATTITUDE (2 instances). Almost every text contained more inscribed than invoked ATTITUDE, though variability was greater among Level II texts. One Level II text contained 41 instances of inscribed ATTITUDE compared to only 15 of invoked ATTITUDE; the other Level II texts actually contained slightly more instances of invoked

ATTITUDE. There was no significant difference across curricular levels in the prevalance of explicitness, χ2(3, N = 100) = 0.45, p = .93.

154

The L1 speaker texts seemed to employ a broader range of implicitness when using invoked ATTITUDE; the ‘flagged’ subtype accounted for 14% of invoked ATTITUDE in L1 speaker texts, compared to 0%, 3.57%, and 2.5% in Levels II, III, and IV respectively.

At a further level of delicacy, implicitness of APPRECIATION resources varied somewhat from level to level. Somewhat counterintuitively, Level II texts contained more invoked (55% of instances) than inscribed APPRECIATION, while L1 speaker texts contained more instances of inscribed APPRECIATION (60.4%). I initially suspected that texts at higher curricular levels would demonstrate a higher incidence of implicit evaluation, given their command of a larger array of linguistic resources. An examination at a greater level of delicacy might reveal differences in how more or less proficiency writers used more or less implicit language. However, in coding this study I found it difficult to distinguish subtypes of invoked ATTITUDE (flagged vs. afforded) from provoked ATTITUDE using the coding scheme I had adopted. I therefore clarified the definitions of each that I used for Study 2.

Table 6. Explicitness in Study 1. Explicitness Level II Level III Level IV L1 N % N % N % N % inscribed 69 58.5% 88 61.1% 53 57.0% 143 58.6% invoked 49 41.5% 56 38.9% 40 43.0% 99 40.6% provoked 0 0.0% 0 9.0% 0 0.0% 2 0.8% Total 118 144 93 244

4.1.5 ATTITUDE within Generic Stages

The distribution of ATTITUDE types seems to vary depending on the generic stage under examination. Though the sample size of this first study is too small to make any sweeping conclusions about whether this pattern holds or what it might mean, preliminary results from this

155

study clarified a research question to be investigated in Study 2. Table 7 displays frequency counts of ATTITUDE tokens at each narrative stage in Study 1 texts.

ATTITUDE within Orientation. Learner texts seemed oriented more strongly toward

JUDGMENT in the Orientation stage. In most of the texts from Levels II–IV, JUDGMENT accounted for the majority of ATTITUDE tokens within the Orientation stage.

Table 7. Raw Frequencies of ATTITUDE Tokens at each Narrative Stage. Level II Level III ORI* COM RES COD ORI COM RES COD

AFFECT 2 21 6 2 10 12 9 1 JUDGMENT 13 36 15 1 13 12 25 1 APPRECIATION 4 13 3 0 14 29 17 1

Level IV L1 ORI COM RES COD ORI COM RES COD

AFFECT 2 5 4 1 10 38 5 2 JUDGMENT 7 27 12 0 12 73 8 1 APPRECIATION 0 18 12 2 10 55 19 12 *ORI = Orientation; COM = Complicating Action; RES = Resolution; COD = Coda.

Within this excerpt from Text si2 in Level II, the author sets up an opposition between the narrator and the sister by ascribing opposing JUDGMENT tokens to each.

Vor langer zeit habe ich mit meiner Schwester gelebt. Sie war sehr schön und fleißig. Aber war ich faul. Weil ich immer sehr krank war, konnte ich nicht wie sie arbeiten. Alle in Dorf haben gedenkt, dass meine Schwester Perfekt war. Meine Mutter wusste die Wahrheit, dass ich war immer zu Krank und konnte nicht arbeiten. Die Familie war sehr arm, weil es gibt ein Teuerung von Brot im Land. Obwohl wir haben nicht so viel für essen, hatte meine Mutter meine Schwester im Haus begrüßte. Weil meine Mutter musste Brot für mich und meine Schwester geben, musste meine Schwester arbeiten um essen zu verdienen.

‘A long time ago, I lived with my sister. She was very beautiful (+APPRECIATION: composition) and hard-working (+JUDGMENT: tenacity). But I was lazy (-JUDGMENT: tenacity). Because I was always very sick (-AFFECT: insecurity), I could not work like she did (-JUDGMENT: capacity). Everyone in the village thought that my sister was perfect (+JUDGMENT: propriety). My mother knew the truth, that I was always too sick

156

(-AFFECT: insecurity) and could not work (-JUDGMENT: capacity). The family was very poor (-APPRECIATION: social valuation), because there is a price increase on bread in the land. Even though we do not have so much to eat (-APPRECIATION: social valuation), my mother had welcomed my sister in the house (+JUDGMENT: propriety). Because my mother had to give bread for me and my sister, my sister had to work to earn food (+JUDGMENT: propriety).’

The ‘good’ sister is evaluated with positive JUDGMENT: tenacity and JUDGMENT: propriety in terms of her ability to work as willingness to earn money for the family. Judgments of the hard-working sister are qualified, in one case by contextualizing the sister’s positive actions in the frame of the family’s need (“because my mother had to…”) and in another by using ENGAGEMENT to place a positive judgment in the voice of others (“everyone in the village thought…”). (This is an instance of an ENGAGEMENT: attribution move, which opens up dialogic space and leaves room for other perspectives.) The narrator also uses JUDGMENT to evaluate herself, this time JUDGMENT: capacity. This evaluation is also qualified via AFFECT with explanations of the narrator’s ill health (“I was always very sick”). APPRECIATION is used sparingly to describe the characters’ physical characteristics and economic situation. In effect, this passage evaluates one sister as hard-working (though by necessity) and one as less able (for reasons beyond her control). The evaluative heavy lifting of this text’s orientation stage is performed by tokens of JUDGMENT, with AFFECT and APPRECIATION playing a clarifying role.

The Orientation stages of L1 speaker texts tend to feature a greater variety of interwoven

ATTITUDE types, as in this excerpt from Text xn2.

Eigentlich begann der heutige Tag, wie alle anderen Tage auch: Meine Mutter und ich saßen zuhause am Tisch und genossen den Tag, während meine Stiefschwester ihre Spinnarbeiten verrichtete und dadurch für unser leibliches Wohlbefinden sorgte. Dabei war das Spinnen keine leichte Arbeit, denn sie sponn meist soviel dass ihr am Abend die Hände bluteten. Allerdings war uns das egal. Meine Mutter, sie war schließlich meine leibliche Mutter, bevorzugte mich und machte mir das Leben so einfach und unanstrengend, wie möglich. So wie es sich gehört. Sie war schließlich nur ihre Stieftochter. Dazu muss ich anmerken, dass sie - entgegen mir - von besonderer Schönheit war. Warum sollte sie dann nicht auf andere Art und Weise etwas schlechtes im Leben erfahren? So hatte ich auch wenig Mitleid mit ihr, wenn sie abends von der Arbeit nach Hause kam und wenigstens ihre Hände verunstaltet

157

waren und ich mich selbst nicht mehr so schlecht und hässlich fühlte.

‘Today actually began like all other days (+APPRECIATION: complexity): My mother and I were sitting at home and enjoying (+AFFECT: satisfaction) the day, while my stepsister carried out her spinning work and thereby saw to our physical well-being. Spinning was no easy work (-APPRECIATION: social valuation), for she usually spun so much that her hands bled in the evening (+JUDGMENT: tenacity). But that did not matter to us (-AFFECT: disinclination). My mother — she was, after all, my biological mother (+JUDGMENT: normality) — preferred me (+AFFECT: happiness) and made my life as easy (+APPRECIATION: social valuation) and relaxing (+APPRECIATION: social valuation) as possible. As it should be (+JUDGMENT: propriety). After all, she was only her stepdaughter (- JUDGMENT: normality). I must admit that she — compared to me — was of great beauty (+APPRECIATION: composition). So why shouldn’t she then experience something bad in her life (-APPRECIATION: social valuation) in another way (+APPRECIATION: social valuation)? So I had little sympathy for her (+AFFECT: happiness), when she came home from work and at least her hands were blemished (-APPRECIATION: composition), and I didn’t feel quite as bad (-AFFECT: happiness) and ugly (-APPRECIATION: composition) myself.’

The narrator sets the stage with tokens of positive APPRECIATION and AFFECT conveying the narrator’s and mother’s emotional state. The sister is briefly introduced with descriptions of her difficult life situation and evocative images of her hard work. The narrator then returns to the narrator and her mother, describing their mutual affection (using AFFECT) and justifying the mother’s one-sided generosity with reference to the daughter’s status as a full-blooded child

(using JUDGMENT: normality). Finally, the narrator uses APPRECIATION to further describe her sister and justify her treatment: The sister is uncommonly beautiful (+APPRECIATION: composition), and it is therefore socially appropriate (+APPRECIATION: social valuation) when she experiences unhappiness (-APPRECIATION: social valuation) in other aspects of her life.

The orientation closes with further details of the narrator’s emotional state. Throughout this stage, the narrator uses all three ATTITUDE types to introduce the three characters: AFFECT to describe the narrator’s impression of her current situation, JUDGMENT of the sister’s ability and the mother’s generosity, and APPRECIATION to compare the sisters’ physical appearances and cast approval on the tenacious sister’s lot in life.

158

ATTITUDE within Complicating Action. The picture is more mixed within the complicating action stage. Level II writers tend to continue to rely on JUDGMENT (with 50–57.14% of instances of ATTITUDE within the complicating action stage), while Level III writers employed a greater proportion of APPRECIATION (54.55% and 55% of the ATTITUDE instances). The two

Level IV texts show a trend back toward reliance on JUDGMENT (55.56% and 53.12%).

ATTITUDE within Resolution. The Resolution stages evince a clear pattern of increasing

APPRECIATION use at higher levels. APPRECIATION made up 13% of ATTITUDE tokens within

Resolution at Level II and increased by at least ten percentage points with each stage, to a majority of all tokens (59%) within Resolution among L1 texts. In this stage, Level II writers used APPRECIATION in order to describe the narrator’s material condition after her encounter with

Frau Holle; she [hat] kein Gold bekommen ‘received no gold’ in contrast to her sister, who was mit Gold bedeckt ‘covered in gold.’ More advanced writers include several similar descriptions of physical condition and beauty, but also broaden their usage of APPRECIATION to include evaluations of the overall situation in general, as in Examples (52) and (53) from Level III:

(52) ich bekam Pech! Es war schrecklich und ich war sehr schmutzig ‘I received pitch (-APPRECIATION: social-valuation: consequence)! It was awful (- APPRECIATION: reaction: impact) and I was very dirty (-APPRECIATION: composition: balance)’

(53) sie hat mein gut Leben geendet ‘she ended my good (+APPRECIATION: social-valuation: consequence) life’

In Level IV texts, the trend toward using APPRECIATION in the Resolution stage for authorial comments on the narrator’s situation becomes even more pronounced. These authors sometimes use APPRECIATION to indicate a reaction of surprise, as in Example (54), or to implicitly evaluate others’ behavior, as in Example (55).

(54) komischerweise haben die Witwe und ihre wirkliche Tochter die Stieftochter eine lange Weile nicht mehr gesehen

159

‘strangely, the widow and her real daughter did not see the stepdaughter again for a while’

(55) so hatte die Familie mehr Platz, mehr Essen und müssten nicht dauernd das Kreischen und Weinen der Stieftochter hören ‘so the family had more room and food and didn’t always have to hear the stepdaughter’s screeching and crying’

In general, within the Resolution stage we can see the development of the learner’s ability to use descriptive terms not only to help the reader envision a scene, but also to convey the author’s mood and opinions.

4.1.6 GRADUATION among Study 1 Texts

FORCE: Intensification and Quantification. Across all levels, texts in the first study included a higher proportion of FORCE than FOCUS among tokens of GRADUATION. All writers also preferred intensification to quantification. Table 8 displays frequency counts of

GRADUATION instances in Study 1 texts.

Table 8. GRADUATION among Study 1 Texts. Graduation Level II Level III Level IV L1 Type N % N % N % N % Force 40 90.9% 24 88.9% 24 80.0% 109 84.5% Focus 4 9.1% 3 11.1% 6 20.0% 20 14.5% Total 44 27 30 129

L1 writers employed FORCE: intensification by intensifying a process (rather than an attribute) relatively more often than learners, though the variance does not appear statistically significant (χ2(3, N = 161) = 5.3276, p = .15). More advanced writers (Level IV and L1) did differ from their Level II and Level III counterparts in their use of FORCE: quantification, quantifying things relatively more often than processes (Level II and III texts used quantification

160

exclusively on processes). Tables 9–10 display frequency counts of GRADUATION instances in

Study 1, broken down by FORCE type (Table 9) and object (Table 10).

Table 9. GRADUATION: FORCE in Study 1 Texts. Force Level II Level III Level IV L1 Type N % N % N % N % Intensification 32 80.0% 18 75.0% 16 66.7% 94 87.2% Quantification 8 20.0% 6 25.0% 8 33.3% 14 12.8% Total 40 24 24 109

Table 10. GRADUATION Targets in Study 1. Intensification Level II Level III Level IV L1 Object N % N % N % N % Attribute 27 84.4% 17 94.4% 13 81.3% 69 73.4% Process 5 15.6% 1 5.6% 3 18.8% 25 26.6% Total 32 18 16 94

Quantification Level II Level III Level IV L1 Object N % N % N % N % Thing 0 0.0% 0 0.0% 5 62.5% 6 42.9% Process 8 100.0% 6 100.0% 3 37.5% 8 57.1% Total 8 6 8 14

Examples may serve to clarify these trends. APPRAISAL is divided into categories not by grammatical form, but by meaning; evaluation can be conferred by modifying attributes, processes, or things. When using FORCE to intensify a process, writers can modify the process with an adjective or adverb, as in Examples (56)–(58), or could use a Process that carries more emotional weight than other available word choices, as in Example (59).

(56) hatte meine Mutter richtig erkannt ‘my mother had correctly recognized’

161

(57) Das geht natürlich nicht! ‘that’s obviously no good!’

(58) was uns natürlich besonders gefiel. ‘which we of course especially liked’

(59) Ich triumphierte, als meine Mutter ‘I rejoiced as my mother’

Similarly, when intensifying an attribute, writers can convey gradations of meaning by using an additional word (60), by using the comparative or superlative form of the existing attribute (61), or labeling the person in question with a particularly strong word of insult or praise (62).

(60) wahrscheinlich war Mutter etwas sentimental ‘my mother was probably somewhat sentimental’

(61) Da ich viel klüger und gewitzter als meine Stiefschwester war… ‘since I was much smarter and cleverer than my stepsister’

(62) Was für eine Lügnerin! ‘What a liar!’

Writers can use FORCE: quantification instead of intensification when evaluative meaning is carried by a process (in which case the writer can increase or decrease its duration) or a thing

(and the writer can effect a change in evaluation by changing that thing’s quantity). In Example

(63) below, the narrator demonstrated her desire to hinder her sister’s happiness by describing that desire’s endurance; similarly, in Example (64), the narrator increases the weight of the lazy sister’s misery by evoking its eternity.

(63) Genau das wollten wir doch immer verhindern… ‘That’s exactly what we had always wanted to prevent’

(64) Sie lebte für immer und ewig in dunkle Traurigkeit ‘She lived forever and ever in gloomy sadness’

162

Evaluation can also be encoded on nouns. Writers can employ grammatical metaphor to describe the emotions they feel — nominalizing, as in Example 68, the process of feeling happy

(glücklich) as an abstract thing, happiness (Glück). As researchers on grammatical metaphor have amply demonstrated, transforming processes into nouns opens up further meaning-making space, allowing those thingified processes to be further “classified and categorized, qualified, quantified, identified and described in various way” (Byrnes, 2009, p. 52). The narrator can also create some distance between themselves and an emotion they are experiencing by elevating that emotion to participant rank, as Examples (65) through (67).

(65) Schließlich wollte meine Mutter ja nur, dass auch ich endlich etwas Glück im Leben fand ‘In the end, my mother just wanted me to find some happiness in life’

(66) So hatte ich auch wenig Mitleid mit ihr ‘So I had little sympathy for her’

(67) doch da bekam sie nur noch mehr unerwartete Schwierigkeiten ‘but there she encountered even more unexpected difficulties’

The relative absence, at lower proficiency levels, of quantification as applied to nouns may reflect an absence of evaluatively laden things to be quantified. Halliday (2002b) observed memorably that written language tends toward a more ‘crystalline’ structure that uses things rather than events as organizing principles; since then, studies of L2 writing have bolstered this claim with evidence that the successful use of grammatical metaphor is integral to the mastery of explanatory and argumentative macro-genres (Ryshina–Pankova, 2015a) and to a learner’s ability to draw general lessons from specific experiences by “‘distill[ing]’ experiences into general and abstract categories” (Ryshina–Pankova, 2015b, p. 26). This pattern within

GRADUATION might then reflect an overall trend toward nominalization among more proficient writers.

163

Realizations of GRADUATION. An additional potential trend suggested by this exploratory analysis involves the way in which GRADUATION is realized. As described in Chapter 1,

GRADUATION can be realized either by an isolated word or phrase (GRADUATION via isolation), or through the use of a verb or nominal group that carries a more intense evaluative meaning

(GRADUATION via infusion). In Examples (68) and (69), the writer employs GRADUATION via isolation, using an additional adjective or adverb to intensify a character’s beauty or anger.

(68) Dazu muss ich anmerken, dass sie… von besonderer Schönheit war ‘I must admit that she was of great beauty’

(69) Ich fand dieses Befehl ganz böse ‘I found this order totally mean’

In cases of GRADUATION via infusion, such as Examples (70) and (71) the writer uses a more heavily laden token of ATTITUDE when another, more evaluatively neutral token could also be used.

(70) kam das Aschenputtel jedoch tränenüberstromt angelaufen und jammerte ‘the Cinderella ran up, drowning in tears, and moaned’

(71) ich fand diese Situation furchtbar! ‘I found this situation frightful!’

The use of infusion was much more common in texts from L1 speakers and Level III learners. Among Level III texts, writers used infusion in 37.5% of instances of GRADUATION, and L1 speakers used infusion in 17.43% of cases; percentages for Levels II and IV were only

5% and 4.17%, respectively. GRADUATION type was significantly correlated with language level,

χ2(3, N = 189) = 16.1361, p < .01. Table 11 displays frequency counts of instances of isolation, infusion, and repetition in Study 1 texts.

The relative prominence of GRADUATION via infusion at Level III could suggest a developing ability, relative to Level II learners, to use a wider array of vocabulary appropriate to

164

the given situation. Successfully employing infusion requires a broader vocabulary than isolation

(in Example 25 above, the writer could say ‘ich fand diese Situation nicht so gut’). This variation in GRADUATION realization prompted me to examine the data in greater detail.

Table 11. Means of Expressing GRADUATION: FORCE in Study 1 Texts.

FORCE- Level II Level III Level IV L1 Mode N % N % N % N % isolation 37 92.5 13 54.2 23 95.8 85 78.0 infusion 2 5.0 9 37.5 1 4.2 19 17.4 repetition 1 2.5 2 8.3 0 0.0 5 5.6 Total 40 100.0 24 100.0 24 100.0 109 100.0

The sheer frequency with which a particular token occurs does not automatically indicate greater mastery of the corresponding concept. In the area of stance, for example, tallying the frequency of stance markers in examples of student writing does not yield as much insight into a writer’s degree of rhetorical control as accounting for the rhetorical function those markers are being used to perform.13 I therefore examined the specific realizations of GRADUATION in these texts. The 37 instances of intensification via isolation among Level II texts included only 17 unique tokens (the word sehr ‘very’ accounted for 14 of these instances).

The sample size is too small to draw firm conclusions, especially since there were very few instances of infusion in Level II and Level IV texts at all. In general, it appeared that writers tended to draw on a wider variety of terms when employing intensification via infusion, relative to intensification via isolation. Level III texts included 8 unique tokens among 13 instances of isolation (61.54%) and 7 unique tokens among 9 instances of infusion (77.78%). Among L1

13 I am indebted to Zak Lancaster for this insight.

165

speaker texts, the 85 tokens of intensification via isolation included 57 unique realizations

(67.06%); 19 instances of infusion included 17 unique realizations (89.47%).

Being a skilled language user means in part having access to a wide array of possible signs and choosing those that, within a particular context of situation, allow the language user to mean what they intend to mean. In some of these learner texts, Level II writers’ focus on intensification via isolation seems to correlate with a reliance on a smaller pool of meaning- making options. If this pattern of lesser diversity within isolation holds, then a GRADUATION analysis of learner texts might also yield insights into where those learners’ breadth of vocabulary is lacking.

4.1.7 Modifications for Study 2

As discussed earlier in this chapter, I made modifications to the APPRAISAL coding scheme for Study 2 based on questions raised or difficulties encountered in the course of analyzing Study 1 texts. These changes included the following:

(1) I coded for both the trigger and source of AFFECT.

(2) Using the scheme recommended by Ngo and Unsworth (2015), I categorized APPRECIATION

tokens as expressing either impact, quality, social valuation, or composition. I found the

guiding questions used by Ngo and Unsworth to categorize APPRECIATION subtypes—for

quality, “Does it indicate a particular standard?” (p. 13)—easier to use in an objective way.

(3) Finally, I recognized both multiple coding and multi-layering in my coding scheme,

following the lead of Lee (2015). In the first study, I did not code for instances of either

polyphony (cases in which a token could be coded as realizing one ATTITUDE type or the

other, or a blend between the two) or nesting, in which one ATTITUDE token is itself a part

of a longer token. However, during the coding process I noticed several of these

166

occurrences and wondered whether either polyphony or nesting would manifest in different

ways at different curricular levels. I therefore added these elements to the modified coding

scheme for Study 2.

Working with the original distinction between impact: reaction and impact: quality, I had found it difficult to distinguish between the two (and sometimes between APPRECIATION: impact and APPRECIATION: composition). The probe questions provided in Martin & White (2005) led me to several borderline cases, such as Examples (72) and (73).

(72) Damals wohnte ich mit meiner schönen Mutti ‘Then I lived with my beautiful (+APPRECIATION: composition or quality?) mother’

(73) Schließlich hätte ich meine zarten Hände verletzen können ‘In the end I could have injured my tender (+APPRECIATION: composition or quality?) hands’

I found the reconceptualization of impact and quality proposed by Ngo and Unsworth

(2015) very useful when interpreting cases such as these.

4.2 Study 2

Study 2 contained 23 texts, including 5 from Level I, 18 from Level II, and 17 from Level

III, as well as three texts written by L1 speakers for comparison purposes. The five Level I texts contained a total of 1111 words (for an average of 185.17 per text), among which I identified 62 instances of ATTITUDE. These included 30 instances of AFFECT, 16 of JUDGMENT, 12 of

APPRECIATION, and 4 of COVERT AFFECT.

The Level II texts comprised 4483 words over 17 texts. As might be expected, more advanced students proved able to write longer texts in the time allotted (for an average of 264.71 words per text). These texts contained 441 instances of ATTITUDE, among them 170 tokens of

167

AFFECT, 129 of JUDGMENT, 117 of APPRECIATION, and 25 of COVERT AFFECT. The proportion of

COVERT AFFECT is similar between Levels I and II, while the proportion of both JUDGMENT and

APPRECIATION tokens increase in relation to AFFECT. The proportion of the three ATTITUDE types varied among Level II texts. AFFECT tokens comprised as few as 18.18%, and as many as

58.14%, of ATTITUDE tokens within these texts; JUDGMENT tokens made up between 5.56% and

52.38% of ATTITUDE tokens at this level; and APPRECIATION tokens varied between 11.43% and

47.83%.

Level III texts contained 5089 total words over 17 texts, for an average of 299.35 per text, which included 508 instances of ATTITUDE, among them 231 tokens of AFFECT, 152 of

JUDGMENT, 109 of APPRECIATION, and 16 of COVERT AFFECT. Table 12 displays frequency counts of each type ATTITUDE across these four levels in Study 2 texts.

Table 12. Frequency Counts of ATTITUDE Types across Curricular Levels in Study 2. Attitude Type Level II Level III Level IV L1 N % N % N % N %

AFFECT 32 50.0% 171 39.0% 235 46.8% 83 48.3% JUDGMENT 13 20.3% 123 28.1% 144 28.7% 43 25.0% APPRECIATION 15 23.4% 118 26.9% 106 21.1% 36 20.9% COVERT AFFECT 4 6.3% 26 5.9% 17 3.4% 10 5.8% Total 64 438 502 172

L1 speakers wrote a total of 1963 words over three texts (averaging to 654.33 per text), including 171 tokens of ATTITUDE. Their texts contained 88 instances of AFFECT, 48 of

JUDGMENT, 26 of APPRECIATION, and 9 of COVERT AFFECT. On average, the L1 speaker texts contain the highest proportion of AFFECT (51%) and lowest of APPRECIATION (15%).

Comparing global patterns of ATTITUDE across each group, a chi-square test showed a significant difference in relative frequency of ATTITUDE type by curricular level, χ2(9, N = 1182)

168

= 17.94, p < .05. In particular, Level II writers used APPRECIATION significantly more often than

L1 writers; a chi-square test comparing ATTITUDE between those two levels was significant, χ2(3,

N = 612) = 11.73, p < .01.

4.2.1 AFFECT among Study 2 Texts

AFFECT Type. Writers at each level varied somewhat in the types of AFFECT upon which they relied, with Level I and II writers preferring AFFECT: un/happiness and Level III and L1 writers leaning more heavily on AFFECT: in/security. Overall, there was less variety within Level

I; these writers relied primarily on AFFECT: un/happiness and dis/inclination to express their emotions, with few instances of in/security (4), dis/satisfaction (1), or surprise (1). Furthermore, within AFFECT: un/happiness, Level I writers used exclusively the subtype of misery/cheer rather than antipathy/affection. More advanced writers seemed to draw on a wider range of resources, with at least ten instances at each level of AFFECT: un/happiness, in/security, and dis/inclination.

However, these frequencies in AFFECT type were not statistically significantly different across curricular level, χ2(12, N = 519) = 19.84, p = .0702.

Source of AFFECT. Writers at the beginner level tended to express AFFECT through the lens of the author, with only 13.33% of AFFECT tokens having someone other than the author as their source. This pattern seems to hold across levels; a chi-square test found no significant difference between curricular level and AFFECT source, χ2(3, N = 519) = 2.465, p = .48.

Directed vs. undirected AFFECT. At each level, AFFECT was primarily directed at a target of some kind rather than being undirected, generalized emotion. There was a slight upward trend in the incidence of directed AFFECT, but this was not statistically significant.

Situational vs. hypothetical AFFECT. Writers at Level I differed from others in their relative use of realis vs. irrealis. Nearly half of AFFECT tokens expressed an emotion that was not

169

directed toward a factual occurrence. However, a chi-square test across all levels found no significant difference in the use of AFFECT: irrealis across curricular levels, χ2(3, N = 422). The trend for both L1 and L2 writers in these kinds of stories approximates a 4:1 ratio; writers across all levels reacted emotionally to real situations about four times as often as they reacted to hypothetical causes.

Explicitness of AFFECT. Level I writers expressed AFFECT explicitly more often than not, with 76.7% of AFFECT tokens being inscribed, 10.0% evoked, and 13.3% provoked. This fits a pattern among all four levels; in each set of texts, between 70–80% of tokens of AFFECT were inscribed.

4.2.2 JUDGMENT among Study 2 Texts

As students were asked to write a story about a friend offering help, we might expect to see a reliance on tokens of JUDGMENT: propriety—praising the friend for their help—and perhaps

JUDGMENT: capacity — for situations in which the friend helped the narrator solve an intractable problem. Indeed, JUDGMENT: propriety occurred by far the most often across all levels, followed by tenacity. (Level I writers used only these two types.) The distribution of JUDGMENT types held steady across all levels; comparing Levels II and III, there was no significant difference across curricular levels for JUDGMENT type, χ2(4, N = 280) = 8.379, p = .0786.

Results from Study 1 suggested a pattern in the shifting targets of JUDGMENT resources across curricular levels. To investigate this within Study 2, I coded instances of JUDGMENT based on whether they targeted the narrator themself, the friend helping them, a family member, or another person. I realized later that having so many categories might hamper a statistical analysis; a chi-square test across all levels was impossible because Level I texts contained no instances of JUDGMENT directed toward someone other than the narrator or the friend.

170

Combining instances of “family” and “other friend” into a larger “other” category allowed me to run a chi-square test across Levels II and III, along with L1 texts. This yielded a significant difference, χ2(4, N = 329) = 13.30, p < .01. (Removing the L1 texts also yielded a significant difference, χ2(3, N = 329) = 10.22, p < .01.) Even without relying on statistical confirmation, the frequency counts suggest a clear trend here: Writers at lower levels evaluate primarily the ethical values of the helping friend (the ostensible focus of the story), while writers at higher levels distribute their JUDGMENT resources in a more balanced way, targeting other characters in the story with greater frequency, perhaps setting up other examples of more or less praiseworthy behavior as an additional yardstick to praise the helping friend. I analyze examples of this trend in the next chapter.

4.2.3 APPRECIATION among Study 2 Texts

As noted earlier, I modified the coding scheme for Study 2 to more easily differentiate between impact and quality, as proposed by Ngo and Unsworth (2015). APPRECIATION types varied more widely than in Study 1 (though it is impossible to make a direct comparison, given the modification in the APPRAISAL scheme). Frequency counts suggested that Level II writers focused more on APPRECIATION: impact (perhaps to describe the usefulness of the friend’s help in terms of its effects), while Level III writers balanced this evaluation with an equal weighty use of APPRECIATION: social valuation to describe the praiseworthy aspects of the friend’s behavior.

Comparing Levels I through III, there was indeed a significant difference across curricular levels for APPRECIATION type, χ2(6, N = 238) = 25.80, p < 0.1. (L2 texts were excluded from the test because there were no instances of APPRECIATION: composition.)

There was also an interesting divergence in the type of APPRECIATION: quality used in

Levels II and III. There were typically not enough instances of each subtype to compare them

171

across levels; here, however, Level II writers made use of both APPRECIATION: quality: aesthetics and APPRECIATION: quality: manageability, while Level III writers used almost exclusively

APPRECIATION: quality: manageability. A good deal of this variance can be attributed to two

Level III texts, which feature 4 and 7 instances of this APPRECIATION subtype. These particular texts described difficult academic situations, and they therefore described aspects of the academic field as higher or lower on a manageability scale.

4.2.4 Inscribed vs. Evoked/Provoked APPRAISAL in Study 2

At all levels, the majority of APPRAISAL tokens were inscribed (that is, explicitly realized). However, the ratio of inscribed to evoked or provoked APPRAISAL varied from group to group; there was a statistically significant difference in the prevalence of explicitness across curricular levels, χ2(3, N = 1116) = 14.187, p < .01. As might be expected, Level II and III writers used implicit evaluation relatively more often than Level I writers, with Level III writers in particular using inscribed evaluation in only 68.9% of instances (compared to 76.4% of instances in Level II and 78.3% in Level I).

Implicit APPRAISAL is divided by Don (2016) into evoked and provoked APPRAISAL. The primary distinction between the two can be phrased as a question: ‘What resources does the reader need to draw upon in order to understand the evaluation being conveyed?’ In the case of provoked APPRAISAL, the reader draws on the text itself (e.g., in the case of lexical metaphor); with evoked APPRAISAL, the reader draws on referents outside the text (e.g., cultural understandings or quotations of texts presumed know to the reader).

Level III writers employed evoked APPRAISAL (requiring references to text-external understandings) most often of any of the curricular groups — though a chi-square comparison between Levels II and III found no significant relationship between curricular level and evoked

172

vs. provoked APPRAISAL (χ2(1, N = 258) = .0038, p = .9506). Interestingly, I found no instances of evoked APPRAISAL among L1 texts. This struck me as curious, and I investigated the use of implicit APPRAISAL within L1 texts in more detail; while the L1 texts do rely on provoked

APPRAISAL, they sometimes deploy ambiguous evaluative tokens within a list of more explicit items; this is a strategy that the learner texts do not use, but which might offer a fruitful avenue for instruction. I discuss this further in the Discussion chapter.

4.2.5 Polyphony in Study 2

I coded each instance of ATTITUDE in Study 2 with a tag indicating whether the token represented an instance of polyphony and/or nesting. This was a phenomenon that I had noticed in Study 1 texts, but which I had not coded for.

Polyphony occurred in a small fraction of cases at each level, with only 6.45% of

ATTITUDE tokens (at Level I) falling into this category, though the percentages were higher at

Levels II (14.29%) and III (13.81%). Although there is not any significant difference by curricular level in the frequency of polyphony (χ2(3, N = 1181) = 4.687, p = .1962), it is striking that polyphony was so much more rare at Level I, where writers more often used tokens of evaluation that clearly belonged to one APPRAISAL type or another, and did not blend aspects of two types. There were also relatively fewer examples of polyphony among L1 writers’ texts.

Table 13. Frequency of Polyphony among Study 2 Texts. Level II Level III Level IV L1 N % N % N % N % Instances of Polyphony 11 17.2% 70 16.0% 71 14.1% 30 17.4%

Total ATTITUDE Instances 64 438 501 172

173

Between the two levels (II and III) with instances of each type of polyphony, there was marked variation in the kinds of polyphony most often seen, with a significant difference by curricular level in the frequency of polyphony type (χ2(3, N = 132) = 12.32, p < .01). Level II writers employed Fused Appraisals much more often than any other type of polyphony; as described in

Chapter 2, these combine aspects of ATTITUDE, GRADUATION, and/or ENGAGEMENT in a single token and include examples of GRADUATION via infusion. Level III writers also use Fused

Appraisals most often but also drew relatively more heavily on Appraisal Blends, which realize meaning-making potential of multiple ATTITUDE subtypes (e.g., both JUDGMENT: tenacity and

JUDGMENT: capacity), with the context of the passage determining which reading is most salient.

Table 14 displays frequency counts of each polyphony type among ATTITUDE tokens in Study 2 texts.

Nesting also occurred less often in both Level I and L1 texts. Level I texts featured instances of nesting at a rate of 1.9 per 1,000 words, compared to 6.9 at Level II and 13.4 at

Level III. Interestingly, L1 texts had only 7.6 instances of nesting per 1,000 words. . Overall, there was a statistically significant difference in the frequency of instances of nesting across curricular levels (χ2(2, N = 1173) = 16.44, p < .01). Nesting typically involves a hypotactic relationship of clauses embedded within clauses; the apparent drop-off in nesting in texts by the writers with the most experience of German may reflect the observation that the texts of more advanced writers tend to be more ‘crystalline’ and use grammatical metaphor to infuse fewer clauses with more information (Ryshina–Pankova, 2015; Ryshina–Pankova & Byrnes, 2013).

174

Table 14. Types of Polyphony among Study 2 Texts. Polyphony Type Level II Level III Level IV L1 N % N % N % N % Invoked Appraisal 3 27.3% 7 10.0% 20 28.2% 7 23.3% Border Phenomenon 0 0.0% 12 17.1% 15 21.1% 6 20.0% Fused Appraisal 4 36.4% 40 57.1% 33 46.5% 17 56.7% Appraisal Blend 4 36.4% 11 15.7% 3 4.2% 0 0.0% Total 11 70 71 30

4.2.6 ATTITUDE within Generic Stages

As in the first study, patterns of ATTITUDE use within generic stages varied among these texts; these variances were more pronounced within the Orientation and Coda stages and typically involved differences between the learner and L1 subcorpora.

Within the Orientation stages of each text, there was a statistically significant difference for ATTITUDE type across curricular levels, χ2(6, N = 245) = 35.62, p < .01. AFFECT was preferred in this stage by Level I writers (though by a small margin), while writers at Levels II and III used AFFECT and APPRECIATION in approximately equal amounts and relatively less

JUDGMENT. L1 writers, however, relied most heavily on JUDGMENT, then on AFFECT, with almost no APPRECIATION.

Writers at different levels also differed in their use of ATTITUDE types in Coda stages.

Here there was also a statistically significant difference, χ2(4, N = 109) = 13.95, p < .01.

Differences between the levels were starkest in this stage; Level II writers used a mix of

ATTITUDE types but preferred JUDGMENT; Level III writers used a higher proportion of AFFECT

175

with some JUDGMENT; L1 writers used AFFECT almost exclusively. Table 15 displays frequency counts of each of these ATTITUDE types withing narrative stages of Study 2 texts.

Table 15. ATTITUDE use within Narrative Stages in Study 2 Texts. Level I Level II ORI COM RES COD ORI COM RES COD

AFFECT 7 6 5 2 33 56 44 17 JUDGMENT 4 0 3 2 17 31 48 21 APPRECIATION 4 1 0 0 37 40 11 13 COVERT AFFECT 2 0 1 0 9 7 5 1

Level III L1 ORI COM RES COD ORI COM RES COD

AFFECT 36 76 21 21 26 16 3 15 JUDGMENT 13 36 32 13 30 8 0 2 APPRECIATION 34 32 6 6 4 4 0 1 COVERT AFFECT 4 3 3 4 4 0 1 2

Differences between the Complicating Action and Resolution stages were less pronounced; here there was no statistically significant difference in ATTITUDE type for curricular level at either the Complicating Action stage (χ2(4, N = 299) = 5.341, p = 0.254) or the

Resolution stage (χ2(2, N = 162) = 0.9268, p = .6291). (L1 texts were excluded from the second test, and Level I was excluded from both tests, because some of the ATTITUDE types yielded zeroes at these stages.)

I will explore the different patterns of ATTITUDE use, particularly the shift in emphasis from APPRECIATION to AFFECT to JUDGMENT in the Orientation stage, in the next chapter.

176

4.2.7 GRADUATION among Study 2 Texts

FORCE: Intensification and Quantification. As in the first study, texts across all levels included a higher proportion of FORCE than FOCUS among tokens of GRADUATION. A chi-square test comparing Levels II and III, as well as L1 texts, showed a statistically significant difference by curricular level in GRADUATION type, χ2(2, N = 588) = 46.87, p < .01. The difference persisted when comparing only the learner texts (χ 2(1, N = 434) = 8.168, p < .01). (Level I texts could not be included in the test because they contained no instances of GRADUATION: FOCUS at all.) Table 16 displays frequency counts of instances of GRADUATION in Study 2 texts.

All writers also preferred intensification to quantification, though this preference seemed to decrease at higher levels. There was a statistically significant difference for curricular level in the frequency of FORCE type, χ2(2, N = 496) = 22.09, p < .01. (Again, Level I texts had to be excluded from this test because of very few instances of FORCE: quantification.) Table 17 displays frequency counts of instances of GRADUATION: FORCE among Study 2 texts.

Level III writers employed FORCE: intensification by intensifying a process (rather than an attribute) relatively more often Level II writers, and this variance proved significant (χ 2(2, N

= 403) = 8.4831, p < .05). There was also a significant difference in the sorts of things writers quantified (χ 2(2, N = 93) = 7.712., p < .05). Table 18 displays frequency counts of GRADUATION:

FORCE instances, categorized by the types of objects they intensify or quantify.

Table 16. GRADUATION among Study 2 Texts.

GRADUATION Level I Level II Level III L1 Type N % N % N % N %

FORCE 21 100.0% 180 94.7% 211 86.8% 105 68.6% FOCUS 0 0.0% 10 5.3% 32 13.2% 48 31.4% Total 21 190 243 153

177

Table 17. GRADUATION: FORCE among Study 2 Texts.

FORCE Type Level I Level II Level III L1 N % N % N % N % Intensification 19 90.5% 164 91.1% 153 72.5% 86 81.9% Quantification 2 9.5% 16 8.9% 58 27.5% 19 18.1% Total 21 180 211 105

Table 18. Means of Realizing GRADUATION: FORCE in Study 2. Intensification Level I Level II Level III L1 Object N % N % N % N % Attribute 18 94.7% 154 93.9% 132 86.3% 71 82.6% Process 1 5.3% 10 6.1% 21 13.7% 15 17.4% Total 19 164 153 86

Quantification Level I Level II Level III L1 Object N % N % N % N % Thing 0 0.0% 9 56.3% 33 56.9% 4 36.8% Process 2 100.0% 7 43.8% 25 43.1% 15 78.9% Total 2 16 58 19

I speculated earlier that measures of GRADUATION might offer a proxy for assessing a writer’s vocabulary size; increased use of GRADUATION via infusion might indicate that the writer has command of a wider array of lexicogrammatical resources in order to fine-tune their meanings. In the case of these texts, however, increased use of FORCE: quantification does not necessarily reflect a broader vocabulary. As shown in Table 19, both Level II and Level III texts included just seven unique instances of FORCE: quantification, if we discount positive and negative polarity (e.g., immer and nicht immer count as one unique realization).

178

The targets of GRADUATION, however, vary significantly at each level, which more advanced writers modifying a nominalized process much more often. I explore these differences, and the ways they might be highlighted for developing writers, in the next chapter.

Results from Study 1 suggested that more advanced writers might be choosing

GRADUATION via infusion relatively more often than intensification. In this study, however, there was not a significant differences in the frequency of isolation vs. infusion seen across levels,

χ2(2, N = 391) = 5.314, p = .0701. As in Study 1, however, the individual realizations of

GRADUATION via infusion are worth more detailed attention (and will receive it in the next chapter).

4.2.8 Lexical Realizations

Research Question 4 asks whether the lexical realizations of APPRAISAL tokens vary across levels. That is, do writers at higher levels rely more heavily on certain Transitivity types or parts of speech in order to express evaluative meanings? UAMCT offers at least two ways to examine this question. Tokens can be manually coded for Transitivity categories within the SFL framework, and the Search function within UAMCT can then generate lists of tokens within each level that are coded as, e.g., both AFFECT and Thing. UAMCT can also scan and automatically code texts in many different languages for formal part-of-speech categories, using the

TreeTagger tool. This automated tool categorizes each token as either a noun, verb, adjective, pronoun, determiner, adverb, number, preposition, conjunction, or separable prefix (Schmid et al., 1999).

179

Table 19. Realizations of GRADUATION: FORCE: Quantification in Study 1.

Level GRADUATION instance Frequency Count Level II einige 2 für die meiste Zeit 1 (nicht) immer 4 jeden Tag 1 nie 1 (nicht) viel 5 weniger 6 Level III immer 10 immer noch 1 manchmal 3 (nicht) mehr 2 (nicht so) viel 24 oft 3 zu viel 2 L1 immer 5 immer noch 2 immer wieder 1 an manchen Tagen 1 mehr 1 nie 3 viel 2 von Anfang an 1 wenig 1

Based on the time required to manually code each text for APPRAISAL, I concluded that it would be infeasible to manually tag each token based on its lexical realization as well. I therefore ran

TreeTagger and ran a search among texts in each level for APPRAISAL tokens coded as nouns, verbs, or adjectives (as well as past participles). I reasoned that this would give a rough

180

indication as to whether trends between levels existed; these trends could then be further investigated if present. Effectively categorizing parts of speech proved difficult even with this automated tagger. The TreeTagger tool typically labels parts of speech in L1 German texts with reasonable accuracy (in some cases up to 97.5%; Schmid et al., 1999). It struggled, however, with learner texts, often mislabeling a misspelled word. I therefore manually checked each tag for accuracy. I was also uncertain how to categorize APPRAISAL tokens in which an entire phrase or clause, rather than a single word or verbal phrase, carried an evaluative meaning, as in

Example (74) below:

(74) ich fühlte als ob ich nicht genug Zeit hatte, um alle meinen Prioritäten fertig zu machen ‘I felt as though I didn’t have enough time to finish all my priorities’

The following results should therefore be interpreted as tentative. A trend toward nominalization was apparent in the Study 2 texts, particularly in realizations of AFFECT. Of the

ATTITUDE tokens I coded through this process, 3.57% were realized by nouns in Level I, while

60.71% were encoded by verbs and 35.71% by adjectives. By contrast, at Level II, 7.32% of these tokens were realized by nouns, 38.21% by verbs, and 54.47% by adjectives; by Level III, the proportion of nouns had reached 17.17% as against 42.42% (verbs) and 40.4% (adjectives).

Among tokens of AFFECT, there was a statistically significant trend toward increased use of nouns at higher curricular levels, χ2(4, N = 391) = 23.58, p < .01. This difference was not significant when assessing JUDGMENT or APPRECIATION. Table 20 displays frequency counts of

ATTITUDE tokens realized by nouns, verbs, and adjectives at each curricular level.

181

Table 20. Lexical Realizations within Study 2 Texts. Level I Level II noun verb adj total noun verb adj total

AFFECT 1 17 10 28 9 47 67 123 JUDGMENT 0 7 7 14 3 19 46 68 APPRECIATION 3 0 9 12 7 10 59 76 COVERT AFFECT 0 0 3 3 2 0 12 14

Level III L1 noun verb adj total noun verb adj total

AFFECT 34 84 80 198 20 33 17 70 JUDGMENT 9 35 47 91 3 17 21 41 APPRECIATION 21 9 64 94 5 6 19 30 COVERT AFFECT 1 0 7 8 2 4 3 9 Note. Adj = adjective.

The proxy values yielded by the automatic TreeTagger coder therefore suggest that writers at higher curricular levels are becoming increasingly able to express evaluation via nominalized processes (and exploit all the broadened meaning-making potential that technique involves), and that the field of emotion (e.g., Hunger ‘hunger,’ Angst ‘fear,’ Freude ‘joy’) offered a range of vocabulary that these learners found particularly manipulable for this purpose.

Manual coding for Transitivity categories could test this hypothesis.

4.3 Summary

In this chapter, I reported patterns and trends suggested by an APPRAISAL analysis that involved hand-coding 32 narratives for ATTITUDE, GRADUATION, explicitness, polyphony, and nesting, and which traced the occurrence of each across curricular levels and stages. I also flagged categories in my original APPRAISAL scheme that proved difficult to distinguish in

182

practice (e.g., flagged, afforded, and provoked ATTITUDE; APPRECIATION: reaction: impact and

APPRECIATION: reaction: quality).

Chi square tests were run to highlight differences in the frequency of use of the three

ATTITUDE types across curricular levels. These tests yielded relatively few trends in the use of

AFFECT, though results from one L1 text prompted me to examine the sources of AFFECT at each level more closely. In the case of JUDGMENT, writers at lower levels tended to restrict the targets of their JUDGMENTs to fewer characters, whereas writers at higher levels tended to use a variety of characters as implicit yardsticks for moral behavior. APPRECIATION-related trends were less clear but suggested that higher-level writers used APPRECIATION: social valuation relatively more often, perhaps to discuss phenomena that affect more characters than just the narrator.

Comparing evaluation across narrative stages yielded additional potential trends; in the

Orientation Stage, APPRECIATION was most common at lower levels; AFFECT at intermediate levels; and JUDGMENT at higher levels.

In the discussion chapter, Chapter 5, I will explore at a more granular level the statistically significant trends that emerged from the results of Study 2 presented in this chapter and discuss the implications that each might have for language instruction. I will also explore individual texts that exemplify trends or characteristics that an APPRAISAL approach can help illuminate.

183

CHAPTER 5. ANALYSIS

This chapter explores the patterns suggested by the quantitative results from the two studies. The qualitative analysis described here examines variation in the use of ATTITUDE types at each curricular level, the increased use of provoked evaluation, the trend toward greater use of

AFFECT in codas, the increased use of GRADUATION: FOCUS, and the various kinds of polyphony common at each curricular level.

The discussion so far has focused on comparisons between entire subcorpora of texts from each curricular level. However, qualitative analyses focused on a few texts are also valuable. For example, O’Donnell (2014) uses the example of one political speech by then-

President Obama in Cairo to explore how an analyst gain a better understanding of the writer’s expressed or implied beliefs by examining which positive or negative traits they consistently emphasize — and by tracking the targets that most often receive those evaluations. In the case of the Obama speech, O’Donnell (2014) argued that “the repeated use of tenacity in positive evaluations tells us that Obama thinks that tenacity is a positive social value” (p. 107). Along similar lines, for a text containing multiple voices, an APPRAISAL analysis could code for each individual voice and then examine the relative frequency of ATTITUDE types expressed through each voice. I explore similar lines of inquiry in the close reading of individual texts in this chapter.

5.1 ATTITUDE Trends across Levels

An analysis of AFFECT, APPRECIATION, and JUDGMENT across curricular levels yielded potential trends in terms of the source or target of each. With increased proficiency, writers tended to express AFFECT more often through voices other than the narrator to generate pathos;

184

to apply APPRECIATION to situations that affected a community beyond the narrator; and to target a broader variety of characters with JUDGMENT in order to make general points about shared ethical values.

5.1.1 AFFECT: Emergence of Multiple Perspectives in Study 1

As I noted in the previous chapter, within Study 1, texts at higher levels displayed greater variety in terms of AFFECT types (e.g., un/happiness, dis/inclination). This result prompted me to look more closely at AFFECT within individual texts in both studies. Writers at lower levels used a narrow range of AFFECT to justify the narrator’s selfish actions by appealing to the reader’s sympathy for her plight, and/or to evoke anger toward her sister. At higher levels, an increase in

AFFECT types tended to correlate with intensified use of additional voices. Expressing emotions through various non-authorial characters allows the author to offer explanations for the narrator’s behavior from less biased sources. Table 21 displays frequency counts of each AFFECT type across Study 1 texts.

Table 21. Frequency Counts of AFFECT Types in Study 1.

AFFECT Type Level II Level III Level IV L1 Un/happiness 9 20 6 17 Dis/satisfaction 3 2 0 4 Dis/inclination 7 3 5 6 In/security 13 6 1 4 Surprise 0 1 0 0

The prompt for the writing task in Study 1 charged students with narrating a story from the perspective of a character typically viewed as the villain. This poses a challenge to a writer wishing to establish an affiliation with the reader. Ordinarily, a writer can use JUDGMENT to assert that affiliation by evaluating as moral behavior that the reader can also be expected to

185

praise; in the case of ‘Pechmarie’ as she is typically portrayed, this is a difficult assignment

(unless the reader also happens to endorse laziness). AFFECT offers another means of establishing an identification with the reader, through the channel of pathos. Level II writers typically did so by casting the narrator as a weak, ill character who lacked the fortitude to complete the same tasks as her sister. In order to do this, they relied primarily on two emotions: un/happiness and in/security.

Some authors at this level deployed AFFECT in relatively unsympathetic ways. The instances of AFFECT listed in Table 22, selected from Text ji1 at Level II, impute to the narrator somewhat selfish emotions, which explain the narrator’s motivations without generating much sympathy:

Table 22. Instances of AFFECT in Text ji1 (Level II).

Token Translation AFFECT type I only wanted the Ich wollte nur das Gold!! (two instances) gold!! dis/inclination Because I wanted weil ich Gold wie meine Stiefschwester haben gold like my wollte stepsister dis/inclination wütend furious dis/satisfaction den ganzen Tag geweint cried the whole day un/happiness so glücklich so happy un/happiness

In the text featured in Table 22, the narrator is characterized by her desire, her happiness at the thought of receiving riches, and her sadness at their absence. This narrative plays on the sister’s frustrated desires, explaining that the narrator acted as she did weil ich Gold wie meine

Stiefschwester haben wollte ‘because I wanted to have gold like my sister [had],’ and that thinking on this possibility made her so glücklich ‘so happy.’ This frustrated expectation explains why the narrator reacts at the end with sadness (Ich habe den ganzen Tag geweint ‘I cried all

186

day’). In this text, the writer seems to have approached the challenge of taking the narrator’s perspective by using emotion to explain, if not excuse, their actions.

When lower-proficiency writers do use AFFECT from a variety of sources, they create more opportunities to generate pathos on the part of the reader. In a Level II text, Text si2, the narrator’s actions are predicated on privation and need; the narrator repeatedly casts herself as hungrig ‘hungry’ and krank ‘ill.’ In this telling, the hardworking sister chooses not to share her wealth with her stepsister and stepmother, prompting the narrator to describe the stepmother as wütend ‘furious’ and traurig ‘sad.’ AFFECT serves to explain many of the narrator’s motivations throughout this text, which is excerpted here:

Aber war ich faul. Weil ich immer sehr krank war, konnte ich nicht wie sie arbeiten. Eine Tage ist meine Schwester mit vielen Geld zurück gekommen. Meine Mutter und ich, ohne sie, hatten nicht so viel für essen. Meine Mutter war sehr wütend und traurig, dass sie hilfte meine Schwester aber meine Schwester mochte nicht das Geld mit uns teilen. Mit anderen Personen war sie sehr gut und freundlich. Aber mit die Familie die sie schützte war sie sehr bose. Meine Mutter und ich musste Geld verdienen um essen zu kaufen. Wir waren sehr hungrig und Geldlos. Aber war ich zu Krank für arbeiten. Eine Abend sagte meine Mutter “Du musst fur Frau Holle arbeiten um Geld zu verdienen.” Ich war sehr krank, aber ich lebte meine Mutter viel und war sehr hungrig. Ich konnte nicht schmutzig sein, weil ich bin taglich krank. Ich möchte nicht mein Kopf verletzen, weil ich habe immer die Kopfschmerzen. Ich konnte nicht das Brot oder das Apfel hilfen, weil meine Gesundheit ist nicht so stark.

‘But I was lazy (-JUDGMENT: capacity). Because I was always sick (-AFFECT: in/security), I could not work like she did. One day my sister returned with a lot of money (+APPRECIATION: social valuation: consequence). My mother and I, without her, did not have much to eat (-AFFECT: in/security: dis/quiet). My mother was very enraged (-AFFECT: dis/pleasure: dis/satisfaction) and sad (-AFFECT: un/happiness: misery/cheer) that she helped my sister, but my sister did not want to share the money with us (Invoked Appraisal: - JUDGMENT: propriety via -AFFECT: dis/inclination: non-desire). With other people (- JUDGMENT: veracity) she was very good and friendly (+JUDGMENT: propriety). But with the family that protected her (+JUDGMENT: propriety) she was very wicked (-JUDGMENT: propriety). My mother and I needed to earn money in order to buy food. “We were very hungry (-AFFECT: in/security: dis/quiet) and broke (-AFFECT: in/security: dis/quiet). But I was too sick (-AFFECT: in/security: dis/quiet) for work. One evening, my mother said, “You must work for Frau Holle in order to earn money.” I was very sick (-AFFECT: in/security: dis/quiet), but I loved my mother (+AFFECT: un/happiness: antipathy/affection) a lot and was very hungry (-AFFECT: in/security: dis/quiet). I could not get dirty (-APPRECIATION: composition: balance), because I am sick (-AFFECT: in/security: dis/quiet) every day. I [did] not want to hurt my head (-AFFECT: dis/inclination: non-desire), because I always have headaches (-AFFECT: in/security: dis/quiet). I could not help (-JUDGMENT: capacity) the

187

bread or the apple, because my health is not so strong (-APPRECIATION: social valuation: consequence).’

In the Orientation stage (which was described in greater detail in Chapter 3), the narrator primarily uses AFFECT to qualify the negative JUDGMENT: capacity of herself with references to her illness (-AFFECT: insecurity), as in the example below:

In the Complicating Action stage in the same narrative, the narrator raises an implicit eyebrow at the actions of the ostensibly admirable Goldmarie by describing the emotional reactions that her actions provoke. When Goldmarie returns with wealth, this provokes anger and sadness on the part of the mother, because Goldmarie declines to share. The reader has already been primed to treat the mother as an ethical source, as she has been the target of positive

JUDGMENT: propriety throughout the orientation. The narrator reinforces her stress concerning her own illness and penury, then uses AFFECT as part of an Invoked Appraisal that invites the reader to judge Goldmarie negatively.

AFFECT also provides a vehicle to criticize the ‘good sister’ directly. The narrator also ascribes emotions to Goldmarie once, writing that she mochte nicht das Geld mit uns teilen

‘didn’t want to share the money with us.’ It is easy to judge as unethical someone who declines to share her wealth, making this instance of AFFECT an invocation of implied negative

JUDGMENT: propriety as well.

Tokens of AFFECT proliferate as the Complicating Action stage continues. The narrator’s reaction to her mother’s command to work for Frau Holle is mediated by expressions of in/security: dis/quiet, referring both to her health and to their family’s financial situation. This negative AFFECT plagues her as she imagines having to navigate Frau Holle’s realm; she is

188

worried that becoming dirty or hitting her head would aggravate her illness.14 Nearly every feature of this stage of the narrative contributes to the narrator’s nervousness about harm to herself or her family, either because of poverty (Wir waren sehr hungrig und Geldlos), disease

(Ich konnte nicht schmutzig sein, weil ich bin taglich krank), or injury (Ich möchte nicht mein

Kopf verletzen, weil ich habe immer die Kopfschmerzen); this succession of evaluations builds up a sense of thematic consistency throughout the tale (see discussion of prosody below and, e.g.,

Thompson, 2014a). This particular text, then, relied almost exclusively on two types of AFFECT

(in/security and un/happiness) — but did so rather effectively, creating a reinforced impression of anxiety on the part of the narrator with which it is easy to sympathize.

The narrator above uses a strategy of deploying AFFECT to evoke sympathy for a reviled character. They make the additional clever choice to have another character—the mother— essentially vouch for the narrator’s version of events. When Goldmarie returns home from Frau

Holle’s realm it is the mother who is angered and saddened by her behavior; since the reader is less predisposed to dislike the mother, it is perhaps easier for the reader to identify with her righteous anger.

This text actually includes a number of sources of AFFECT — primarily the narrator, but also the mother, Goldmarie, and Frau Holle. The narrator situates the emotional tokens wütend

(furious) and traurig (said) with the mother; Frau Holle also expresses rage (wütend again). In the other texts at this level, students express most of the AFFECT via the narrator, but in each text,

14 It could be argued that the narrator’s references to illness actually represent instances of JUDGMENT: capacity. For consistency, I chose from the outset to label expressions of illness as AFFECT: insecurity, since labeling someone as ill does not reflect an evaluation of their ethics or morality. In this case, the AFFECT: insecurity coding is further justified when we reflect on the importance of keeping context in mind when coding, taking into account the evaluative prosody of the narrative as a whole (Macken–Horarik & Isaac, 2014; Thompson, 2014a).

189

AFFECT is expressed by at least one character (the mother once in Text ji1, Goldmarie three times in Text si2).

The L1 texts include a greater variety of AFFECT types, which in part seems to reflect the use of additional characters as sources to express certain emotions. In these texts, authors make greater use of the opportunity (exploited by just one Level II author) to situation emotions within many characters to give those evaluations greater or lesser weight. In Text xn1 (an L1 text), three characters apart from the narrator express emotions: Goldmarie (who expresses only unhappiness: misery); the mother (characterized by happiness: affection), and Frau Holle

(unhappiness: antipathy). The narrator, in her interactions with these characters, explores a range of emotions beginning with AFFECT: satisfaction: pleasure and AFFECT: security: quiet at her favored position in the household, to AFFECT: disinclination (toward Frau Holle’s chores) and inclination (toward the riches she expects to receive) and eventually AFFECT: unhappiness: misery at her final fate.

Results from Study 2 offer further tentative support to the conclusion that more proficient writers tend to distribute tokens of AFFECT among a wider array of voices. Table 23 displays the breakdown of authorial vs. non-authorial AFFECT at each level in Study 2, as well as the frequency with which non-authorial sources expressed AFFECT. While authorial AFFECT actually represented a lower percentage of AFFECT tokens within Level I texts, more proficient authors distributed AFFECT more evenly among characters other than the narrator.

The Humanities Assessment Project at Georgetown identified the inclusion of multiple voices as one key criterion of progress toward humanities learning goals (Ryshina–Pankova,

2015b); by “incorporating into the narrative the perspectives of various characters and their underlying literary, historical, and political positioning” (Cunningham et al., 2018, p. 148),

190

writers can demonstrate their understanding of the various identities, value systems, and implicit understandings that coexist in a linguistic community. While these retold fairy tales may not evince as much specific awareness of contemporary German-speaking culture, they provide further instances in which the use of multiple voices helps the author achieve their goal in composing this text; this analysis therefore lends support to the value of identifying multiple voices when evaluating student work.

Table 23. AFFECT Sources in Study 2.

AFFECT Source Level I Level II Level III L1 N % N % N % N % Author 27 84.4 151 88.3 214 91.1 71 84.5 Non-Author 5 15.6 20 11.7 21 8.9 13 15.5 Total 32 100 171 100 235 100 84 100

Non-Authorial Level I Level II Level III L1 Sources N % N % N % N % Friend 5 100 8 40.0 12 57.1 9 69.2 Family Member 0 0.0 1 5.0 5 23.8 2 15.4 Other 0 0.0 11 55.0 4 19.0 2 15.4 Total 5 100 20 100 21 100 13 100

5.1.2 APPRECIATION in Study 2: From Impact on Self to Value for Others

In the case of AFFECT, texts at higher proficiency levels showed greater variety in the number of voices expressing emotions. Greater variety also characterized the use of

APPRECIATION resources at higher proficiency levels; at higher levels, writers drew on

APPRECIATION types in a more balanced way, relied less on adjectives and more on verbs and nominalizations in order to realize APPRECIATION, and used APPRECIATION (particularly

191

APPRECIATION: social valuation) more often in order to evoke shared values, rather than simply commenting on the narrator’s individual reactions.

5.1.2.1 APPRECIATION Realizations at Level I

Level I writers used APPRECIATION sparingly and typically in the form of adjectives that commented on an event that had just occurred; e.g., Dies Kampf war nicht dramatik ‘This fight was not dramatic.’ Most of the instances of APPRECIATION involved appraising something as either gut ‘good’ or nicht gut ‘not good.’

5.1.2.2 APPRECIATION Realizations at Level II

APPRECIATION proved more common in Level II and was used most often to describe the narrator’s personal reactions. Specifically, most Level II texts contained at least three instances of APPRECIATION: impact. In most cases, these tokens described the narrator’s reaction to an element of their environment, such as their academic situation in Examples (1)–(3), a social situation in Examples (4)–(5), or classes or groups important to the narrator’s life in Example

(6).

(1) Georgetown ist oft nicht so interessant ‘Georgetown is often not so interesting’

(2) Dieses Semester ist sehr gut ‘This semester is very good’

(3) Während meine erste Jahr toll war, gab es schwer Zeit ‘While my first year was good, there were difficult times’

(4) es war sehr schwer, weil ich nicht sie schlechte fühlen wollte ‘It was very difficult, because I didn’t want to make her feel bad’

(5) Ich muss sagen, dass ein Aspekte meines Lebens im Universitaet ein bisschen schwer ist: mein persönliche Leben ‘I must say that one aspect of my life in the university is a little difficult: my personal life;

192

(6) Diese Treffen mit die NAIMUN gruppe war seht laut und schwer. ‘This meeting with the NAIMUN group was very loud and difficult’

Throughout Level II texts, one commonality among impact tokens is that they refer either explicitly or implicitly to an effect on the narrator specifically. In Example (7), repeated references to the narrator’s internal state, including an instance of COVERT AFFECT, make it clear that they mean the semester has been good for them, not necessarily for the Georgetown population in general.

(7) Dieses Semester ist sehr gut und froh, aber ich habe viele viele Dinge zu tun ‘This semester is very good (+APPRECIATION: impact) and happy (+COVERT AFFECT), but I have a lot of things to do (-AFFECT: in/security)’

In the previous examples, one could add the tag für mich ‘for me’ to the impact token while preserving the narrator’s meaning. Nearly every instance of impact falls into this category; almost all refer specifically or implicitly to the narrator’s experience. Just a few tokens of impact appraise an aspect of the environment that affects a wider group, as in Example (8).

(8) Wenn es in Georgetown geschneit, gibt es fünf oder sechs inches - das ist so viel! ‘[When] it snowed in Georgetown, there [were] five or six inches - that is so much!’

Expressions of impact that refer to an effect on other characters are even more rare; the only such instances occur in Examples (9) through (11).

(9) Und Sie würden das Stadt lieben; es ist aufgeregt, und es hat vielen Aktivitäten, dass Sie interssant finden würden. ‘And you would love the city; it is excit[ing], and it has many activities that you would find interesting’

(10) Ich hoffe dass, alles da gut ist! ‘I hope everything there is good!’

(11) Seiner Hand ist blau und groß, und es schmerzte zu viel ‘His hand is blue and large, and it hurt too much’

193

In short, Level II writers seem to use APPRECIATION: impact resources primarily to add personal commentary to situational description. This lack of functional variety is matched by the lexicogrammatical forms the narrators use to realize this particular resource of ATTITUDE. As mentioned in Chapter 3, the vast majority of impact tokens are realized in the form of adjectives.

Verbal and nominal realizations displayed in Table 24 represent the only exceptions.

Table 24. Verbal and Nominal Realizations of APPRECIATION at Level II. Token Translation Appreciation Type Ich mag diese schwierig Diskutieren nicht I don’t like these difficult impact discussions

Ich mochte gern dieses Konversation I liked this conversation impact

Ich hatte Tanzen (. . .) nur dieses I only started dance this impact Semester beginnt, aber ich liebe es semester, but I love it

Kekse macht mir immer besser fühlen Cookies always make me feel impact better

Am meisten von meine Klassen waren Most of my classes were very impact sehr interessant, ich habe sie geliebt und interesting, I loved them and my meine Noten waren sehr gut grades were very good

Seiner Hand ist blau und groß, und es His hand is blue and large, and it impact schmerzte zu viel hurt too much

Es gab einer Mal, in der ich hatte eigene There was one time when I had valuation Schwierigkeiten some problems

Das war natürlich eine Problem That was of course a problem valuation

194

Other APPRECIATION resources display slightly different patterns of use. In the case of

APPRECIATION: social valuation, for instance, Level II writers use nominalizations slightly more often to describe behavior or occurrences that are considered socially beneficial or detrimental.

In Example (12), the narrator discusses their efforts to address their roommate’s disorderly behavior.

(12) Obwohl das Febreze hatte der Geruch verbessert, seine Kraft war nicht genug ‘Although the Febreze improved (+APPRECIATION: quality) the odor (-APPRECIATION: social valuation), its power was not enough (-APPRECIATION: quality)’

This is a rather banal example (Level II texts contained relatively few instances of

APPRECIATION: social valuation from which to draw), but it still serves to illustrate the meaning- making potential of nominalizations. The narrator avoids make any argument about what level of hygiene is appropriate to a college dormitory (even though personal experience suggests that a wide range of perspectives exists on this topic). Instead, the undesirability of the roommate’s uncleanliness is simply taken for granted, and nominalizing as der Geruch ‘the odor’ allows the author to comment on the effectiveness of her cleaning methods and her efforts to improve her situation.

5.1.2.3 APPRECIATION Realizations at Level III

Writers at Level III, when they used tokens of APPRECIATION: impact, used them in almost all cases to describe an effect on the narrator individually. Example (13) displays an instance in which the narrator gestures toward other characters’ reactions to certain situations; in

Example (14), the narrator describes as an idea as being schrecklich ‘appalling’ to her roommate, implying a mental state on the part of that character.

(13) Jetzt sagen wie einander kaum ein Wort, obwohl sie immer noch sehr eng sind ‘Now we hardly speak to each other, although they are still very close’

195

(14) Rachel, meine Mitbewohnerin, die eine Krankenschwester ist, sagt mir, dass es war ein schreckliche Idee, mit so gross eine Halsschmerzen zu trinken ‘Rachel, my roommate, who is a nurse, told me that it was a terrible idea to drink with so much throat pain’

The group of Level III texts differs from Level II, however, in its relatively higher incidence of APPRECIATION: social valuation. This type of evaluation “has to do with our assessment of the social significance of the text/process” (Bednarek, 2009a, p. 153); it reflects the degree to which an event or behavior matches shared understandings of what is prized by the community. Example (15 )provides an instance that is fairly low-stakes but nonetheless instructive:

(15) Als ich meine Note zurück bekomme, war ich ganz enttäuscht. Es war ein C-. ‘When I received my grade back, I was very disappointed. It was a C- (-APPRECIATION: social valuation).’

In describing their performance with a letter grade, the narrator assumes that the reader is familiar with the American university grading system and shares the implicit belief (not necessarily shared in universities with less grade inflation) that this grade represents poor work.

Other instances of APPRECIATION: social valuation in Level III texts, such as Example

(16), similarly evaluate situations and relationships based on their conformity with what is commonly thought desirable in the narrator’s community.

(16) Meine erste Semester bei Georgetown ist sehr gut gelaufen. Ich habe einen engen Freundeskreis aufgebaut, viel gelernt, und wirklich eine zweite Heimat gefunden. ‘My first semester at Georgetown went very well (+APPRECIATION: impact). I built a tight (+APPRECIATION: social valuation) circle of friends, learned a lot, and really found a second home (+APPRECIATION: social valuation).’

The narrator’s evaluations of their community at Georgetown presuppose the shared understanding that in college, it is important or desirable to have a strong circle of friends. This meshes with a concept of postsecondary study that involves communal living and the

196

development of interpersonal skills, aspects of university life that may be less emphasized in academic institutions in Europe.

5.1.2.4 APPRECIATION Realizations in L1 Texts

Within L1 texts, tokens of APPRECIATION: impact did reflect in most cases the narrator’s reaction to events and behaviors. They did, however, reflect other characters’ reactions relatively more often. In Example (17), the narrator describes a friend’s response to another character’s kind words, evaluating that second character’s kindness in reference to the friend’s reaction. In

Examples (18), the narrator identifies another character for whom a certain situation was terrible; in Example (19), the writer asserts that other readers (not only the narrator) would find the following story surprising.

(17) konnte ich merken, dass allein seine Stimme sie sehr beruhigt hat ‘I could see that his voice alone calmed her down very much’

(18) Meiner besten Freundin aus der Uni ist etwas ganz schreckliches passiert ‘Something completely terrible happened to my best friend from the university’

(19) Ihr werdet nicht glauben, was alles passiert ist ‘You won’t believe what happened’

It is also worth noting that the use of impact in Example (19) reflects the writer’s awareness of the multiple perspectives being expressed in their story; in contrast to most instances of impact in Level II, this writer is evaluating situations based on how they affect other characters, not just the narrator.

The trend toward increasing reliance on social valuation in higher-proficiency texts parallels a process described by Derewianka (2007), who in a study of Australian students’

English essays argued that “[o]ne way of viewing the development of evaluation strategies is in terms of the increasing ‘institutionalization of feeling’” (p. 160). Less experienced students in

Derewianka’s study tended to evaluate events primarily with emotional, personal responses,

197

while students who had undergone more years of schooling tended to recontextualize emotional responses as evaluations “grounded in social norms” and expressed by JUDGMENT and

APPRECIATION. In a similar way, lower-proficiency writers in Study 2 used APPRECIATION primarily to describe the effect of a situation on the narrator personally (reflected in the prevalence of APPRECIATION: impact tokens), while higher-proficiency writers began to explore those situations’ effects on, or value to, their community at large.

5.1.3 JUDGMENT in Study 2: A Broadening of Targets

Results from Study 2 showed a significant effect of curricular level on the target of

JUDGMENT used by the writer. Lower-proficiency writers tended to judge the narrative’s focal friend character, while more experienced writers tended to judge other characters (including the narrator) more often.

Tokens of JUDGMENT at higher levels also tended to perform more evaluative work.

Writers at lower proficiency levels typically used Judgment in an explicit, targeted way, highlighting the ways in which their individual friend showed moral fiber in a particular circumstance — e.g., describing a friend as freundlich ‘friendly,’ nett ‘kind’ or die beste ‘the best.’ Texts at higher levels included more instances of JUDGMENT that performed double duty, evaluating one character while implicitly making an appraisal of another character — for instance, juxtaposing a passive security guard with a much more assertive friend.

198

5.1.3.1 JUDGMENT Targets in Level I Texts

In Level I texts, authors primarily targeted JUDGMENT toward their friends. They expressed JUDGMENT of themselves only twice, and only in the context of their ability to keep up with coursework, as in Example (20):

(20) Ich habe studiert und studiert aber ich lernte den Information nicht ‘I studied and studied (+JUDGMENT: tenacity) but I did not learn the information (-JUDGMENT: capacity)

These texts included twelve other instances of JUDGMENT, all targeting the friend character. These ethical evaluations of friends were fairly surface-level; most of these tokens, including Examples (21) and (22), involve general expressions of affection or explicit descriptions of helpfulness or goodness:

(21) Ich mogen meiner Zimmerkollegin ‘I [like] (Blended + AFFECT: un/happiness: affection / +JUDGMENT: propriety) my roommate

(22) Matt ist sehr schön und ein sehr gut Freund! ‘Matt is very handsome and a very good (+JUDGMENT: propriety) friend!’

In a few cases, such as Examples (23) and (24), friends are more implicitly praised for their helpful actions. These appraisals occur immediately after other positive expressions concerning the friend, making it clear that the reader is meant to interpret this comment as an example of moral behavior.

(23) Francesca ist sehr freundlich und ruhig. Sie hat geweißt, dass ich gestresst war. ‘Francesca is very friendly (+JUDGMENT: propriety) and quiet. She knew that I was stressed (+JUDGMENT: propriety)’

(24) Es war Maria! Sie hilf mir und sie rief einen Krankenwagen. Gott sie Dank! ‘It was Maria! She helped me and she called an ambulance (+JUDGMENT: propriety). Thank God!’

In all but one of these cases, tokens of JUDGMENT co-occur only with other tokens of

JUDGMENT, not of any other ATTITUDE type. There is just one instance of an Invoked Appraisal,

199

in which an expression of another character’s AFFECT can also be read as implicit JUDGMENT. In a previous example, the author described her roommate as expressing antipathy toward her other friends, writing that sie mochte nicht meinen Freunden in Zimmer ‘she doesn’t like my friends in the room (-AFFECT: un/happiness: antipathy/affection; invoked -JUDGMENT: propriety).’

As mentioned in Chapter 2, a token of AFFECT can also invoke JUDGMENT if the object of the emotion is something that is universally admired or despised; the source of the AFFECT can be judged based on their attitude toward that object. In this case, one would probably be expected to like the narrator’s other friends. In describing their roommate’s dislike of these other friends, the narrator could be evaluating their roommate as cold or unsociable. This instance of Invoked

Appraisal, however, is the exception; most the instances of JUDGMENT in Level I texts do not involve this kind of polyphony.

5.1.3.2 JUDGMENT Targets in Level II Texts

In Level II texts, the balance of JUDGMENT targets shifts to focus on the author’s friends, whom Level II writers praised both directly and via contrast with secondary characters. As Table

25 indicates, the author’s friends represented the most common JUDGMENT target at this level and were most often evaluated with tokens of JUDGMENT: propriety. Level II writers evaluated the narrator relatively less often and tended to use JUDGMENT: capacity rather than propriety.

Table 25. Frequency Counts of JUDGMENT Tokens at Level II. normality propriety veracity capacity tenacity total self 1 3 0 12 2 28 friend 0 63 0 10 1 73 family 0 1 0 0 0 1 others 1 19 1 7 0 28

200

JUDGMENTs of self are again typically related to the narrator’s abilities rather than moral gifts and failings, with Examples (25)–(27) representing the only three instances of JUDGMENT: propriety:

(25) Ich habe das richtige Ding gemacht ‘I did the right thing‘

(26) obwohl ich ein Desaster war ‘Although I was a disaster’

(27) Sie hilft mich und ich helfe sich ‘She helps me and I help [her]’

As in Level I, other self-judgments realize JUDGMENT: capacity or tenacity and involve the narrator’s ability to keep up with coursework or other activities. This level includes 14 instances of JUDGMENT: capacity or tenacity involving the self, such as those in Examples

(28)–(30).

(28) Ich bin faul und unordentlich ‘I am lazy (-JUDGMENT: tenacity) and disorganized (-JUDGMENT: capacity)

(29) Ich hatte keine Idee was zu tun ‘I had no idea what to do (-JUDGMENT: capacity)’

(30) konnte ich die Aufgabe nicht machen weil ich die Information nicht verstehen ‘I could not do the assignment because I do not understand the information (-JUDGMENT: capacity)’

Level II texts are also similar to those in Level I in that they evaluate friends’ morality much more often than the narrator’s; these texts included 74 instances of JUDGMENT implicating the main friend character.

The friend is sometimes praised for their intelligence or ability, as in Examples (31)–(33).

(31) weil er sehr intelligent und freundlich ist ‘Because he is very intelligent (+JUDGMENT: capacity) and friendly (+JUDGMENT: propriety)’

201

(32) In unter zwei Uhren, nahm sie mein officialische Transcript aus dem Büro ‘In under two hours (+JUDGMENT: capacity), she took my official transcript out of the office’

(33) sie hat immer ihre Hausaufgaben mindestens zwei Tage im Voraus ‘She always [completed] her homework at least two days in advance (+JUDGMENT: capacity)’

More often, however, the friend is praised or condemned using JUDGMENT: propriety, with 63 of the 74 tokens belonging to this type, including Examples (34) and (35):

(34) Heute habe ich zwei sehr nett Mitbewohnerin ‘Today I have two very nice roommate[s]’

(35) sie hat meine Leben gerettet, so ist es mir wichtig, dass Sie über sie wissen ‘She saved my life, so it is important for me that you know about her’

As in the previous level, the friend is often described either with generically positive language or adjectives describing their kindness. But writers at this level also evaluated the friend more often in superlative ways, as in Examples (36) through (38).

(36) meine Freundin, Gen ist, am besten ‘My friend, Gen, is the best (+JUDGMENT: propriety)’

(37) Sie ist die beste! ‘She is the best (+JUDGMENT: propriety)!’

(38) Und sie ist die letzte Person, die ich auf eine Party zu sehen (to expect) würden ‘And she is the last person I would expect to see at a party (+JUDGMENT: propriety)’

At the lexicogrammatical level, these constructions indicate a greater command of superlatives and relative clauses than are typical at for Level I writers. Some of these instances of

JUDGMENT also go beyond an explicit evaluation of one target, implicitly including in their praise a general behavior the narrator finds laudable. As O’Donnell (2014) points out, ethical evaluations can cut both ways, shedding light not only on an author’s impression of the people they describe, but also on the values they see as most praiseworthy. In his analysis of the 2009

Obama speech in Cairo, O’Donnell interprets repeated references to the Egyptian people’s

202

tenacity as implying that “Obama thinks that tenacity is a positive social value” (p. 108); when a writer consistently praises a target for holding a particular characteristic, they implicitly praise that characteristic as well. This may seem tautological until one considers that many attributes can be used to praise or condemn essentially the same behavior; one might praise a hard- working, goal-oriented scientist as ‘ambitious,’ then use the same epithet to criticize a politician with his eye on the next-highest office. (Indeed, there is evidence from the field of political science that voters often do this unconsciously, applying negative evaluations to ambitious female politicians while male politicians receive praise for that same ambition; see Okimoto &

Brescoll, 2010). The Level II writer of Example (38) has made a similar kind of move, praising their friend by reference to a behavior that one might otherwise find suspect.

Text fi9 also described a friend as kind and ethical with reference to specific behaviors.

The narrator reflects on a situation in which they needed rapid assistance; in order to complete an application for a scholarship, the author needed a transcript from the Georgetown’s registrar, but the narrator was not physically on campus. They appealed to a classmate, who rose to the occasion, as described in the following excerpt.

Eine Freundin von mir, Suna, war sehr bereit mir zu helfen. In unter zwei Uhren, nahm sie mein officialische Transcript aus dem Büro des Registrars und gab mir ein Scan daran ein. Gott sei Dank, die letzte Ort für das Scholarship war komplett! (. . .) Nachdem sie mir hilfte, war ich viel mehr dankbar für ihr. Ich lernte dass, war und ist sie sehr bereit und glücklich Menschen zu helfen. Sie war sehr nett über das, und wollte kein Geld oder andere Bezahlen. Suna ist ein Beispiel des gutes Freunds. Sie hilfte mir in einem angstvol Zeit, und war sehr nett darüber. Man soll Menschen als ihr nahe zu haben.

‘A friend of mine, Suna, was very ready to help me (+JUDGMENT: propriety). In under two hours (+JUDGMENT: capacity), she took my official transcript out of the registrar’s office and gave me a scan of it. Thank God, the last place[?] for the scholarship was complete! (. . .) After she helped me, I was much more thankful for her. I learned that she was and is very ready and happy to help people (+JUDGMENT: propriety). She was very nice (+JUDGMENT: propriety) about this and wanted no money or other payment (+JUDGMENT: propriety). Suna is an example of a good friend (+JUDGMENT: propriety). She helped me in a stressful time

203

(+JUDGMENT: propriety) and was very nice (+JUDGMENT: propriety) about it. One should [] have people like her nearby (+JUDGMENT: propriety).’

The narrator makes it clear that we are to interpret Suna’s actions as highly moral by explicitly evaluating her as ein Beispiel des gutes Freunds ‘an example of a good friend,’ a person whom ‘one should have nearby.’ These categorical evaluations follow a description of the actions that make Suna such a praiseworthy person: She offers help quickly, is happy to do so, is effective in her efforts, and does not request any financial reward. From these evaluations, we can glean much more about the narrator’s own values than we could from a Level I text; beyond merely describing one situation, here the narrator asserts the overarching value of offering aid without expecting a reward. To the extent that the reader shares these values, the narrator has established a degree of affiliation between them by relating the actions of a friend that the reader is expected to laud.

Level II writers also use JUDGMENT more often to evaluate characters other than the author and the primary friend. Other characters are evaluated with JUDGMENT in 29 cases, as opposed to zero in Level I. These evaluations often serve as a dual purpose; they not only appraise the character in question, but also provide contrast to another figure in the narrative. In the following excerpt from Text fi11, the author praises her current roommates by contrasting them favorably with her previous one, highlighting her former roommate’s deficiencies as a means of lifting up her new friends.

Wie ich letztes Feiern gesagt habe, hatte ich eine sehr, sehr dreckig Mitbewohnerin. Sie duschte nicht gerne und sie hat nie Wäsche gemacht. Meine groesstes Probleme war dass diese Maedchen war eine Freundin von meinem Internat, so es war sehr schwer, weil ich nicht sie schlecht fühlen wollte. Heute habe ich zwei sehr nett Mitwebohnerin (. . .). Ich bin sehr glücklich, weil sie sind so sauber wie ich und jedes Mal, wenn sie eine Feier haben, fragen sie mich, wie ich mit ihnen zu kommen mochte.

‘As I said during the last break, I had a very, very dirty (-APPRECIATION: quality: aesthetics / - JUDGMENT: propriety) roommate. She never showered (-JUDGMENT: propriety) and never did the laundry (-JUDGMENT: propriety). My biggest problem was that this girl was a friend

204

from my boarding school, so it was very difficult, because I didn’t want to make her feel bad. Today I have two very nice (+JUDGMENT: propriety) roommate[s] (. . .). I am very happy, because they are as clean as I am (+JUDGMENT: propriety) and every time they have a party, they ask me [whether] I would like to come with them (+JUDGMENT: propriety).

She first introduces her previous roommate as one who does not adhere to standards of cleanliness, who is dreckig ‘dirty.’ As mentioned in Chapter 3, such physical descriptions are typically coded as APPRECIATION: quality: aesthetics. However, this description is followed by additional examples of the former roommate’s objectionable behavior and lack of consideration.

The author confirms this impression when she describes her groesstes Probleme ‘biggest problem’ as the roommate’s belonging to her school. If this is the biggest problem, the other factors mentioned are implied to be problems as well. As the narrative concludes, the narrator secures two new roommates, who are described primarily in contrast to the first roommate.

Again, it would not necessarily be clear that the new roommates’ cleanliness is meant to be such a positive evaluation (for example, ‘neat freak’ is not usually a term of endearment), had the narrator not drawn such a contrast in the form of the dirty roommate. By mentioning the first roommate as an example of unacceptable roommate behavior, the narrator is able to praise their new roommates more economically.

In general, then, we see two trends as we move from Level I to Level II — a broadening of targets of JUDGMENT and an increase in implicit evaluation of particular traits and characteristics, not merely individuals.

5.1.3.3 JUDGMENT Targets in Level III Texts

Level III texts include relatively more frequent appraisals of the narrator and characters other than the primary friend. Judgments of the self, as in previous levels, typically involved the

205

narrator’s abilities to cope with the situation they described. Table 26 displays the frequencies with which each character is targeted for JUDGMENT in Level III texts.

Table 26. Frequency Counts of JUDGMENT Tokens at Level III. normality propriety veracity capacity tenacity self 1 8 2 20 10 friend 1 17 2 10 2 family 0 10 0 0 1 other 1 19 2 10 2

Just as Level II narrators more often evaluated secondary characters in order to praise their friend by contrast, Level III narrators sometimes used their praiseworthy friends as yardsticks by which to judge (and sometimes find wanting) their own behavior. In Example (41), for instance, the narrator describes their Georgetown classmates as intelligent and capable by way of casting doubt on their own ability:

(41) Wie kann ich antrete mit alle andere Studenten, die klüger als ich sind? ‘How can I compete with all the other students, who are smarter than I am (+JUDGMENT: capacity)?’

Level III texts include several other instances in which the narrator evaluates their own or others’ moral standing implicitly, using the examples of others as contrasts. Examples of these contrasts describing a friend as very aktiv relative to the narrator’s own unsure attitude, describing new friends as treuer ‘more loyal’ in comparison to their former associates, and (in the example below), praising a friend as positiv to throw their own poor attitude into relief.

In Text wa9, selections of which are presented in the following excerpts, the overarching theme is the author’s doubt in their own abilities and their efforts to improve, which are clarified

206

by comparison with a close friend. The author feels discouraged and overwhelmed at the outset of their tale:

In dieser Zeit bin ich als ein Person gewachsen. Aber zuerst hatte ich viele Probleme. Solche Probleme hatte ich nie bevor begegneten: ich hatte keine Freunde, ich kriegte schlimme Noten, ich hatte ein schlechte Wohnungssituation.

During this time I grew as a person. But first I had many problems. I had never before encountered such problems: I had no friends, I received bad grades, I had a bad living situation.

The narrator then introduces her friend Miranda as something of a foil; Miranda exemplifies the qualities the narrator wishes they themselves possessed.

Spezifisch hat meine Freund Miranda mir geholfen. (. . .) Sie ist immer positive und hat ein Geburtstagsfeier für mich geplant.

‘Specifically, my friend Miranda helped me (+JUDGMENT: propriety). She is always positive (+JUDGMENT: propriety) and planned a birthday party for me.

Reflecting on Miranda’s actions, the narrator makes specific commitments regarding the behaviors they wish to improve.

Um meine Wohnungsituation zu verbessern, habe ich jeden Tag meine Mitbewohner verziehen, damit ich ein verständlicher Person werden kann. Ich kann besser werden. Langsam, aber trotz besser.

‘In order to improve my living situation, I forgave my roommates every day, so that I can become a more understanding (+JUDGMENT: propriety) person. I can become better (+JUDGMENT: propriety). Slowly, but still better.

While Miranda does not solve the narrator’s problems, Miranda provides an example of a positive attitude that inspires the narrator to treat her roommates with the same generous intent.

The narrator also implicitly criticizes themself by heaping praise on Miranda for qualities that the narrator does not always show.

These texts also include examples of secondary characters being evaluated as comparisons or contrasts to the primary friend character. In Text fsa1, excerpted here, the narrator is disturbed by a man who seems to be stalking them while they attempt to study. The

207

narrator enlists a nearby friend, and both approach a security officer. The entire passage contributes to the impression of the friend as both helpful and capable — not only through descriptions of the friend, but also in the contrasting depictions of the narrator and the security officer.

Ich wär sehr glücklich dass sie dort mit mir war, weil sie immer weiss, wass man tuen soll. Wir haben beide zu die Securität des Bibliotek gegangen und zu ihnen die Situation erzählen. Die man von Securität hat nichts zu dieser Situation reagiert. Er hat nur gesagt, dass ein solche Situation immer passiert. Meine Freundin aber hat mehr mit ihn gesprochen und zunächst hat er die Securität des Universität geruft. Ich habe nicht von dieser Sprach von ihnen gehört aber bald hat die Securität gekommen und habe dieser Mann versucht. (. . .) Meine Freundin war sehr halfbar und ich wusste nicht was wir tuen sollten. Sie war sehr aktiv in dieser Situation und ich war sehr hoffnungsfoll dass sie mir helfen konnte.

‘I was very happy that she was there with me, because she always knows what to do (+JUDGMENT: capacity). We both went to library security and told them the situation. The man at security did not react to this situation (-JUDGMENT: propriety). He only said that such situations always happen (-JUDGMENT: propriety). But my friend talked with him more (+JUDGMENT: tenacity) and then he called university security. I didn’t hear anything of this conversation, but soon security came and searched for the man. (. . .) My friend was very helpful (+JUDGMENT: propriety) and I didn’t know what we should do (-JUDGMENT: capacity). She was very active (+JUDGMENT: tenacity) in this situation and I was very hopeful that she could help me.’

It is expected that security officers would respond to a complaint of an intruder; this officer’s response, therefore, invokes implicit JUDGMENT against him. The narrator describes a friend willing to do the things the security officer will not — to advocate on her behalf and call for additional help.

In the following longer excerpt from Text ma4, the narrator introduces the reader to three friends, two of whom have recently treated the narrator quite coldly, and one of whom has become by contrast a valued source of support. In this text, many instances of JUDGMENT

208

perform additional ‘work’ by evaluating not only their target, but also (implicitly) the character with whom they are juxtaposed.

During the Orientation Stage, after a reference to happier times, the narrator described a period of alienation from her former friends, who seem (from the narrator’s perspective) to have suddenly excluded them from the friend group.

Es ist als ob, sie ploetzlich entscheidet haben, dass ich ihnen nicht mehr gefaellt habe. Jetzt sagen wir einander kaum ein Wort, obwohl sie immer noch sehr eng sind.

‘It is as though they suddenly (-JUDGMENT: propriety / others) decided that they didn’t like me. Now we hardly say one word to another, although we used to be quite close.’

After describing the cold, distant actions of her former friends, the narrator provides an immediate contrast, in the form of their new friend Sarah, who goes out of her way to encourage and support the narrator.

Deshalb war diese Semester fuer mich schwieriger als der letzte. Der Anfang, als sie begonnen haben, mit mir kalt zu behandeln, war der schwierigste. Meine Freundin Sarah hat mir waehrend dieser Zeit viel geholfen. Ich habe mich gefuehlt, dass ich keine andere Freunden hatte, und das ich bei Georgetown allein war, aber sie hat mir gezeigt, dass ich viele andere Leute kennte, die mich unterstuezen wollten. Sie hat mir die Mut gegeben, andere Schuelern zu treffen, und hat mich zugehoert, als ich mich meine Probleme mit Dylan und Emily beschwert habe. (. . .)

‘That’s why this semester was harder for me than the last. The beginning, when they started to treat me coldly (-JUDGMENT: propriety / others), was the hardest. My friend Sarah helped me (+JUDGMENT: propriety / friend) during this time. I felt as though I had no other friends, and that I was alone at Georgetown, but she showed me that I knew many other people who wanted to support me (+JUDGMENT: propriety / friend). She gave me the strength to meet other students and listened to me (+JUDGMENT: propriety / friend) while I complained about my problems with Dylan and Emily. (. . .)’

The mention of other people who would support the narrator can be read as an implicit indictment of her former friends; the examples of Sarah’s generosity throw the more distant actions of the narrator’s former friends into even sharper relief. As the narrative moves toward

209

the Resolution stage, the narrator provides further juxtapositions of her past and current circles of friends.

Jetzt haben Dinge zwischen ich und meiner alten Freunden nicht verbessert. Wir reden fast nicht und es tut mir immer noch weh, sie zusammen zu sehen, weil ich erinnere mich an die Zeit, wann wir alle zusammen waren. Die waren gute Zeiten. Aber ich weiss auch, dass ich sehr gluecklich bin, andere Freunden jetzt zu haben, die mir treuer sind. Ich weiss mindestens, dass Sarah immer bei mir standen wird.

‘Now things between me and my old friends have not improved. We almost never talk, and it hurts me to see them together, because I remember the time when we were all together. They were good times. But I also know that I am very happy to have other friends who are more loyal (+JUDGMENT: propriety/ others / -JUDGMENT: propriety/ others) to me. I know at least that Sarah will always stand with me (+JUDGMENT: propriety/ friend / -JUDGMENT: propriety/ others).’

The comparative treuer ‘more loyal’ is ostensibly a token of JUDGMENT that positively evaluates the narrator’s current friends; however, it can only do its evaluative work by contrasting these friends with the previous ones. Likewise, the narrator’s concluding sentence both praises Sarah for her steadfastness and, by omission, condemns their previous friends for not standing by them.

From lower to higher proficiency levels, then, we see a trend in the use of Judgment: In more advanced texts, the narrators tend to use JUDGMENT not only to evaluate a single character, but instead to set up oppositions between one character offered as a moral exemplar and others

(sometimes the narrator, sometimes secondary characters) whom the narrator implicitly evaluates less generously.

210

5.2 ATTITUDE within Narrative Stages

Apart from cross-level differences in the use of AFFECT, APPRECIATION, JUDGMENT, one of the clearest differences between curricular levels involved the functions of these ATTITUDE types in the context of narratives.

An author’s decision to vary the ATTITUDE types they use from stage to stage may serve a variety of purposes. Shifts in patterns of ATTITUDE usage can coincide with a shift in narrative phase in order to appeal to the audience’s emotions; Macken–Horarik et al. (2018) offer the example of the narrative of an Aboriginal woman’s life, in which the Complication phase is characterized not only by a contrastive conjunction, but also by an increase in amplified markers of AFFECT (p. 161). The ATTITUDE patterns of a later narrative stage may also allow an author to cast previous stage in a different light (Macken–Horarik & Morgan, 2011).

In general, authors tended to use APPRECIATION to describe a contrast between the narrator’s initial and final situations, with the scope broadening out to characters beyond the narrator at higher proficiency levels; accordingly, it was most common in Orientations and (at lower levels) Codas. JUDGMENT tokens came into play more often in later stages as authors used them to evaluate the helpfulness of the main characters, and (at higher levels) to reflect on the traits they most prized in those characters. It was difficult to draw as clear a conclusion regarding

AFFECT — except in the case of L1 texts, where authors often used AFFECT in Codas in order to provide further evidence of their ethical evaluations.

5.2.1 APPRECIATION in Narrative Stages

APPRECIATION within lower-level texts served to describe the narrator’s initial situation, and to set up a baseline for comparison with the narrator’s circumstances at the end of the tale.

211

Particularly at Levels I and II, it was most prevalent during Orientation and sometimes Coda stages.

In beginner texts, APPRECIATION tokens tended to be overtly positive or negative; in the

Orientation excerpted in Example (39), the narrator used APPRECIATION in order to give a broad- strokes overview of the events they were about to describe:

(39) Es war nicht ein gutes Mal, aber meine Freundinen und ich denken es war sehr lustig. ‘It was not a good (+APPRECIATION: quality) time, but my friends and I think it was very funny (+APPRECIATION: impact)’

Level II writers used APPRECIATION most often in the Orientation stage (in terms of raw frequency counts across all texts; after taking into account individual texts with disproportionately high frequencies of AFFECT or JUDGMENT, APPRECIATION also emerged as the most consistently used marker of an Orientation stage at this level) and generally continued to use it throughout their narratives; APPRECIATION tokens were distributed evenly across Level II texts, with almost every text including at least two such instances. These texts typically begin with a description of the college atmosphere as either positive or stressful (or both), or

Washington, D.C.’s weather as dark and rainy (a surprisingly common preoccupation among these writers). These comments on the external situation nearly always reflected that atmosphere’s impact on the internal world of the narrator, rather than any other character.

Example (40), which includes APPRECIATION tokens describing the university environment, was characteristic:

(40) Meine Leben in Georgetown ist oft nicht so interessant. Ich habe meine Klassen, meine Texte, meine Briefe, und so weiter. Ich habe keine große Angst über die Alltagsdingen von ein Student. Aber ich muss sagen, dass ein Aspekte meines Lebens im Universitaet ein bisschen schwer ist ‘My life in Georgetown is often not so interesting. I have my classes, my texts, my letters, and so on. I don’t have a lot of worries about the daily things of a student. I must say that one aspect of my life in the university is a little difficult’

212

APPRECIATION tokens in Level II Orientation stages fell into the following categories with regard to the ideational element that they evaluated: Stress of university life, relevance of the event about to be described, in Examples (41)–(42), the narrator’s relationships (43), physical description of a major character (44), and the narrator’s reasons for writing the story (45).

(41) Nachdem ich ein beeindruckendes Ereignis an meiner Universität erlebt habe, möchte ich es dir erzählen ‘After I experienced an impressive event at my university, I want to tell you about it’

(42) Ich habe eine lustige Geschichte für sie! ‘I have a funny story for you!’

(43) Wir würde Uber die tag gesprochen und ich mochte gern dieses Konversation ‘We would [talk] about the day and I liked these conversation[s]’

(44) Sie ist groß und elegant ‘She is tall and elegant’

(45) so ist es mir wichtig, dass Sie über sie wissen ‘so it is important to me that you know about her’

In almost every text from Levels I–III, APPRECIATION contributed most of the evaluative weight of the Orientation stage — though at higher levels the trend was toward relatively greater reliance on AFFECT. The following excerpt provides an example of how one writer at Level III used APPRECIATION to describe the conditions that led to a change in their emotional state.

Dieses Jahr hier in Georgetown war besser als ich vorgestellt hatte. Ich habe viele neue Freunde getroffen und tolle Klassen gehabt. Letzte Semester waren meistens meinen Klassen ganz leicht, deswegen war es nicht zu schwierig ein gute Note zu bekommen. Aber diese Semester hatte ich ein kleines Problem mit eine Klasse über Politik in Südamerika. Mein Südamerika Klasse ist ein “elective”, dass meine Freundin Maria mir vorstellen hatte. Sie ist eine “junior” und ich bin noch eine “freshman”, aber diese Klasse ist niveau 400, das bedeutet es ist ein von die schwierigste Klassen hier in Georgetown. Natürich wusste ich das nicht während “preregistration”.

‘This year at Georgetown was better (+APPRECIATION: quality: manageability) than I had imagined. I met many new friends (+AFFECT: in/security: dis/quiet) and had cool (+APPRECIATION: quality: manageability) classes. Last semester most of my classes were very easy (+APPRECIATION: quality: manageability), and for that reason it was not too hard (+APPRECIATION: quality: manageability) to get a good (+APPRECIATION: social valuation: consequence) grade. But this semester I had a little problem (-APPRECIATION: social

213

valuation: consequence) with a class about politics in South America. My class on South America is an elective that my friend Maria introduced me to. She is a junior, and I am still a freshman, but this class is level 400 (-APPRECIATION: quality: manageability), which means it is one of the hardest (-APPRECIATION: quality: manageability) classes here at Georgetown. Of course (+JUDGMENT: normality), I didn’t know that during preregistration.’

In the first part of this Orientation, the author describes their environment using explicit positive APPRECIATION, mostly drawing from the field of schooling; in this context, it is understood that ‘easy’ is a positive attribute. Both the overall atmosphere in Georgetown and individual classes are described within this context. We can expect that this will contribute to the narrator’s sense of positive AFFECT: security, which is also implied by her statement that she has found a supportive community. The Orientation continues as follows.

Als ich zu erste Klasse gegangen bin, habe ich gefühlt dass meinen Klassenkameraden wussten schon mehr über Südamerika und Politik als mir. Meisten hätten schon anderen Klassen über die Region gehabt und konnten viel besser argumentieren. Aber obwohl es sehr (intimidating) war, dachte ich die Klasse sehr Interessant war, deswegen wollte ich es lassen. Die Problem ist mit die erste Midterm gekommen.

‘When I went to the first class, I felt that my classmates knew more than me about South America and politics (-AFFECT: in/security: dis/quiet; +JUDGMENT: capacity). Most had already had other classes about the region and could argue much better (+JUDGMENT: capacity). But although it was very intimidating (-APPRECIATION: impact), I thought [that] the class was very interesting (+APPRECIATION : impact), and for that reason I wanted (+AFFECT: dis/inclination: desire) to leave it. The problem (-APPRECIATION : social valuation: consequence) came with the first midterm.’

JUDGMENT does creep into the Orientation at this point, but mostly in order to explain the author’s growing feeling of inadequacy and insecurity (habe ich gefühlt dass meinen

Klassenkameraden wussten schon mehr über Südamerika und Politik als mir); they flag their classmates’ advanced knowledge and abilities, and they implicitly evaluate their own capacity negatively as a way of explaining the shift in their AFFECT from positive (security) to negative

(insecurity). The author then returns to adjectival realizations of APPRECIATION that explain the affective state of interest in the class with which they close the Orientation.

214

The Complicating Action stage in this text drew more heavily on JUDGMENT as the narrators’ friends or antagonists entered the picture and were evaluated. At higher levels,

APPRECIATION was comparatively rare in this stage; the narrators may have used tokens of

APPRECIATION, but the primary evaluative prosody in these stages addressed either the narrator’s assessment of their own or others’ behavior, the narrator’s changing emotional reactions to the events around them, or both.

APPRECIATION is often used in the Coda to describe the narrator’s new external situation

(most often at Level II). One purpose of a Coda is to reflect on how the situation has changed;

APPRECIATION proves a useful vehicle for this purpose, particularly when focusing on situational aspects that affect the narrator. APPRECIATION was much less common in Codas at Level III, giving way to JUDGMENT.

5.2.2 JUDGMENT in Narrative Stages

Level I texts included relatively few instances of ATTITUDE at any particular stage. Two texts primarily use JUDGMENT in their Orientation stages, in order to introduce and appraise an important character who will be further evaluated later, as in Examples (46)–(47).

(46) Ich mogen meiner Zimmerkollegin, aber manchmal wir widersprechen ‘I like (+JUDGMENT: propriety) my roommate, but sometimes we disagree’

(47) Ich habe eine wunderbare Freundin ‘I have a wonderful (+JUDGMENT: propriety) friend’

At lower levels, JUDGMENT was generally rare in the Orientation stage.15 It might be employed a few times to describe a character who will be prominent later in the story, as in the

Example (48).

15 One text (Text wa5) that did include ample JUDGMENT throughout the story did not in fact follow the stages of a narrative or describe a specific occurrence; instead, it described the narrator’s relationships with mentors and friends at various times in their life.

215

(48) Letzte woche ist etwas schwierig passiert, und ich will immer ernst und ehrlich mit euch sein. Sie haben mein freund, Bri, schön getroffen. Sie ist einfach ein wunderbares freund, und dafür bin ich ganz glücklich. Also, letzte woche war ganz schlecht. ‘Last week something very difficult happened, and I always want to be forthright and honest with you. You’ve already met my friend Bri. She is simply a wonderful (+JUDGMENT: propriety) friend, and I am very happy for that. So, last week was very bad.’

However, JUDGMENT more often served to evaluate the characters who played starring roles in the Complicating Action. The Complicating Action stage of Text ma2 provides an example of this trend. In the section excerpted here, the narrator describes their own response to a challenging exam.

Ich wusste dass die Prüfung nicht leicht wurde, die Professorin hat das klar erzählt. Deswegen habe ich sehr viel dafür gelernt. Fast zwei Woche früher habe ich mein “study guide” begonnen und alle Texten noch einmal gelesen. Ich war sicher dass ich alle gemacht hatte als es moglich war. Als ich meine Note zurück bekomme, war ich ganz enttäuscht. Es war ein C-. Ich fühlte mich als ob ich nicht intelligent genug für Georgetown war.

‘I knew that the test would not [be] easy (-APPRECIATION: quality: manageability), the professor had clearly (+JUDGMENT: capacity) explained that. For that reason I studied for it a lot. Almost two weeks beforehand I had begun my study guide (+JUDGMENT: capacity) and read all the texts one more time (+JUDGMENT: tenacity). I was sure (+AFFECT: in/security: dis/quiet) that I had done everyyhing possible (+JUDGMENT: tenacity). When I receive[d] my grade, I was very disappointed (-AFFECT: dis/satisfaction: dis/pleasure). It was a C- (- APPRECIATION: social valuation: consequence). I felt as though I was not intelligent (- JUDGMENT: capacity) enough for Georgetown (+AFFECT: in/security: dis/quiet).’

The first test provides an inflection point in this narrative. Aware that it will be a challenge, the author draws on their academic skills, which they at first positively evaluate with examples of their advance planning (+capacity; Fast zwei Woche früher habe ich mein “study guide” begonnen) and effort (+tenacity; alle Texten noch einmal gelesen). Receiving their grade, however, precipitates a shift in their AFFECT; they both express displeasure (ganz enttäuscht) and repeat the insecurity (Ich fühlte mich als ob ich nicht intelligent genug für Georgetown war) that they first mentioned in the Orientation. They also negate their previous evaluation of their own ability.

216

The Resolution stages at this level typically draw on both AFFECT and JUDGMENT to describe the author’s reaction to the event’s conclusion, and/or their reflections on the motivations of those who helped or hindered them. The Resolution stage of Text ma2, excerpted here, is representative of other texts at this level in this regard.

Ich war wirklich traurig aber glücklicherweise war meine Freundin Maria in die Klasse, und sie hat mir geholfen. Sie hat mir alle ihren Noten seit freshman year gezeigt damit ich wusste dass als wir älter werden, werden wir mehr gewöhnt mit die Georgetown system. Ihre Noten ware ähnlich wie meine und sie (assured me) dass es wird immer leichter mit zeit.

I was very sad (-AFFECT: un/happiness: misery/cheer), but fortunately (+COVERT AFFECT) my friend Maria was in the class, and she helped me (+JUDGMENT: propriety). She showed me all her notes since freshman year so that I knew that as we got older, we got more used to the Georgetown system (+JUDGMENT: capacity). Her notes were similar to mine, and she assured me that it always gets easier (+APPRECIATION: quality: manageability) with time.

In Level II, writers sometimes used JUDGMENT in the Coda stage to offer a final reflection on their friend’s helpfulness (JUDGMENT appeared fairly prevalent in Codas at this stage in terms of raw frequency counts, but these numbers were slightly warped by a higher-than-usual incidence of Judgment tokens in one text (Text fi9)). This reflection might underline the particular way in which the friend was helpful, or the qualities they showed, thereby indicating the narrator’s beliefs on the subject of friendship. In the texts of higher-proficiency writers, the Coda stage tended to function as a reflection on the valued characteristics of people and behaviors, rather than on aspects of a situation or environment and their impact on a single character; accordingly, at these higher levels, JUDGMENT was relatively more common compared to APPRECIATION.

217

5.2.3 AFFECT within Narrative Stages

AFFECT proved a more common ATTITUDE type among L1 texts across all narrative stages, with the most marked contrasts to other levels appearing in the Orientation and Coda stages.

Text xn4 provides an interesting example of this trend, as it contains a particularly long

Orientation stage. The author finds themself unhappy with their chosen major and out of place among their classmates. They begin the Orientation with a description of their emotional state, drawing on a series of -AFFECT: un/happiness and -AFFECT: in/security tokens. Instances of other

ATTITUDE tokens are also typically related to emotions, such as the tokens of APPRECIATION and

COVERT AFFECT in Examples (49) and (50).

(49) Es hat mir von Anfang an überhaupt nicht gefallen ‘From the beginning, I didn’t like it (-APPRECIATION: impact) at all’

(50) Ich habe das Gefühl, dass das Mathematik-Studium mein Leben an der Universität stark beeinflusst, und zwar negative. ‘I have the feeling that studying math influences (-COVERT AFFECT) my life at college strongly — negatively so.’

The first section of the Orientation conveys the author’s feelings of stress and insufficiency as they grapple with difficult math topics, leaving room for an empathetic reaction on the part of the reader. The author then shifts their tone with an extended reflection on the failings of their classmates, whom they evaluate in almost exclusively negative terms in contrast to their friends at home. (For an analysis of these JUDGMENT moves, see section 4.1.3 above.) These classmates strike the author as unsociable, and crucially they do not assist the author when they have trouble with their workload. The classmates’ unavailability prompts the author to react viscerally to the prospect of going to campus: …sodass ich schon Bauchschmerzen kriege, wenn ich dorthin fahren muss ‘…so that I get a stomachache when I have to drive there.’

218

The author further contrasts their classmates in math with those in their second major,

English, whom they evaluate using positive JUDGMENT: propriety. This mention of supportive friends turns quickly to negative emotion, however, as the author complains that they have little time to spend with their English classmates. The Orientation of this text is excerpted here.

…weil ich immer für Mathe lernen muss. Das macht mich sehr traurig, weil ich mich dann schnell ausgeschlossen fühle. Ich möchte wirklich gerne das Fach zum nächsten Jahr wechseln, aber ich weiß nicht, welches Fach ich stattdessen machen möchte. Ich möchte euch nicht enttäuschen, sondern ich möchte, dass ihr stolz auf mich seid. Ich will auch nicht, dass ihr euch Sorgen um mich machen müsst.

‘…because I always have to study for math. That makes me very sad (-AFFECT: un/happiness), because I then quickly feel shut out (-AFFECT: in/security). I would really like to (+AFFECT: dis/inclination) switch majors next year, but I don’t know (-AFFECT: in/security) which major I want to do instead. I would not want to disappoint you (-AFFECT: dis/inclination); instead I want (+AFFECT: dis/inclination) you to be proud of me (+AFFECT: dis/satisfaction). I also don’t want (-AFFECT: dis/inclination) you to have to worry about me (-AFFECT: in/security).’

The author closes the Orientation with a description of the ways their emotions combine to create a miserable emotional state; their conflicting desires to teach and to work with children, and also to avoid disappointing their parents, have created an internal situation of uncertainty and despair. Strikingly, this lengthy Orientation contains very little APPRECIATION; in addition to the token of APPRECIATION: impact mentioned above, Examples (51) and (52) represent the only instances.

(51) Alle haben immer gesagt, dass die Jahre an der Universität die schönsten Jahre ihres Lebens waren ‘Everyone always said that their university years were the best (+APPRECIATION: valuation: consequence) of their life’

(52) Die Fächer in Mathematik sind viel zu schwer für mich ‘The math subjects are much too hard (-APPRECIATION: quality: manageability) for me’

This author uses APPRECIATION less often in general than they use other types of

ATTITUDE. However, APPRECIATION does appear somewhat more often in their Complicating

Action and Resolution stages, so it is not the case that the author simply never evaluates the

219

value of situations and things. Even without using much APPRECIATION in the Orientation, the author fulfills the expectations of this stage of the genre; they are able to set the stage for an emotional journey and argue for a standard of behavior among university students.

Other L1 texts included significantly shorter Orientation stages, often briefly using

APPRECIATION to comment on the nature of story to come (e.g, heute war in meiner Universität ein doch eher düsterer Tag…) and then launching into the Complicating Action.

Text xn4 concludes with one such AFFECT-laden Coda. The author has grappled with stress concerning their field of study, and their friends have supported them both by assuring them of their capabilities and suggesting a possible course change.

Ich habe wieder eine Perspektive und meine Kopfschmerzen sind seitdem auch viel weniger geworden. Ich habe keine schlechten Gedanken und Gefühle mehr, wenn ich an die Universität nächstes Jahr denke, und ich freue mich richtig auf das Geographie-Studium. Ich bin zwar auch ein bisschen nervös, weil ich wenig Vorwissen habe, aber das sei wohl nicht schlimm und ich könnte das schnell nachholen. Mir geht es jetzt viel, viel besser!

‘I now have perspective again (+AFFECT: in/security) and my headaches have also lessened since then (+AFFECT: in/security). I don’t have any more bad thoughts and feelings (+AFFECT: in/security) when I think about university next year, and I’m really excited (+AFFECT: dis/inclination) about the study of geography. True, I’m a little nervous (- AFFECT: in/security), but that shouldn’t be so bad (-APPRECIATION: impact) and I could catch up quickly. I’m doing much, much better (+AFFECT: un/happiness)!’

This passage relies almost entirely on AFFECT to summarize the narrator’s new situation, and those expressions of AFFECT refer only to the narrator’s internal state. At the same time, the entire paragraph could also be construed as an implicit evaluation of the author’s friends; they effected this change in the narrator’s emotional state by encouraging her to attend a geography class.

Recall that the prompt for this writing task asked participants to narrate an experience in which a friend helped them solve a problem or recover from a difficult time. The action of narrating allows the writers to use evaluation in a variety of ways — to give their impression of

220

their environment; to delve into the emotional states they experienced at a time of crisis; and to rate their friends’ character as expressed during crunch time. Recall also that each of these evaluative moves has an interpersonal function; the narrator’s expression of anguish may help generate pathos and strengthen their connection with the reader, and their evaluation of their friend according to an ethical yardstick can evoke beliefs and expectations that they and their readers have in common.

In a broad sense, the narrators of all of these stories tell a story not only about a particular friend, but about the nature of friendship in general. By repeatedly praising a positively- identified character for their generosity, the narrator emphasizes the importance of generosity in general (O’Donnell, 2014); this, the narrators implicitly argue, is how one displays good friendship. While explicit JUDGMENT of the narrator’s friend is relatively less prevalent in these

L1 texts, the narrators achieve a similar purpose — evaluating their friends as more or less helpful — by describing the internal emotional effects that their friends’ actions wreak.

Another L1 text approaches this point more concisely in Example (53), drawn from the

Coda stage of this text.

(53) Euer Sohn war am verzweifeln und selbst, sei es durch Verzweiflung oder Ungeschick, nicht in der Lage sein eigenes Problem zu lösen. Manchmal muss man wohl akzeptieren, dass andere Menschen gute Helfer sein können.

‘Your son was despairing (-AFFECT: un/happiness) and, whether because of despair (- AFFECT: un/happiness) or inadequacy (-JUDGMENT: capacity), not able to solve his own problem (-JUDGMENT: capacity). Sometimes you do just have to accept that other people can be good helpers (+JUDGMENT: capacity/propriety).’

The Coda of this text mixes AFFECT and JUDGMENT (of the narrator, not their friend) with a final tag including a qualified JUDGMENT of the friend character. Each instance of negative AFFECT

221

describing the narrator’s previously dejected state gives additional weight to the positive effects his friend’s actions have had on him.

5.3 GRADUATION in Study 2

Initial results from Study 1 suggested that learners’ use of GRADUATION might prove a fruitful area for more detailed analysis. In particular, writers in Study 1 tended to employ a broader vocabulary when using GRADUATION via infusion compared to GRADUATION via isolation, and more advanced writers tended to use infusion relatively more frequently. I speculated that if Study 2 texts displayed the same trend, this might suggest a use for

GRADUATION analysis as a tool for instructors interested in their students’ writing development in narrative genres; instructors could call attention to students’ use of GRADUATION via isolation and juxtapose these with examples from model texts of GRADUATION via infusion. This is hardly a new suggestion; highlighting students’ overuse of terms like ‘very’ and ‘really’ has long been a tool in a writing tutor’s toolbox (Strunk & White, 1959). Viewing a text through the lens of

GRADUATION, however, might support an overall effort to assess learner texts holistically, rather than simply flagging individual errors to improve.

In any event, texts from Study 2 did not contain the same trend toward greater use of

GRADUATION via infusion in more texts. Instead, this larger sample contained more unique instances of GRADUATION via infusion at higher curricular levels (where the same tokens were more often used repeatedly at lower levels), as well as a greater awareness of nominalization and, perhaps most relevant to a discussion of implicit evaluation, an increased reliance on

GRADUATION: FOCUS as opposed to FORCE. Table 27 displays individual instances of

222

GRADUATION via Infusion in Level II and III texts, as well as the frequencies with which those particular terms occurred.

Table 27. Instances of GRADUATION via Infusion in Levels II and III. Level II Level III

Realizations Appearing in Both Levels

Realization Frequency Realization Frequency besser 14 besser 1 am besten / beste 7 (nicht die) beste 1 liebe/geliebt 6 liebst/liebte 2 fantastisch 2 fantastisch 1 hoffenlos 1 hoffnungslos 1 schlechter 1 das schlechteste 1 Realizations Appearing at Only One Level Realization Frequency Realization Frequency wütend 3 schwieriger/schwierigste 4 habe keine Ahnung / keine geschrien, leichter, Panik, schrecklich, 2 2 Idee, überwällt schwerer ein Beispiel des guten, ein ein Alptraum, angegriffen, Desaster, erschrecklich, ausgezeichnet, effizienter, entfremdet, kalter, eine Katastrophe, 1 Lieblingsfreunden, schlimm, treuer, 1 letzte, lustigste, perfekt, Unheil, verständlicher, wie den Tod, schwierigsten, shockte, wunderbar trauriger,

Unique Tokens 21 24 Total Realizations 49 33

Among tokens of GRADUATION via infusion, there was a slight difference between Levels

II and III in the variety of different terms used. If lexemes of the same lemma are counted as the same term, all Level II texts contained 21 unique tokens of infusion, while Level III texts

223

contained 24. This disparity is perhaps not immediately striking, though it increases if we discount cognates such as ‘Katastrophe’ and ‘fantastisch.’ It is also worth noting that Level III texts included fewer instances of GRADUATION via infusion overall; thus, a greater proportion of those instances were unique.

5.3.1 Targets of GRADUATION

The targets of GRADUATION, however, varied more widely at each level, with more advanced writers modifying a process much more often. As Table 28 indicates, the percentage of tokens modifying a process increased from Level II onward (the number of GRADUATION instances among Level I texts was low overall).

Table 28. GRADUATION Targets in Study 2. Level I Level II Level III L1

FORCE Type Intensification 19 90.48% 164 91.11% 153 72.51% 86 81.90% Quantification 2 9.52% 16 8.89% 58 27.49% 19 18.10% Total 21 180 211 105

FORCE Target Intensification 18 85.7% 154 85.56% 132 62.56% 71 67.62% of Attribute Intensification 1 4.76% 10 5.56% 21 9.95% 15 14.29% of Process Quantification 0 0.00% 9 5.00% 33 15.64% 4 3.81% of Thing Quantification 2 9.52% 7 3.89% 25 11.85% 15 14.29% of Process

224

L1 writers, for instance, were relatively more likely to use GRADUATION: quantification to indicate the endurance of processes throughout time, as in Example (54).

(54) weil ich immer gerne zur Schule gegangen bin ‘because I always (GRADUATION: quantification: process: upscale) liked to go to school (+AFFECT: dis/inclination)’

Level III writers also use quantification relatively more often than lower-proficiency writers. This form of GRADUATION often modifies a nominalization of some kind, as in Example

(55).

(55) Ich habe dir gesagt während der Sommer, dass ich habe gefüllt, dass es so viel (GRADUATION: quantification: thing: upscale) Unsicherheit in mein Leben gibt ‘I told you during the summer that there is so much uncertainty in my life’

Level II writers only rarely use GRADUATION in order to up- or downscale nominalized

ATITTUDE tokens, while most Level III texts include at least one such instance. Among Level

II writers, quantification is used in only two instances to modify a nominalization involving either viel ‘a lot of’ or weniger ‘less’ Angst ‘fear.’ One of these is shown in Example (56).

(56) Nachdem dieses SMS Diskutieren, hatte ich weniger angst. Aber auch angstvoll war die Tanzenkompetion später. ‘After the SMS discussion I had less (GRADUATION: quantification: thing: downscale) fear. But the dance competition later was also fear-inducing.’

By using a nominalization, the narrator distances themself somewhat from the feeling of fear; rather than describing themself as experiencing fear (or describing themself as a ‘fearful person,’ which might imply a negative judgment), they construe the fear as a separate entity.

This opens up the possibility of describing fear as an actor—e.g., my fear spurred me to greater productivity; I defeated my fear—though Level II writers do not take this road.

Research by Ryshina–Pankova (2015a, 2015b) indicates that increased use of nominalizations is characteristic of greater advancedness in foreign language writing; these examples help illustrate why this increased use can offer benefits in expanded meaning-making

225

potential in narratives. Consider Example (57), in which the author construes their negative emotions as a thing:

(57) Aber mein Beusch bei der Apple Store hat meiner Panik wirklich nicht geholfen ‘But my visit to the Apple Store didn’t really help my panic’

By nominalizing the emotion of panic, the narrator conceptualizes it as an external phenomenon that others can affect; this move allows the narrator to implicitly blame other characters (the techs at the Apple Store) for the persistence of this negative emotion.

5.3.2 Increased Use of FOCUS

One of the clearest trends from level to level involves the use of GRADUATION: FOCUS.

Writers use FOCUS to “‘sharpen’ or ‘soften’ the boundaries of categorical meanings of an experiential phenomenon or attitudinal value” (Oteíza, 2017, p. 463), describing something by reference to an archetypal example of that thing.

Level I writers did not use FOCUS at all. There are a few instances in Level II, primarily adverbs that strengthen or hedge the writer’s commitment to an assertion. About half of these instances used wirklich ‘really’ or a synonym, as in Example (58).

(58) Charlie und Jordan sind wirklich fantastische Leute ‘Charlie and Jordan are really fantastic people’

Care must be taken when interpreting some of these tokens of GRADUATION; at times, they slip into the realm of ENGAGEMENT. Consider wirklich ‘really’; as Carretero and Taboada (2014) point out, it can be used either epistemically or evaluatively, depending on whether it reinforces the validity of a proposition or scales up the intensity of an emotion or attribute. Example (59), for instance, is somewhat ambiguous; the writer might be using echt ‘truly’ to refer to the difficulty of the day or to the entire proposition.

226

(59) Heute war echt ein schweres Tag ‘Today was really a hard day’

In most of these instances, tokens of FOCUS served to emphasize whatever attribute the writer described; friends were described, for example, as being ever closer to the model of a true friend as in Example (60). In uncommon cases, Level II writers used FOCUS to introduce a caveat of some kind, as in Example (61).

(60) Suna ist ein Beispiel des gutes Freunds. ‘Suna is an example of a good friend’

(61) Sie waren äußerlich meine Typ ‘She was physically my type’

As in Level II, the majority of FOCUS tokens in Level III were instances in which the author increased the valence of an evaluative token, often by indicating that an emotion was experienced or an attribute embodied to its fullest extent. The token ganz ‘completely’ was used several times, as in Example (62).

(62) Ich fühlte ganz allein ‘I felt completely alone’

Level III writers also used adverbs such as besonders ‘especially,’ eigentlich ‘actually,’ total ‘totally,’ vollig ‘fully,’ and wirklich ‘really’ to achieve a similar effect. Writers at this level also used FOCUS more often in order to soften the effect of a particular evaluation. For instance, they might describe a character in positive ways, but mention that he only sometimes fulfills those attributes, as in Example (63), or as a subtle means of criticizing other characters, as in

Example (64).

(63) er ist sehr nett und normaler Weiße cool ‘he is very nice and usually cool’

(64) Sie waren offentlich freundlich mit mir, aber… ‘They were openly friendly with me, but…’

227

FOCUS also allows authors to mention an emotion while assuring their readers that they do not experience it constantly. In Example (65), the writer might not want their parents to worry that they are in a state of constant depression:

(65) Ich kann oft hier allein fühle ‘I can often feel alone here’

The use of FOCUS at Level III also differed from its use in Level II in terms of lexical realization of elements that GRADUATION tokens modify. At this level, there was an increase in the number of nouns and verbal phrases (in contrast to a reliance on adjectives at Level II) that were emphasized by means of FOCUS; these included Examples (66) and (67).

(66) Dann hatte Bri mich besucht, ein total Überraschung ‘Then Bri visited me, a total surprise (+APPRECIATION: impact)’

(67) Wegen meines Kopfwehs konnte ich gar nicht studieren ‘Because of my headache I could not study (-AFFECT: in/security) at all’

L1 writers used similar tokens to sharpen or soften their evaluations, but modified nouns and verbs even more often, relatively speaking, than learners did; Examples (68)–(70) include some such instances.

(68) Immer, wenn ich an Mathe denke, kriege ich schlechte Laune und möchte am liebsten einfach nur wegrennen ‘Always, wenn I think about math, I get a bad mood and would like most of all to just simply run away’

(69) Ich wollte ihr doch unbedingt helfen ‘I absolutely wanted to help her, though’

(70) Meine Kommilitonen in Mathe reden nur über Mathematik und haben kein Leben außerhalb des Studiums, was überhaupt nicht mein Ding ist ‘My classmates in math only talk about math and have no life outside of their studies, which is not my thing at all’

This trend toward increased use of GRADUATION: FOCUS: soften dovetails somewhat with the trend discussed in the sections on JUDGMENT and invoked evaluation. In those section, we

228

saw that higher-proficiency writers began to use certain characters as archetypes or moral yardsticks, making statements on other characters behavior, or on morality generally, by reference to one character’s example. Increased use of FOCUS allows authors to deploy these archetypes in a more nuanced way, indicating the degree to which the reader might interpret each character as an instantiation of the author’s beliefs about praiseworthy behavior.

5.4 Invoked Evaluation

Results from Study 2 flagged a trend in the ways in which learners used implicit evaluation in their narratives; higher-proficiency writers tended to use implicit evaluation more often, and L1 writers relied particularly heavily on provoked evaluation.

The ability to deploy both explicit and implicit evaluations represents a component of competent writing (Rothery & Stenglin, 2000). Apart from more effectively positioning a reader toward their point of view, implicit forms of evaluation can contribute to the strength of writers’ individual textual voices, since implicit evaluation often relies on the writer’s assumptions of the reader’s prior knowledge and views; “[t]hus, instances of invoked attitude have a key role in construing a textual identity, relying as they do on authorial assumptions about the knowledge, beliefs, values and expectations of the addressee (s) and readers” (Don, 2016, p. 5).

In this section, I discuss the ways in which learner and L1 writers use provoked

ATTITUDE, employing textual signals to convey implicit evaluation of some kind. While L1 writers employed this strategy most often, writers at every level show some level of command over it, suggesting that lower-proficiency writers might benefit from overt instruction and awareness raising in this area.

229

5.4.1 Provoked ATTITUDE in Learner Texts

Provoked ATTITUDE (implicit evaluation relying to some extent on in-text signals rather than background knowledge; see Don [2016]) was the more common form of implicit evaluation across all levels. Occurrences of provoked ATTITUDE in learner texts—such as in Examples (71) and (72)—included numerous instances of a behavior that reinforced a previously mentioned, explicit evaluation, weaving both types of evaluation together as Rothery & Stenglin (2000) described.

(71) Sie war sehr nett und toll. Sie sah, dass ich traurig war, und machte mich besser fühlen. ‘She was very nice and cool (+JUDGMENT: propriety; inscribed). She saw that I was sad and made me feel better (+JUDGMENT: propriety; provoked).’

(72) Meine Freundin Sadie hilfte mich. Sie sind auch die beste, und sagte was ich !sollte trage, wo ich sollte stande, und war sehr nett zu mir. ‘My friend Sadie helped me (+JUDGMENT: propriety: provoked). She is also the best (+JUDGMENT: propriety: inscribed) and said what I should wear, where I should stand and was very nice to me (+JUDGMENT: propriety: provoked).

In these cases, the friend’s behavior clarifies an attribute already applied to the friend, serving as evidence of their positive qualities. Even if it is not explicitly stated, it is clear that the narrator in Example 46 intends to praise their friend’s behavior in seeing dass ich traurig war; in a similar vein, the narrator in Example (72) explicitly identifies their friend Sadie as die best while also describing the specific ways in which Sadie helped them.

Texts from Level II also include some narratives in which the author uses a list of mixed explicit and implicit evaluations in order to evaluate someone. One narrator, reflecting on a woman with whom he was set up on a date, describes her positively in Example (73).

(73) Ich sagte ja, weil sie waren äußerlich meine Typ. Sie ist schlau, konservativ, und sie sieht gut aus. Darfur war ich interessiert. ‘I said yes, because she was outwardly my type (+APPRECIATION: quality: aesthetics; inscribed). She is smart (+JUDGMENT: capacity; inscribed), conservative (+JUDGMENT: propriety; provoked), and good-looking (+APPRECIATION: quality: aesthetics; inscribed). That’s why I was interested (+AFFECT: dis/inclination: desire; inscribed).’

230

In this case, konservativ is clearly meant as a positive attribute when included with such unambiguous positive markers as schlau and sieht gut aus. This is a deft way of asserting values in common with the audience. As O’Donnell (2014) points out, “Attitude resources can be used to affiliate with or distance from other people or organizations: by favouring an entity with positive attitudinal evaluations, we voice an affiliation with them by showing we value them”

(p. 111). At no point does the author explicitly claim that ‘conservative’ is a good thing to be; it is simply assumed because of the context in which the evaluation appears. The author thereby not only praises the woman in question, but also those who share her traits.

5.4.2 Provoked ATTITUDE in L1 Texts

L1 writers used provoked ATTITUDE in a few additional ways: by reinforcing an expression of emotion with physical manifestations of that emotion; by including ambiguous descriptors within a list of more explicit judgments, and by setting up oppositions between two groups, one of which is evaluated more explicitly.

In Text xn7, for instance, the narrator describes a friend who has suffered a painful loss.

The narrator provides vivid descriptions of the friend’s mental state in Examples (74) and (75).

(74) Sie hat ganz schrecklich geweint und kein Wort richtig aussprechen können. ‘She cried (-AFFECT: un/happiness: misery/cheer; inscribed) terribly and couldn’t get a word out right (-AFFECT: un/happiness: misery/cheer; provoked).’

(75) Ihre gesamte Lebensfreude war von einem Moment auf den anderen wie weggeblasen und sie starrte nur noch vor sich hin. ‘Her entire joy in life was blown away (-AFFECT: un/happiness: misery/cheer; inscribed) from one moment to the next, and she just stared straight ahead (-AFFECT: un/happiness: misery/cheer; provoked).’

Some of these, such as Sie hat ganz schrecklich geweint, clearly indicate a particular emotion. Others, such as kein Wort richtig aussprechen können and starrte nur noch vor sich hin, could be interpreted to express a variety of feelings; one might be speechless because of shock,

231

deep sadness, joy, or confusion. In these cases, the conjunction und indicates that both tokens are meant to express similar evaluations.

In a similar vein, the narrator in Text xn4 uses the conjunction aber in Example (76) to indicate that her physical ailments express a sentiment similar to the one she hopes her parents do not experience:

(76) Ich will auch nicht, dass ihr euch Sorgen um mich machen müsst. Ich habe aber starke Kopfschmerzen, wenn ich an nächstes Jahr denke, weil ich nicht weiß, was ich studieren soll.

‘I don’t want (-AFFECT: dis/inclination: non-desitr; inscribed) you to have to be worried about me (-AFFECT: in/security: dis/quiet: inscribed). However, I have severe headaches (- AFFECT: in/security: dis/quiet: provoked) when I think about next year, because I don’t know (-AFFECT: in/security: dis/quiet: inscribed) what I should study.’

A more detailed look at this text reveals a number of other instances of provoked

ATTITUDE in which the author makes her feelings toward her friends and classmates clear despite using some tokens that might be ambiguous in isolation, both by combining explicit and implicit tokens in lists, and juxtaposing two groups to make clear their overall evaluation of each.

Learner texts contained a few examples of this move, but it was more frequently evident in L1 texts, such as this longer excerpt, again from Text xn4.

Die Fächer in Mathematik sind viel zu schwer für mich und ich mag die Kommilitonen in Mathe auch nicht, weil sie ganz anders sind, als meine Freunde zuhause. Meine Freunde zuhause sind offen, lustig, lieb, unterstützend und entspannt. Meine Kommilitonen in Mathe reden nur über Mathematik und haben kein Leben außerhalb des Studiums, was überhaupt nicht mein Ding ist. Sie gehen nicht raus, sie machen nichts mit Freunden, sie sind sehr ehrgeizig und denken nur an sich selbst. Wir müssen jede Woche einen Übungszettel in kleinen Gruppen machen, aber ich verstehe nie, was ich überhaupt machen muss. Die anderen Mitglieder in meiner Gruppe sind sehr ehrgeizig und helfen mir nicht, wenn ich eine Frage habe. (. . .) Mein zweites Fach Englisch macht mir sehr viel Spaß und meine Kommilitonen sind sehr nett, lustig und offen. Allerdings habe ich fast gar keine Zeit, was mit meinen Freunden aus Englisch zu machen, weil ich immer für Mathe lernen muss. Das macht mich sehr traurig, weil ich mich dann schnell ausgeschlossen fühle.

The topics in math are much too difficult (-APPRECIATION: quality: manageability; inscribed) for me, and I also don’t like (-JUDGMENT: propriety; inscribed) my classmates in math, because they are totally different (-JUDGMENT: normality; provoked) from my friends at

232

home. My friends at home are open (+JUDGMENT: propriety; provoked), funny, kind, supportive (+JUDGMENT: propriety; inscribed), and relaxed (+JUDGMENT: propriety; provoked). My classmates in math only talk about math and have no life outside of their studies (-JUDGMENT: propriety; provoked), which is not my thing at all (-APPRECIATION: social valuation; inscribed). They never go out, they do nothing with friends, they are very ambitious (-JUDGMENT: propriety; provoked) and don’t help me (-JUDGMENT: propriety; inscribed) when I have questions. My second major in English is a lot of fun for me (+COVERT AFFECT; inscribed), and my classmates are very nice, fun (+JUDGMENT: propriety; inscribed) and open (+JUDGMENT: propriety; provoked). However, I have almost no time to do things with my friends from English, because I always have to study for math. This makes me very sad (-COVERT AFFECT; inscribed), because I then quickly feel shut out (-AFFECT: in/security: dis/quiet; inscribed).

As an example of the aspects of her chosen major that she finds distressing, the author sets up an opposition between her friends at home and her classmates in her math class. She uses several tokens of positive JUDGMENT that quite explicitly demonstrate her positive impressions of her friends; e.g., lustig ‘fun,’ lieb ‘kind,’ and unterstützend ‘supportive.’ Some of the other adjectives she uses to describe her friends might be ambiguous in isolation. Offen ‘open’ is often a positive attribute, but a blabbermouth can be negatively evaluated as ‘too open’ with their feelings; similarly, if a stern fiscal conservative believes a twentysomething living with their parents should be showing more industry, describing that person as entspannt ‘relaxed’ would not be a compliment. In the context of a list of other positive attributes, however, it is clear that all these adjectives are meant to positively evaluate the author’s friends.

In a similar vein, the author conveys her impression of her classmates not only in the tokens she uses to describe their behavior, but also by juxtaposing this section with her description of her friends. She has reported already that ich mag die Kommilitonen in Mathe auch nicht, that they are anders (…) als meine Freunde zuhause, and that her friends at home are

233

characterized by their sociability. ‘Only talking about math,’ then, is clearly meant to be read in a negative light, even though a single-minded focus can be a positive trait in other contexts.

5.5 Polyphony in Study 2

Results from Study 2 flagged an apparent relationship between curricular level and the particular way authors used polyphony—a means of encoding multiple meanings in a single

APPRAISAL token—in their narratives. As discussed in Chapter 2, polyphony has been broken down by Bednarek (2007, 2009b) into the categories of Fused Appraisals (simultaneous realization of two APPRAISAL types, e.g., ATTITUDE and GRADUATION); Appraisal Blends

(simultaneous expression of meanings from multiple APPRAISAL subtypes, e.g. JUDGMENT: propriety and JUDGMENT: capacity); Border Phemonena (borderline cases in which a token could be coded as either one type or another); and Invoked Appraisal (tokens that have one explicit, inscribed meaning but also invoke a meanings from another APPRAISAL type).16 Border

Phenomena and Invoked Appraisal can be difficult to differentiate from one another. According to Bednarek (2007), Invoked Appraisals involve a primary, inscribed ATTITUDE type and a second, implicit type that is typically prompted by one of three triggers: AFFECT related to a behavior that the reader likely feels strongly about, metonymic implication (e.g., a logical connection between a product being evaluated and its producer), or counter-expectancy. Border

Phenomena, by contrast, could justifiably be coded as, e.g., either AFFECT or APPRECIATION, depending on the analyst’s chosen coding criteria; Bednarek (2007) gives the example of analysts interested only in AFFECT, who might include in their analysis tokens such as exciting

16 Since this analysis discusses GRADUATION in another section, and does not deal with ENGAGEMENT, I will not devote a great deal of space to Fused Appraisals.

234

(based on its implication of an actor who is excited), although in other situations it would be coded as APPRECIATION.

With increasing advancedness, we see an increase in the occurrence of Appraisal Blends and Invoked Appraisals. The latter allow the narrator to evaluate other characters more implicitly, by drawing on implicit prior assumptions—assumptions the reader is expected to share—about what constitutes acceptable behavior.

5.5.1 Polyphony within Level I

As mentioned earlier, Level I texts include very few instances of polyphony, perhaps because novice writers may only have access to a vocabulary or evaluative resources that are less open to interpretation. Example (77) represents one case of AFFECT that is also coded as invoked JUDGMENT: propriety:

(77) Wir haben immer kleinen Kämpf - das Zimmer ist unordentlich, sie mochte nicht meinen Freunden in Zimmer ‘We always have small fights - the room is disorganized, she doesn’t like my friends in the room (-AFFECT: un/happiness: antipathy/affection; invoked -JUDGMENT: propriety)’

This token of AFFECT (the roommate’s antipathy) invites judgment upon her because of the shared understanding that liking one’s roommate’s friends constitutes praiseworthy behavior; here the narrator describes the roommate’s lack of affection to implicitly evaluate their roommate as unfriendly. Level I texts include a few instances of Fused Appraisals as well, such as that in Example (78); these are tokens that evoke two APPRAISAL systems simultaneously; e.g., both ATTITUDE and GRADUATION, or ATTITUDE and ENGAGEMENT.

(78) Sie ist gewohnt mit der Larm und der Durcheinander ‘She is used to the noise and the chaos’

In Level I, these took the form of GRADUATION via Infusion; that is, these are tokens that convey attitudinal meaning in an intensified way.

235

5.5.2 Polyphony within Level II

Level II texts include just three more examples of the sort of Invoked Appraisal first seen in Level I. In each case, a token of explicit AFFECT also invokes JUDGMENT in some way. In

Example (79), the author imputes an emotional state to the professor which serves to evaluate the professor’s moral standing. Since professors are expected to help their students, this is not a favorable judgment. Likewise, in Example (80) the narrator ascribes an emotional state to their friend that implies benevolence on the part of that friend.

(79) Der Professor will auch nicht mir helfen ‘The professor also doesn’t want to help me (-AFFECT: dis/inclination / invoked -JUDGMENT: propriety)’

(80) Aber es freut ihn, dass diese Schnee so wichtig fur mich waere ‘He is happy (+AFFECT: un/happiness; invoked +JUDGMENT: propriety) that this snow was so important for me’

In Example (81) below, the narrator evaluates their friend in a roundabout way. Overtly this is also an instance of AFFECT; the narrator desires that the reader of this letter have a chance to meet their friend. It would follow that their friend, then, is a person worth meeting.

(81) Ich hoffe, dass sie beide können ihnen treffen ‘I hope (+AFFECT: dis/inclination) you both can meet her’

Apart from including numerous Invoked Appraisals, Level II texts also differ from those at Level I in their exploration of Border Phenomena. These APPRAISAL tokens can be read in multiple ways, classified as either AFFECT, JUDGMENT, or APPRECIATION depending on the context of the rest of the passage. One such instance, in which the narrator described their roommate’s habits as dreckig ‘dirty, has received attention already. Dreckig receives a primary coding as a token of negative JUDGMENT: propriety; rather than APPRECIATION that simply describes their appearance, the narrator here negatively assesses their former roommate’s apparent unwillingness to adhere to social expectations by maintaining good hygiene — as is

236

made clear by the following sentences, in which the narrator makes continued references to disorderly behavior that clarify how the reader is meant to interpret dreckig.

As the example of dreckig demonstrates, it is important to take into account the context within which a token is embedded. In Example (82), the narrator’s expression of uncertainty

(hatte ich keine Ahnung) might be considered as AFFECT or JUDGMENT; the narrator might be implicitly criticizing themself for their lack of insight.

(82) Während ich die Pre-registration vorbereiten musste, hatte ich keine Ahnung über welche Kursen zu nehmen. Ich konnte nicht was zu machen, sondern ginge ich zu meinem „Dean", aber sie hat mir überhaupt nicht geholfen. I war ganz überrascht

‘While I had to prepare for pre-registration, I had no idea (-AFFECT: in/security / - JUDGMENT: capacity) which courses to take. I did not know (-AFFECT: in/security) what to do; instead I went to my dean, but she didn’t help me at all. I was very surprised (-AFFECT: surprise).’

In this instance, the ambiguous token occurs within an overall context of AFFECT: insecurity and AFFECT: surprise, where there are no other unambiguous instances of JUDGMENT; it would make more sense to code this instance as AFFECT as well. There are, however, several other instances at this level of what will become a more marked trend: descriptions of uncertain mental states that might imply either AFFECT: insecurity or JUDGMENT: capacity. In Example

(83), descriptions of other characters’ mental processes are clearly meant not to evoke sympathy, but to criticize:

(83) Jeden Tag denke ich ϋber meinen Freund, weil er in Pennsylvania ist und ich in DC bin. Ich hoffe, dass er gute – nicht krank – werden, aber die Ärzten habe keinen Ahnung, was schlecht ist.

‘Every day I think about my friend, because he is in Pennsylvania and I am in DC. I hope he is doing well – not sick – but the doctors have no idea (-JUDGMENT: capacity / -AFFECT: in/security), which is bad.’

In the context of this example, having no idea can indicate an emotional reaction. Here, however, the narrator seems to be using the phrase to indicate that the doctors are unskilled. To

237

hammer this home, the narrator explicitly indicates that he finds the doctors’ lack of insight schlecht ‘bad.’

In/security is not the only emotion that prompts consideration of a coding other than

AFFECT at this level. Another narrator uses emotions in Example (84) as a shorthand in order to label one of the characters.

(84) Obwohl ich noch traurig über den wütend Patron war, machte meine Freunde mich so glücklich. ‘Although I was sad about the angry (-AFFECT: un/happiness: antipathy / -JUDGMENT: propriety) patron, my friends made me so happy.’

In this case, the patron’s anger has caused him to act aggressively toward the narrator, and so this label carries negative JUDGMENT connotations.

5.5.3 Polyphony within Level III

5.5.3.1 Border Phenomena

This level contains several more examples of ATTITUDE that could be read one of two ways. As discussed in Chapter 2, I adopted coding guidelines to help reduce the ambiguity in categorizing border phenomena that have to do with responses of affection to a particular trigger.

Following the guidelines in Chapter 2 (Emotion on the part of the subject; Trigger refers to an aesthetic quality), I coded Example (85) and similar instances as AFFECT / APPRECIATION:

(85) Ich genisse die Deutsche kurs sehr ‘I enjoy (+AFFECT: un/happiness /+APPRECIATION: impact) the German course very much’

There are, however, relatively few instances of this type of Border Phenomenon. These Level III texts contain many more instances, such as Example (86), in which the narrator mentions being unable to make a decision, which might reflect an instance of AFFECT: in/security or JUDGMENT: capacity.

238

(86) Meine Freundin war sehr halfbar und ich wusste nicht was wir tuen sollten ‘My friend was very helpful (+JUDGMENT), and I didn’t know what we should do (-AFFECT: in/security or - JUDGMENT: capacity)’

The narrator’s statement that they wusste nicht was wir tuen sollten echoes other statements to the effect that the narrator felt unsure and unsafe after an unsettling encounter. This instance could also be read as one of several contrasts between the friend’s competence and the narrator’s hesitation; the narrator found themself unable (-JUDGMENT: capacity) to deal with a difficult challenge, and their friend demonstrates both their ethics and skill by appearing to help solve it. Depending on the perspective of the reader, the token could justifiably be read in either way.

5.5.3.2 Invoked Appraisals

At higher level, writers also seem much more comfortable using descriptions of characters’ desires (AFFECT) in order to evaluate their integrity (JUDGMENT). Level III texts include instances such as Example (87), in which a character is implicitly praised by referencing their affection toward things one is ‘supposed’ to love:

(87) ich erinnere mich, wie viel du meine Katze liebst ‘I remember how much you love my cat (+AFFECT: un/happiness: affection)’

Here the narrator imputes an emotional state to their letter-reader and also implicitly praises them as someone who, like the narrator, is a cat lover.

These sorts of appraisals use what Don (2016) calls evoked (as opposed to provoked) evaluation, which relies on shared understandings from outside the text in order for its full meaning to be realized. Using these sorts of implicit evaluation allows the writer to build up the sense that they and the reader have a set of shared beliefs in common; in describing their friend in this way, the writer might be read as communicating the sense ‘my friend is a cat lover, which, as we both know, is a good thing to me.’ Through an entire text, “the more a writer relies on

239

evocation, the more they construe a relationship of affiliation and axiological alignment with the intended/putative reader” (p. 23). To illustrate this point, Don described a previous study

(Wigboldus et al., 1999) of posts on an electronic discussion forum written in the 1990s and

2000s. The findings suggested that implicit evaluation has particular relevance for the maintenance of group norms of behavior; this study, which was bolstered by Don’s own work, found that implicit ATTITUDE was “strongly associated with the positive assessment of people outside the group and with the negative assessment of people in the group” (p. 6). In the example above, the shared understanding might simply be that cats are lovable pets, and ‘cat people’ are admirable for their recognition of this fact — a fairly low-stakes moral assertion, though militant dog people might disagree.

Texts at Level III contain nine such instances, in which a description of a character’s feelings or behavior can also be read as an evaluation of their moral fiber. This represents a slight shift from Level II, in which characters were implicitly evaluated by reference to their emotions or internal states; here we have a more specific reference to what characters do or do not desire or endorse.

5.5.4 Polyphony within L1 Texts

In terms of raw frequency counts, L1 texts contain fewer instances of polyphony than texts from Level III. I initially found this result puzzling; one possible explanation is that instances of polyphony within L1 texts were less obvious because their surface meaning matched the overall evaluative prosody of their immediate context. Indeed, on my first attempt at coding these texts I found I had overlooked the tokens discussed below and coded them as single- voiced. Comments from second coders analyzing the texts to gauge interrater reliability prompted me to reassess my codings in this set of texts. L1 writers may also use tokens of

240

APPRAISAL with more precise connotations whose meaning-making potential is easier to interpret unambiguously.

L1 texts included instances of Border Phenomena similar to those seen at previous levels, such as uses of AFFECT: in/security that can also be read as JUDGMENT: capacity, as in Example

(88).

(88) Wir müssen jede Woche einen Übungszettel in kleinen Gruppen machen, aber ich verstehe nie, was ich überhaupt machen muss. ‘We have to write an exercise sheet in small groups every week, but I never understand (- AFFECT: in/security / -JUDGMENT: capacity) just what it is I have to do.’

One new way of realizing these Border Phenomena is by means of modality. In this text

(teased apart at greater length below), the author discusses her attempts to comfort a grief- stricken friend, writing in Example (89):

(89) Was sollte ich denn machen, außer sie in den Arm nehmen und zu sagen, dass alles gut werden würde. ‘What should I do (-AFFECT: in/security / - JUDGMENT: capacity), other than taking her in my arms and telling her, that everything would be well again’

Taken literally, this could be read as an instance of hesitation; the narrator could be voicing her uncertainty about how best to comfort their friend. The question might also be meant rhetorically, with the narrator proactively defending themself against the idea that they did not do enough to help their friend; in this case, the unanswered question could be read as a token of

JUDGMENT: propriety. L1 and Level III writers more frequently use modality to realize these

Border Phenomena.

5.5.4.1 Invoked Appraisals in L1 Texts

One text demonstrates the meaning-making potential of polyphonic appraisal tokens particularly well. In Text xn7, the author describes a traumatic event in which several characters act somewhat unsympathetically. The focal characters include the narrator’s close friend and

241

their family, which might give the author an incentive not to condemn them too explicitly. This may explain why there are very few instances of explicit JUDGMENT in this text. Despite that lack, however, the author is able to convey a clear impression of the narrator, their stricken friend, another helping friend, and an ungenerous family member, primarily through instances of invoked APPRAISAL.

The author begins with an account of a phone call from their friend, whose brother has unexpectedly died. The author uses numerous tokens of AFFECT to describe the friend’s grief, then begins to describe their search for assistance. Just before this excerpt begins, the friend has asked the narrator to drive them home so that the friend can contact other family members.

Obwohl ich ja so ungerne Auto fahre, habe ich es trotzdem gemacht , aber es war eine ziemliche Herausforderung, weil ich immer wieder auf sie geachtet habe. Im Nachhinein hätte ich besser nicht fahren sollen, da es nicht wirklich verantwortungsvoll war.

‘Although I don’t like to drive (-AFFECT: dis/inclination), I did it anyway (+JUDGMENT: tenacity), but it was quite a challenge (-APPRECIATION: impact), because I was always paying attention to her. In hindsight I shouldn’t have driven (-JUDGMENT: propriety), since it was not really responsible (-JUDGMENT: propriety).’

Interestingly, this passage represents the only instances of inscribed JUDGMENT: propriety in the text. Throughout the rest of the narrative, the author uses characters’ thoughts and intentions in order to invoke positive and negative appraisals of their ability and altruism. In the section immediately following, the narrator describes their and their friend’s emotional states. The narrator’s ruminations on their inner doubt could be categorized as purely AFFECT: in/security; however, the elaboration weil ich ihr einfach nicht richtig helfen konnte suggests an assessment of the narrator’s powerlessness to affect the situation; in other words, this can be read as an indictment of the narrator’s capacity.

242

Immer wieder hat sie ganz schrecklich geweint und ich habe mich so hilflos gefühlt, weil ich ihr einfach nicht richtig helfen konnte. Was sollte ich denn tun?

‘She kept on crying (-AFFECT: un/happiness) horribly, and I felt so helpless (-AFFECT: in/security / -JUDGMENT: capacity), because I simply couldn’t help her properly (- JUDGMENT: capacity). What should I do? (-AFFECT: in/security)’

The tag Was sollte ich denn tun? is also somewhat ambiguous. It could be read as further emphasizing the narrator’s uncertainty; it might also be used as an implicit plea for understanding, a justification of their behavior. The narrator continues to tell their story through the lens of their emotional state, introducing a thought with a token of affect and continuing along a cline of more explicit JUDGMENT. In the following section, their emotions can again be interpreted as instances of invoked JUDGMENT.

(…) Ich wollte ihr doch unbedingt helfen und konnte nichts tun außer sie immer wieder in den Arm zu nehmen. (…)

‘But I absolutely wanted to help her (+AFFECT: dis/inclination; Invoked +JUDGMENT: propriety) and couldn’t do anything but take her in my arms again (-JUDGMENT: capacity; +JUDGMENT: propriety)’

Most explicitly, wanting to help someone is an expression of inclination: desire; however, the desire to assist is also generally considered a praiseworthy attribute. This can thus be coded as invoked JUDGMENT: propriety; the narrator is casting themself in a sympathetic role. In a parallel way, the statement that the narrator ‘couldn’t do anything but’ seems to realize negative

JUDGMENT: capacity; it again indicates a lack of ability, reflecting that the narrator is paralyzed or otherwise not up to the challenge of facing this crisis. In another sense, however, the narrator could be describing their deeper character, asserting that they are the sort of person who cannot do other than to assist their friend. This would typically be read as an instance of praise. (If

Martin Luther had actually said “Here I stand; I can do no other” when defying papal authority at the Diet of Worms, this would have been something of a boast.)

243

The narration continues with a further description of the friend’s distraught emotional state. Here, the narrative narrows its focus to AFFECT, with fewer instances of Invoked Appraisal or Border Phenomena. (It could be argued that helpless counts as a Border Phenomenon, but the overall thrust of the passage speaks repeatedly to the friend’s mental state; it would require fewer additional assumptions to read the token helpless as contributing to that evaluative prosody.)

(…) Sie ist eine so starke Persönlichkeit und für mich war es sehr ungewohnt, sie so hilflos zu sehen. Sie hatte bereits einen anderen Todesfall erlebt und war gerade wieder richtig gut drauf. Ihre gesamte Lebensfreude war von einem Moment auf den anderen wie weggeblasen und sie starrte nur noch vor sich hin. (…)

‘She is such a strong personality (+JUDGMENT: capacity), and it was very unusual for me (COVERT AFFECT) to see her so helpless (-AFFECT: in/security). She had just lived through another death and had just gotten well again (+AFFECT: un/happiness). It was as if her entire joy in life was blown away (-AFFECT: un/happiness) from one moment to the next, and she just stared straight ahead (-AFFECT: un/happiness).’

The narrator has so far used tokens of AFFECT that also serve as invoked JUDGMENT in order to position themself as a figure acting in their friend’s best interest. They now describe the reactions of other characters, some of whom act in less praiseworthy ways.

Richtig schlimm fand ich auch, dass ihre Tante nicht die Mutter ans Telefon holen wollte, weil sie dachte, dass das zu viel sei. Das finde ich immer noch ziemlich suspekt.

‘I also found it really awful (-APPRECIATION: valuation) that her aunt did not want to put her mother on the phone (-AFFECT: dis/inclination), because she thought that would be [subjunctive] too much (-APPRECIATION: impact). I still find that fairly questionable (- APPRECIATION: valuation).

Here, the narrator sidesteps making an explicit JUDGMENT, yet through use of AFFECT leaves the reader in no doubt of the aunt’s qualities. Schlimm ‘bad’ and suspekt ‘questionable’ might be considered instances of JUDGMENT, but in each case they apply to a description of behavior, rather to a person directly. In the case of schlimm, the use of of a dass-clause distances the appraisal token from the human agent; likewise, in the case of suspekt, the narrator is commenting on the entire previous sentence — evaluating not the aunt personally, but how the

244

aunt behaved. The aunt’s mental state (ihre Tante [wollte] (-AFFECT: dis/inclination) nicht die

Mutter ans Telefon holen) damns her in the reader’s eyes. Every other character in the story expresses willingness to help the narrator’s friend; the aunt’s lack of desire to do so indicates that she is not meant to be the hero of this story. The narrator reinforces the impression that the aunt’s perspective is not to be endorsed by using the subjunctive form sei, which provides the narrator epistemic distance.

In the succeeding passage, the narrator introduces a third character, the friend’s roommate, who provides an immediate contrast. The roommate immediately expresses their inclination to bring the friend to her parents.

Als ihr Mitbewohner dann endlich wieder kam, war der sofort damit einverstanden, dass wir sie nach Hause zu ihren Eltern bringen. Mir war es egal, die ganz Nacht im Auto sitzen zu müssen um sie sicher nach Hause zu bringen. Hauptsache jemand war bei ihr.

‘When her roommate finally came back, he immediately agreed that we should (+AFFECT: dis/inclination / invoked + JUDGMENT: propriety) bring her home to her parents. It didn’t matter to me (-AFFECT: dis/inclination) if I had to sit in a car all night in order to bring her safe home. The most important thing (+APPRECIATION: valuation) was for someone to be with her.’

Once again, the friend’s desire to help marks him as a praiseworthy individual. The narrator reinforces this impression with a token of APPRECIATION. Characters are evaluated by reference to an implicit yardstick of friend-worthy behavior.

One way in which Text xn4 is effective, though it does not employ the full range of evaluative resources, is that it conveys a clear authorial voice through the sustained use of a particular evaluative palette throughout the text. As the following section illustrates, this use of prosody can lend a text cohesion and situate it as an instance of its genre.

245

5.6 Evaluative Prosody

In the examination of higher-level texts, we have seen various authors draw on expanded meaning-making potential in tokens of ATTITUDE. In the case of JUDGMENT, for instance, higher- proficiency writers use individual tokens not only to grade the behavior of one character, but to implicitly evaluate other characters by contrast, and to make overarching statements about the behavior itself. More advanced writers can exploit the meaning-making potential of evaluation to a greater degree by contributing to an evaluative prosody that infuses an entire text.

The examination of prosody “explores overlaps in choices, possibilities of multiple coding, implicit (evoked) and explicit (inscribed) patterns of APPRAISAL, gradient features such as amplitude (FORCE and FOCUS within GRADUATION) (Martin 2000; White 2002) and

ENGAGEMENT” (Macken–Horarik, 2003b, pp. 313–314) that combine to yield a cumulative impression of the lens through which the reader is meant to interpret the events. Through skillful use of prosody, advanced language users can “distribute both explicit and implicit interpersonal meanings through the whole text so that the audience is ‘positioned’ toward a particular point of view or interpretation of the content” (Rothery & Stenglin, 2000, p. 240). The effect may be to strengthen the author’s individual voice or style, and/or to convey an overall mood that goes beyond individual, piecemeal evaluations, lending greater coherence to the text as a genre- representative whole.

Prosody is related to the concept of ‘evaluative key,’ described by Martin and White

(2005) as “highly regular patterns involving the presence/absence of a sub-set of evaluative meanings” (p. 166) that recur over a large set of public-facing texts that share other generic similarities. Heffer (2007) defines key more generally as “a set of normative constraints on the appraisal options available to the speaker” (p. 172). An author may adhere to the conventions of

246

their perceived social role, as molded by the constraints of their chosen genre, by building up chains of meaning with reliance on particular ATTITUDE types. This is particularly visible in the field of journalism, where Martin and White (2005) and Coffin (2002) have identified keys in the discourses of news reporting and history that are characterized by the use or avoidance of particular appraisal resources. Key in this sense is also referred to by Coffin as ‘voice’ (though it is related only indirectly to the many other constructs labeled as ‘voice’ described in Chapter 1).

Reporter voice, for instance, avoids attributing evaluations to itself and uses JUDGMENT: propriety or veracity sparingly if at all (Swain, 2012). In a study of reporting in Italy, where individual newspapers often advance a particular party line, Swain (2013) describes how journalists use “a typical configuration of linguistically disparate appraisal resources realizing attitude, intensity and focus (. . .) [to position] authors and readers intertextually and dialogistically with regard to thematic content and the views of other authors and readers” (p.

60).

Key plays out at the level of large groups of texts; Martin and White (2005) situate it at an analogous level of delicacy to the concept of register. A given author’s use of evaluative prosody in their writing may reflect their intent to adopt the attributes of a particular key — to write their text as an archetypal reporter, commentator, or historian ‘should.’ At the level of individual texts, use of consistent prosody may also reflect an author’s claim of belonging to an identity group, as well as “personal aspects (. . .) conditioned by our individually held social values (what social values do we respect, and which do we disvalue), and how do we feel about the individuals and events that we are talking about in our text” (O’Donnell, 2014, p. 99).

O’Donnell (2014) uses a speech by then-president Barack Obama to demonstrate one means of concretizing what O’Donnell calls an author’s evaluative style. One can explore the

247

characteristics the author values and disvalues; e.g., does the author consistently praise particular targets for holding, or lacking, certain characteristics? (In this speech, Obama consistently valued propriety, tenacity, and capacity.) The analyst can also track an author’s affiliations by

“deriv[ing] a list of the targets that each voice favours with positive evaluations, and those that they disfavour with negative evaluations” [p. 111]. In this case, Obama somewhat unsurprisingly singled out terrorist groups for condemnation, while evaluating the greater pro-democracy community in Egypt positively, signaling his desire to affiliate with them.

Incidentally, O’Donnell’s approach provides one way to address otherwise ambiguous evaluations. He includes an additional level of delicacy in his codings of JUDGMENT, indicating whether a target holds or lacks a characteristic and whether they are positively or negatively evaluated for it. This makes explicit a step in the coding process that we take when referring to context to assess whether a JUDGMENT token should be read with positive or negative polarity.

Example (90), taken from an excerpt in section 4.2.2, provides one such instance:

(90) Sie gehen nicht raus, sie machen nichts mit Freunden, sie sind sehr ehrgeizig und denken nur an sich selbst. ‘They never go out, they do nothing with friends, they are very ambitious (-JUDGMENT: propriety; provoked) and don’t help me (-JUDGMENT: propriety; inscribed).’

As discussed in that section, the author clearly does not hold these classmates in high regard; throughout their texts they consistently praise some classmates for their friendliness and what we might call ‘work–life balance,’ while criticizing others for displaying exclusive interest in their studies. O’Donnell would describe this as a pattern of evaluation in which JUDGMENT: tenacity is less valued by the author, while propriety is more greatly valued.

Prosody is intertwined with invoked APPRAISAL, which is “typically realized prosodically, often through wordings that imply attitude and generate connotations through their

248

association with other choices in a stream of other words” (Macken–Horarik et al., 2018, p. 45).

As implicit evaluations proliferate throughout a text, each gains an evaluative weight that it would not confer alone. Macken–Horarik et al. (2018) give the example of the loathsome chicken farmers in Fantastic Mr. Fox, whom the author characterizes in part with repeated reference to the intensity and frequency of their eating. As Macken–Horarik and Isaac (2014) point out, “it is the prosodies of attitude associated with each character that shape our responses to them” (p. 80). As we encounter more and more references to the chicken farmers’ frenzy to eat, we as readers are primed to interpret further descriptions of the characters through the lens of distasteful gluttony.

One of the Level III texts (Text ma4) achieves a similar effect with repeated reference to their friends’ kindness or lack thereof. The author’s evaluation of multiple characters facilitates a prosody of JUDGMENT throughout the text; because the author frames each interaction with the various characters with ethical evaluations of their behavior, the text as a whole can be read as a statement of ethical treatment (rather than as containing instances of JUDGMENT here and there).

In addition to clarifying the author’s conformity with the patterns of the evaluative key particular to a profession or discipline or strengthening the author’s affiliation with a group defined by shared values, prosody can lend a text overall cohesion, making it easier to read as an intentional whole. Evaluative prosody, argues Partington (2017), “is the ‘invisible’ non-obvious cohesive glue by which texts cohere evaluatively (and not just propositionally). It creates evaluative harmony, to avoid mixed evaluative messaging” (p. 202). Some learner texts in this project show that higher-level writers can use evaluation not just in isolated instances, but to lend cohesion to the entire text. Text ma1 (see also Example 62) is saturated with AFFECT (primarily

249

un/happiness and in/security). In the Orientation, the narrator sets the stage with Example (91), which includes an expression of feeling isolated and overwhelmed.

(91) Ich hatte viele Exams, Aufgaben und dingen das ich machen musste. Ich hatte viele Heimweh, weil sie auf Ferien waren und ich nicht mitkommen konnte. ‘I had many exams, assignments and things I had to do (-AFFECT: in/security). I was very homesick (-AFFECT: in/security), because they were on vacation and I could not come with.’

The author goes on to describe the various challenges of their workload and the emotional toll the stress took, and then the unexpected relief they felt when their friend visited with a care package. The story concludes Example (92), a recapitulation of the author’s much- improved mental state.

(92) Ich bin dankbar, dass ich nicht allein in diesem ersten Jahr, weit weck von euch, bin. ‘I’m thankful (+AFFECT: un/happiness) that I’m not alone (+AFFECT: un/happiness) in this first year, far away from you.’

This return to the author’s emotional state signals that the narrative has reached its logical conclusion, and the emotional trajectory, from stressed to alone to thankful, lends the story a sense of completeness.

By contrast, in texts characterized by individual, disconnected instances of evaluation, the effect can be dissonant. In the following excerpt from Text si8, from Level II, the narrator relates an experience that likely felt bittersweet. The narrator attends an elegant party characterized by humorous experiences but finds themselves suddenly struck by sadness.

Ich habe eine lustige Geschichte für sie! Letzte woche hatte Georgetown’s SFS eine schöne Danze als die Library von Kongress hier in D.C. Sie war fantastisch! (. . .) Bevor wir zu die Librarie gehen konnten, haben Charlie und ich unserer Rocken trugen. Aber... meinen Rock war zu klein! Ich hatte auf die Grund liegen musste, während Charlie setze auf mir und macht das Zipper schlug. Wir waren hysterischen lachen und auch weinen, die Erfahrung zu lustig war. Ja, dies Nacht war so spaß.

I have a funny (+APPRECIATION: impact) story for you! Last week Georgetown’s SFS had a nice (+APPRECIATION: quality) dance [at] the Library of Congress here in DC. It was fantastic (+APPRECIATION: impact)! (. . .) Before we could go to the library, Charlie and I wore our dresses. But… my skirt was too small (-APPRECIATION: quality)! I had to lie on the ground while Charlie sat on my and closed the zipper. We were laughing hysterically

250

(+AFFECT: un/happiness) and also crying (+AFFECT: un/happiness), the experience was so funny (+APPRECIATION: impact). Yes, this night was so fun (+APPRECIATION: impact).

Throughout this excerpt, each evaluative token matches an overall feeling of happiness and fun. To the extent that any tokens are ambiguous, it is easy to interpret them in light of this overall prosody; weinen ‘crying’ is clearly meant as a token of AFFECT: happiness given a context that includes hysterischen lachen and lustig. In the following paragraph, however, the tone of the narrative shifts in polarity.

Es war einen guten Zeit als ein insgesamt Zeit, dass der anderes nicht so gut ist. Sie kennen jedoch, dass mein Freund sehr krank ist. Er konnte natürlich nicht zum Ball gehen, oder wirklich aus seiner Hause gehen. Ich habe sehr traurig sein. Jeden Tag denke ich über meinen Freund, weil er in Pennsylvania ist und ich in DC bin. Ich hoffe, dass er gute -- nicht krank -- werden, aber die Ärzten habe keinen Ahnung, was schlecht ist. Seiner Hand ist blau und groß, und es schmerzte zu viel, also er konnte nicht gut schlafen oder seiner Hausaufgaben machen (weil er konnte nicht auch mit seiner Hand schrieben)!

It was a good (+APPRECIATION: impact) time [in] a general time, that otherwise was not so good (-APPRECIATION: impact). You know anyway that my friend is very sick (-AFFECT: in/security). He of course could not go to the ball or really go out of his house. I was very sad (-AFFECT: un/happiness). Every day I think about my friend (-AFFECT: un/happiness), because he is in Pennsylvania and I am in DC. I hope (+AFFECT: dis/inclination) that he will be good (+AFFECT: in/security) – not sick (-AFFECT: in/security) – but the doctors have no idea (-JUDGMENT: capacity), which is bad (-APPRECIATION: quality). His hand is blue (- APPRECIATION: quality) and big (-APPRECIATION: quality), and it hurts too much (- APPRECIATION: impact), so he cannot sleep well (-AFFECT: in/security) or do his homework (because he also can’t write with his hand)!

In this excerpt (which immediately follows the previous one), every evaluative token contributes to an overall sense of sadness and worry about the sick friend. One of the few ambiguous tokens here, Jeden Tag denke ich über meinen Freund ‘I think about my friend every day,’ also invokes sadness given that the following sentences reinforce the friend’s dire situation.

The following section reverses the polarity again:

Die Nacht des Ball war für mich sehr wichtig und gut. Es hilft mir die Traurigkeit vergessen. Charlie und Jordan sind wirklich fantastische Leute.

The night of the ball was very important (+COVERT AFFECT) for me and good (+COVERT AFFECT). It helps me to forget (+APPRECIATION: impact) the sadness

251

(-AFFECT: un/happiness). Charlie and Jordan are really fantastic (+JUDGMENT: propriety) people.

Evaluative prosody is clearly present in individual sections of this text; the author uses repeated references to a particular emotion to evoke affective surges that color any more ambiguous emotions they describe. However, different sections of the anecdote convey opposite prosodic impressions, flipping from delighted to worried without any transition. The author’s intent in writing this text seems clear: to describe a situation in which close friends helped them alleviate their feelings of worry and sadness about their ill friend. However, their choice to frame the beginning and end of the story with positive evaluations (Ich habe eine lustige Geschichte für sie… Die Nacht des Ball war für mich sehr wichtig und gut) works against this goal. This lower- level text represents an instance in which inconsistent prosody detracts from the cohesion of the story in general.

5.7 Summary

This analysis described several trends in the use of ATTITUDE, GRADUATION, and implicature across curricular levels.

AFFECT proved an effective tool for generating sympathy for an unlikeable character.

Authors in Study 1 employed emotional evaluation in a variety of ways: to bookend or explain away negative JUDGMENTs of themselves; to offer what we might call ‘character witnesses’ to the character’s worth; or to implicitly criticize other characters for holding antisocial attitudes.

Higher-proficiency writers tended to include instances of AFFECT from a greater variety of characters; this opens up a greater range of meaning-making potential, as a character’s expression of emotion can also serve as an implicit evaluation of that character. Use of

APPRECIATION at higher levels also involved a broader perspective more often, with more

252

experienced writers employing APPRECIATION to describe effects of a situation on characters beyond the narrator. JUDGMENT provided a vehicle for implicit evaluation, especially in higher- proficiency texts. Writers at higher curricular levels used JUDGMENT more often to evaluate multiple characters using one character’s example as a basis for comparison, and to make an implicit statement about what kind of behavior counts as worthy.

In general, these trends suggest a growing awareness on the part of more advanced writers that evaluative tokens do not exist in a vacuum. The interpretation of an instance of

AFFECT may depend on which character expresses that emotion; a JUDGMENT of one character may make a broader statement about ethics, depending on what we already know about that character. Writers’ use of context-dependent evaluations is also reflected in more apparent evaluative prosody in the texts of more proficient writers.

That said, it is too simplistic to argue that the texts in this sample represent a smooth cline of progress, with each level delineated nicely from the previous one. In general, we are well served to remember Derewianka’s (2007) admonition not to read overarching language development trends into the results from a small number of texts; in this particular project, certain writers demonstrated nuanced command of some functions that was uncommon at their level. The breadth of examples of invoked APPRAISAL described in this chapter, for instance, suggests that choices that allow for nuanced evaluation are available to even less experienced L2 writers.

In the following section, I propose ways in which these findings could contribute to pedagogical practices, and I reflect on the process of designing and conducting this study in order to suggest methodological improvements for future studies and potential adjustments to the

APPRAISAL scheme.

253

CHAPTER 6. DISCUSSION AND CONCLUSION

In this concluding section, I summarize the results of the two studies and discuss the limitations inherent in this project, as well as implications for instruction and future research yielded by the results of these studies.

6.1 Summary of Research

This dissertation aimed to better understand some of the tools that language learners can use in order to achieve interpersonal goals in the context of a particular genre, and to explore the value of APPRAISAL as an objective tool of analysis of evaluative language in German. In the previous chapter, I outlined potential instructional activities that could draw on this analysis, and

I highlighted aspects of the APPRAISAL framework that could be refined in future analyses. I close with a reflection on the importance to language learners of gaining fluency in the narrative genre and master of evaluative language in general, and I offer suggestions for further research.

This project involved an APPRAISAL analysis of two sets of student texts. Fairy tale texts from Study 1 were analyzed in order to identify potential cross-level trends, these trends were further explored and tested for statistical significance in the larger text sample collected for

Study 2. The APPRAISAL analysis revealed differences between curricular levels in terms of

APPRAISAL type, incidence of implicit evaluation, and the scope over which evaluation tended to apply. In-depth exploration of the results, particularly in the areas of AFFECT and JUDGMENT, revealed several ways in which more proficient L2 writers (and L1 writers) used those resources to engage in implicit evaluation. Lower-proficiency writers made similar moves (albeit less frequently) suggesting that students might benefit from additional instructional time in this area.

254

Several trends concerning APPRAISAL use across levels involved a broadening of the scope of evaluation. In the case of APPRECIATION, for instance, Level II writers in Study 2 relied more heavily on APPRECIATION: impact to describe the effects of a friend’s help on the narrator; more proficient writers typically used APPRECIATION: social valuation to describe the friend’s behavior in more general terms, evaluating the actions not in terms of their individual effects, but in terms of their value to the community.

Further examples of scope-broadening involved JUDGMENT; higher-proficiency writers more often used JUDGMENT to identify particular traits that constitute moral behavior, while lower-level writers focused on the moral virtue of the helping friend specifically. JUDGMENT was also more often implied in higher-level texts, sometimes in provoked evaluations involving

AFFECT that required shared moral understandings (e.g., this character is the sort of person who likes x, and such people are not to be trusted). These were particularly common in L1 texts, where the authors often used characters’ stated thoughts and intentions in order to frame them as more or less generous or reliable.

The process of APPRAISAL coding also highlighted choice points in the APPRAISAL scheme (particularly at the border between AFFECT and APPRECIATION) that would benefit from increased conceptual clarity. Choosing between categories often required subjectivity on the part of the coder; edits to the scheme for Study 2 provided additional guidance, though a test of interrater reliability indicates that even coders familiar with APPRAISAL often disagree in individual cases. Before moving on to general limitations of the dissertation, I explore ways in which this subjectivity may have affected the project and how it could be mitigated in future research.

255

6.2 Potential Bias of APPRAISAL Categories

As discussed at length in Chapter 2, the boundaries between APPRAISAL categories are porous in many cases (Thompson & Alba–Juez, 2014). I discuss the three areas—differentiating between APPRECIATION: impact and APPRECIATION: quality, distinguishing JUDGMENT: capacity from JUDGMENT: tenacity, and coding for Appraisal Blends in a principled way—in which I updated the APPRAISAL scheme for Study 2 in the interest of following coding guidelines that others could follow and obtain similar results. Despite these updates, discrepancies between coders suggest that the APPRAISAL scheme as employed for German texts retains a degree of subjectivity that makes unbiased analysis more difficult.

To assess the objectivity of the APPRAISAL scheme in general, 20% of the ATTITUDE tokens in Study 2 were coded by an additional coder. The additional coders were both members of the German department, though none of the L1 German speakers who responded to the writing prompts also acted as additional coders.

The second coder and I agreed on a given token’s ATTITUDE type in 75% of cases (this includes cases in which coded a token as expressing JUDGMENT while disagreeing on whether it reflected capacity or tenacity). This yielded a cohen’s kappa of .66. We agreed fully on a given token’s ATTITUDE subtypes in 57% of cases, yielding a cohen’s kappa of .54. The third coder and

I agreed on the ATTITUDE types in 77% of cases, yielding a cohen’s kappa of .69, and agreed on subtypes in 62% of cases, yielding a cohen’s kappa of .60. These results are considered acceptable in many research contexts (Multon, 2010), though it is striking that agreement on

ATTITUDE subtypes in both cases was not much better than 50–50. Instances of non-agreement were not resolved, though this could be a fruitful avenue for future work aiming to further refine

ATTITUDE categories.

256

In future research on narratives, it would also be advisable to enlist the assistance of an additional coder to judge the boundaries of narrative stages as well. While there are ample descriptions of the characteristics of these stages in the literature on narrative analysis, the precise boundaries between stages might be open to debate.

6.2.1 Ambiguities in Coding APPRECIATION

In terms of APPRECIATION, I adopted the distinction between impact and quality proposed by Ngo and Unsworth (2015). These authors suggest using two probe questions to differentiate these subtypes from each other and from APPRECIATION: valuation. Attributes that (could) trigger an emotional response in the character are coded as APPRECIATION: impact; attributes that can be graded on a scale of some kind comprise tokens of APPRECIATION: quality. (Tokens of

APPRECIATION: impact have the greatest in common with AFFECT; tokens of APPRECIATION: valuation, with JUDGMENT.) While this seemed, subjectively, easier to use than the previous distinction between reaction and quality, an interrater reliability test revealed several discrepancies. Six of 54 disagreements (11.11%) between the first two coders involved

APPRECIATION subtypes — though only between impact and valuation, not quality. There were several more discrepancies when comparing results from the first and third coders; ten out of 70

(14.29%) involved APPRECIATION subtypes. If this distinction is to be retained in the APPRAISAL scheme, future analysts might hold a more detailed pre-coding session in which they highlighted these probing questions.

6.2.2 Ambiguities in Coding JUDGMENT

More work is also required to yield a principled distinction between the JUDGMENT categories of tenacity and capacity. Disagreements over tenacity vs. capacity accounted for 50%

257

of the interrater coding discrepancies between the first and third coders involving JUDGMENT types. In Example (1), it is difficult to tell which coding is most appropriate:

(1) im Gegensatz zu mir war anscheinend keiner in dieser Gruppe in der Lage, etwas zum Projekt beizutragen ‘In contrast to me, apparently no one in this group was in the position to contribute something to the project’

[Nicht] in der Lage ‘[not] in the position’ is somewhat vague here; the author might be expressing frustration at their classmates’ lack of knowledge, preparedness, or energy at that moment. That this is an instance of JUDGMENT seems clear; as the following sentences explain, the classmates have provoked the author’s negative emotions with their inaction, and it is easy to hear a note of scorn in the word anscheinend. However, it is unclear how one would code this as either tenacity or capacity in a principled way.

JUDGMENT: capacity can also be conflated with AFFECT: in/security; indeed, this combination reflected the most common interrater discrepancy, between all three coders, among

AFFECT / JUDGMENT disagreements. The case in Example (2) is somewhat ambiguous:

(2) Ich dachte nicht, dass ich mein Wohnheim finden könnte ‘I didn’t think that I could find my residence’

This could represent an indictment of the author’s competence (especially if they are in a familiar place where they should be able to find their way home), or it could reflect the author’s feelings of uncertainty and fear in an unfamiliar location. It also remained difficult to distinguish in an objective way between COVERT AFFECT, AFFECT, and APPRECIATION: impact. Among the

54 coding disagreements between the first two coders, eight (or 14.81%) involved COVERT

AFFECT; between the first and third coders, the proportion was 8/70 or 11.43%.

258

6.2.3 Ambiguities in Coding Appraisal Blends

Appraisal Blends proved difficult to identify and code in an objective way. This difficulty was likely compounded by a degree of overlap between AFFECT: un/happiness and dis/satisfaction. Martin and White (2005) distinguish AFFECT: dis/satisfaction from un/happiness by asserting that dis/satisfaction “covers emotions concerned with telos (the pursuit of goals) — ennui, displeasure, curiosity, respect” that “are sensitive to how active a role we are playing in the activity we’re reacting to” (pp. 49, 51). These emotions would seem to involve a stimulus of some kind, a trigger that prompts the goal-seeker’s pleasure, while manifestations of un/happiness sometimes occur without a particular trigger. In practice, however, it proved difficult to disentangle dis/satisfaction and un/happiness. Out of 28 instances in which the third coder and I agreed on the ATTITUDE type but not the subtype (e.g., one of us coded a token as

JUDGMENT: capacity, one as JUDGMENT: tenacity), nine of those instances (or 32%) involved a disagreement as to whether an AFFECT token represented un/happiness or dis/satisfaction. (For comparison, there were no more than three instances of disagreement on any other specific

ATTITUDE subtype pair.)

Appraisal Blends also involved interplay of AFFECT and a trigger of some kind. I adopted a somewhat cumbersome process to code examples of blended AFFECT, APPRECIATION, and

JUDGMENT in the case of expressions of affection and love, based on the triggers and sources of the emotions (see Chapter 2). While this process should theoretically make coding choices less subjective, its multistep nature invites human error. It was also difficult to reflect these distinctions in the UAM Corpus Tool application. Example (3) offers one such instance of blended AFFECT / APPRECIATION, in which the token liebe would receive the following codes:

(3) Ich liebe meine Zeit an der Georgetown Universität I love my time at Georgetown University

259

APPRECIATION Impact Polyphony_Yes Appraisal Blend Blend_Affect

This instance would be counted as one of APPRECIATION but not of AFFECT. (Using this software, it does not seem to be possible to label a token as expressing multiple ATTITUDE types.) This problem could be averted by coding this token twice; e.g., as an instance of AFFECT that is an Appraisal Blend, and as an instance of APPRECIATION that is an Appraisal Blend.

However, this then results in double counting of all Appraisal Blends.

The perceived need for these cascading labels may simply stem from the recognition that

AFFECT is an evaluative tool just as JUDGMENT and APPRECIATION are. The description used for blended AFFECT / JUDGMENTof the emoter, in which the trigger refers to a widely praised/condemned moral behavior, is difficult in the end to distinguish from an Invoked

APPRAISAL, in which emotional expressions generally (beyond affection/antipathy) might prompt the reader to judge that character more or less harshly. This conflation could be addressed by a refinement to AFFECT proposed by Benítez–Castro and Hidalgo–Tenorio (2019). Since the original formulation of APPRAISAL in Martin & White (2005), several linguists have proposed modifications to its various categories (especially AFFECT) to bring them closer into congruence with the distinctive characteristics of human emotions as expressed in real language. Bednarek

(2008), for instance, drew on evidence from corpus linguistics to argue for a redrawing of

AFFECT categories based on observed patterns in lexical realizations of emotions. Benítez–Castro and Hidalgo–Tenorio argue that the blurriness between AFFECT category boundaries results in part from a reliance “on folk concepts and intuitions” (p. 302) and propose a further restructuring of AFFECT types that draws on evidence from three psychological theories of emotion.

260

Based on a view of emotions as “neurophysiological processes emerging from our perception of an event, situation or entity as relevant, beneficial or harmful to our goals, needs or values” (p. 316), Benítez–Castro and Hidalgo–Tenorio (2019) structure their proposed scheme around the axes of goals and valence. They group surprise, (dis)interest, and (dis)inclination as goal-seeking emotions, triggered when “our psychic energy is directed to particular stimuli”

(p. 319) and distinguished from each other based on our degree of active cognitive engagement with, or pursuit of, such stimuli. These are contrasted with goal-achievement emotions such as

(dis)satisfaction (which includes (un)happiness and (in)security), which involve the pleasurable feelings that reflect the degree congruence between our goals and our external reality. Finally, there are goal-relation emotions, attraction and repulsion, which represent “a generalized valenced focus and stance” (p. 323) toward particular people and concepts. Benítez–Castro &

Hidalgo–Tenorio propose a further reconceptualization of ATTITUDE in general, reframing

ATTITUDE as AFFECT and subdividing it into EMOTION (formerly AFFECT) and OPINION

(encapsulating JUDGMENT and APPRECIATION).

As noted above, I found it difficult to distinguish between un/happiness and dis/satisfaction. I used the presence of absence of a trigger to determine the coding decision; as

Benítez–Castro and Hidalgo–Tenorio (2019) point out, however, the presence of a trigger is often implicit, making coding decisions arguable in many cases. They resolve this ambiguity by recogizing the causal relationship that sometimes connects these two emotions: “happiness seems to be most likely an emotion category built upon a more general feeling of satisfaction triggered by our success in attaining relevant goals” (p. 310). Future APPRAISAL studies would benefit from adopting a framing of AFFECT such as that of Benítez–Castro and Hidalgo–Tenorio that clarifies the role of triggers and goals in distinguishing between un/happiness and

261

dis/satisfaction. It would also be beneficial to use coding software that allows coding a token as both AFFECT and APPRECIATION, to better handle cases of Appraisal Blends.

All this being said, linguists striving to improve the objectivity and replicability of

APPRAISAL coding should bear in mind that some level of indeterminacy is likely inherent in its subject matter, and that trying to impose strict categories on blurry-edged phenomena can obscure rather than illuminate the object of study (Dawkins, 2011). One benefit of using the

APPRAISAL framework is that it can suggest patterns across texts and curricular levels that can then be explored in greater detail; data gathered using even the most ‘accurate’ APPRAISAL scheme (whatever that might mean) would require analysis in context to explore, for instance, how certain evaluative tokens play into the prosody of an entire key or text. As Heffer (2007) puts it, “[t]he appraisal framework does provide a rich metalanguage for describing evaluative phenomena at a textual level and seems particularly useful in illuminating macrolinguistic patterns of evaluation across text. The problems come when it ceases to be a useful heuristic and starts to become an uncomfortable straitjacket, when the focus goes on the labelling rather than on the explication” (pp. 176–177). One major reason for selecting a research design that involved both quantitative and qualitative analysis was to include contextual information in the use of evaluative tokens, rather than relying on frequency counts alone.

6.3 Limitations

Several factors caution against interpreting the findings of this dissertation overly broadly. In particular, Study 1 and 2 were hampered by a relatively small sample size, reflected students’ experience with a specific language curriculum, and may have reflected implicit bias in the use of an APPRAISAL scheme with subjectively delineated categories.

262

6.3.1 Small Sample Size

The present dissertation involved a total of 44 texts from learners of German, in addition to five texts written by native German speakers. Four additional texts were collected but not used because the writers did not follow the guidelines given in the writing prompt. While coding this number of texts still proved time-consuming, the number of instances of ATTITUDE (599 in Study

1, 1176 in Study 2) was low enough that any conclusions drawn from these data must be offered tentatively. In particular, it was sometimes impossible to make statistical comparisons across texts, as chi-square tests should not be run when there are fewer than five instances of a given phenomenon (and cannot be used when there are zero). Comparing the use of AFFECT subtypes across levels was inadvisable, since there were fewer than five instances of AFFECT: dis/satisfaction at multiple curricular levels. I was also interested in cross-level differences in the use of JUDGMENT to evaluate family members, but Level I texts in Study 2 contained no instances of JUDGMENT directed toward these characters. (It was particularly challenging to compare Level I texts with texts at other levels; Level I texts were relatively short, containing only 64 total instances of ATTITUDE in Study 2.)

Drawing on a larger sample size, of course, requires additional time spent on coding.

While the process of coding each of these texts proved time-consuming, increased familiarity and experience with the APPRAISAL scheme should make the process more rapid, as would implementation of coding guidelines—such as those proposed by Benítez–Castro and Hidalgo–

Tenorio (2019)—that further demystify the distinction between ATTITUDE subtypes. It is also possible that deep learning models could be trained to code for APPRAISAL, given a large enough sample size from which to draw. Human raters could then review the codings and adjust as necessary, likely saving time overall. Studies in the field of psychology have already begun to

263

rely on machine learning tools to code large sets of data based on typical stages of talk therapy

(e.g., risk check, set goals, future planning) in order to assess the effectiveness of clinical therapeutic practices (Ewbank et al., 2020); training a machine learning system to code for

APPRAISAL, based on initial input from human codes, would seem to have had a similar level of difficulty.17

6.3.2 Unique Curricular Context

With two exceptions, the texts analyzed in this project were written by undergraduate or graduate students at Georgetown University. (Two texts were written by students in summer courses.) About half of the participants had taken at least two previous semesters of German at

Georgetown. Most of the participants, then, had had sustained experience with the articulated sequence of genres that characterizes the GUGD undergraduate curriculum. Students in GUGD courses progress through a sequenced examination of genres, jointly analyzing model texts in order to understand the meaning-making affordances of each genre. Student work is assessed in part based on students’ use of discourse markers appropriate to the given genre (Georgetown

University, 2020). One outcome of this approach is that students gain experience outlining and composing entire texts. To the extent that students at more advanced levels were able to compose texts with a consistent evaluative prosody that imparted a high degree of cohesion, this may reflect their familiarity with similar assignments at Georgetown; other students who have studied

German for similar lengths of time may not have as much experience in this area. In some ways, this is also a virtue; in selecting participants with this shared curricular experience, I was able to control for knowledge of genre, sidelining absence of genre knowledge as a potential

17 I am grateful to David Jangraw for this suggestion.

264

confounding variable in the results. That said, collecting data from writers with less instruction in or familiarity with genres might have revealed a wider range of ATTITUDE resources that appeal to developing writers.

The Georgetown community also includes a large proportion of multilingual students.

Ten of the German language learners across the two studies claimed at least advanced proficiency in at least one other language (other than their L1). If knowledge of one L2 fosters speedier acquisition of a second L2, we might expect this population of students to progress toward advancedness relatively more quickly than a comparable group of students who had no additional language learning background. In short, the ability of students at Georgetown to deploy effective evaluation in narratives should not be used thoughtlessly as a yardstick to measure the performance of students in other instructional contexts. (On the other hand, it would hardly be accurate to argue that the archetypal L2 German learner is one who speaks no other languages besides English. Given the landscape of commonly offered languages in the United

States, L2 German students may actually be more likely than not to have experience with an additional second language, such as Spanish or French. It may be that working with L2 German writers offers researchers more opportunities to study multilingual subjects.

For language instructors interested in guiding their students toward greater mastery of narrative writing, these considerations caution against adopting recommendations from this project without first teaching the fundamentals of this genre. (Further reading on the development of the GUGD curriculum can be found at https://german.georgetown.edu/multipleliteracies030217; additional readings pertinent to genre- based instruction include, e.g., Christie, 1997, 1999; Martin, 1997; Martin & Rose, 2003;

265

Rothery & Stenglin, 1997; Ryshina–Pankova, 2016; Swaffar, Arens, & Byrnes, 1991; Warren &

Winkler, 2016).

Limitations inherent in these studies included their small sample sizes, a context specific to Georgetown University, and a coding scheme that could be made even more objective.

Despite these limitations, trends emerging from the two studies suggest potential instructional avenues, such as working with beginner writers on the effect of different AFFECT choices, analyzing JUDGMENT patterns to gauge which traits an author particularly values, and incorporating implicit evaluation into humanities learning assessment rubrics.

6.4 Pedagogical Applications

This project was undertaken in part to yield a better understanding of the evaluative meaning-making tools to which L2 language writers at various curricular levels can be expected to have access, the better to design thoughtful pedagogical activities that help students stretch their capabilities just beyond their previous levels of literacy. As this project has hopefully demonstrated, evaluative language enhances the meaning-making potential of student writing in a variety of ways; offering students overt guidance on this front would likely support them in more effectively achieving their communicative goals (e.g., generating sympathy, telling an emotionally affecting story, making a point about the nature of friendship or scholarship) and in writing texts with greater overall cohesion. Devoting additional time to evaluative language may also offer a benefit in terms of student motivation, since “instructional focus on evaluative language allows students to reflect on their personal connection to the topic and enables them to say something personally meaningful about the foreign language culture” (Warren & Winkler,

266

2016, p. 51); at a time of falling enrollments in modern language programs, solidifying students’ perceived connection to the L2 is a meaningful benefit.

Any discussion of pedagogy requires considering the curricular context. As has now been discussed at length, the texts analyzed in this project were collected from students with at least some experience with genre-based instruction. As an instructor trained to teach texts within this curricular framework as an integral part of my graduate education (Byrnes, 2001), I am likely inclined toward teaching recommendations that presume a similar instructional context.

Instruction within a genre-based curriculum lends itself to teaching activities such as the reading for meaning model proposed by Swaffar, Arens, and Byrnes (1991). This approach exploits the staged nature of genres with a five-step sequence that begins with a broad examination of the stages of a model text, then analyzes word- and sentence-level details before widening the scope again. When using this model, teachers first engage in a preview by evoking students’ background knowledge of the genre and their overall impressions of the text based on title, images, captions, et cetera. In the episodic structure step, students and teachers delineate the divisions between stages based on discourse markers that indicate a shift in time or affect.

The subsequent word- and phrase-level reconstruction step involves providing students matrices they can use to identify individual tokens that contribute to key ideational and interpersonal meanings that occur throughout the text. Students then move from the phrase-level to the sentence/paragraph-level stage, engaging in suprasentential/sentential reconstruction of parts of the text using the tokens in their matrices. Finally, students are prompted to offer their opinion about textual information and use instances of language from the text to express their own ideas.

This model underlies many of the textual analysis activities practiced in the GUGD (Ryshina–

Pankova, 2021).

267

Even absent a sequenced curriculum, a baseline level of familiarity with the genres of texts one is instructing is valuable. Instructors can support their students by familiarizing themselves with the typical evaluative moves that recur across texts; “[l]anguage teachers who are familiar with the linguistic resources that typically do (or do not) occur across different genres are in the advantageous position to support their learners in developing advanced literacy abilities through evaluating and providing meaningful feedback on student work” (Crane, 2016, p. 122). A range of pedagogical activities may be appropriate to genre-based instruction; of central importance is that instructors provide scaffolding to help learners practice the lexicogrammatical tools appropriate to their language ability, and that instructors direct students to links between lexicogrammatical form and interersonal meaning, and at a higher level to “the connections between social purposes and the staged structure of genres and varying lexicogrammatical realizations (. . .) that support such generic staging” (Ryshina–Pankova, 2016, p. 66). By reflecting on the characteristic structure and goals of genres, instructors can guide learners to discover specific language resources that are instrumental in achieving those communicative goals; for instance, in the context of academic writing in history, students can learn how particular lexicogrammatical resources can help them “present[] a partial stance as objective and (. . .) lead[] readers toward acceptance of the authorial viewpoint (e.g., reference to the voices of other experts, juxtaposition of opinions through concessive structures, etc.)”

(Ryshina–Pankova, 2016, p. 66).

Many instructional activities involve intentionally drawing students’ attention to evaluative features of texts during class time. Results from an action research project conducted by Warren and Winkler (2016) suggest that this overt instruction yields results (as measured by the frequency of evaluative tokens present in student speaking tasks produced after the

268

instructional intervention) when incorporated into an instructional cycle characterized by review of the subject matter and engagement with model texts. Once example of such an instructional cycle is laid out by Troyan (2016), who engaged in a five-lesson sequence on travel writing in which students move from (1) an examination of the field of castles and tourism, to (2) a focus on the stages of model texts and the verbal processes common to each stage, then (3) an analysis of evaluation and (4) circumstantial adjuncts that describe a castle’s physical situation and accoutrements. In a comparison between a pretest and posttest written by one student, Troyan argues that students were able to produce texts more characteristic (in terms of interpersonal distance from the reader and interconnected stages) of the genre they had studied.

In an exploration of evaluative resources that similarly emphasizes the staged nature of texts, Kuhn (in press) proposes pedagogical activities in which students examine the effects of

AFFECT and JUDGMENT in an autobiographical text by an East German educator. Kuhn proposes using guiding questions to help student recognize the words that signal the beginning of each new stage, and she offers matrices instructors can use to direct students’ attention toward relevant emotions, distinct voices, and implicit judgments at various stages.

Other pedagogical activities described by SFL practitioners have included inviting students to adopt the perspective of both writer and reader(s); prompting students with probe questions; and considering a range of related texts.

Awareness of the presumed audience is crucial to any effective narrative; in letter- writing, the reader’s role is particularly evident. In a study on letter writing in German, Crane

(2016) argues that “having students consider the voices not just of the writer, but those of the reader as well, can help L2 writers to make informed linguistic choices rooted in critical thinking about the subject matter” (p. 133).

269

Student readers can use a variety of probe questions to examine elements of a narrative through the interpersonal lens. As Macken–Horarik et al. (2018) illustrate with a list of probe questions displayed in Table 29, students can be guided to use such questions to flesh out how a narrative text positions its readers. Kuhn (in press) proposes similar questions to help students identify successive narrative stages.

Table 29. Probe Questions to Identify Reader Positioning. Interpersonal Probe Questions Follow-up Questions Aspects Overarching What is the point of Question view from which you are being positioned to experience the text? Narration Who tells the story? — a narrator or character? (. . .) internal or external to events in the story? First/third person? (. . .)

Focalization Who sees? How is this viewpoint mediated verbally? (. . .)

Interpersonal Probe Questions Follow-up Questions Aspects Dialogue or Who speaks? How do characters address one another? (. . .) Does voicing the dialogue reveal character? How does it do this? (. . .)

Evaluation What attitudes are (. . .) AFFECT: How are characters’ feelings of expressed? un/happiness, in/security and dis/satisfaction portrayed? (. . .) JUDGMENT: How are personal traits (capacity, normality, tenacity) of the characters portrayed? How are evaluations of truth or moral and ethical issues conveyed? (. . .) APPRECIATION: How is a particular atmosphere or ambience conveyed?

Intensity How loud or soft is Consider GRADUATION — use exclamatives, Evaluation? intensifiers or repetition in language (. . .)? (Macken–Horarik et al., 2018, p. 70).

270

Yasuda (2017) also used probe questions as an instructional tool in a study of academic summaries. She presented students with multiple versions of a text that described the same content but with different audiences in mind. Using scaffolding questions, she then encouraged students to explore how each author provides details that support their thesis statement, condenses information, and indicates awareness of the reader. This approach echoes a framing device Eggins (2004) employs in her illustration of the metafunctions within SFL; Eggins uses the image of a spider biting a beetle to demonstrate how the meaning of a sentence can change when one keeps two strands of meaning constant and varies a third. In Examples (4)–(6), the interpersonal function of the clause remains the same (each offers information), as does the ideational function (the scene described is the same), while the textual meaning, or point of departure of the sentence, is different in each.

(4) The redback spider gave the captured beetle a poisonous bite. (5) A poisonous bite was what the redback spider gave the captured beetle. (6) To the captured beetle the redback spider gave a poisonous bite. (p. 138)

Eggins repeats this illustration with examples that vary only in terms of the ideational function, or only in terms of the interpersonal function. This technique of considering how a text’s tone and purpose would change if certain elements were removed or altered, which

Dennett (2014) calls ‘twiddling the knobs’ in the context of thought experiments, can be a powerful instructional tool because it invites students to isolate the function or meaning-making potential of a particular utterance.

Using these examples of instructional activities in the context of SFL-informed instruction in a genre-based curriculum, I offer a few directions instructors might pursue, suggested by results from this project.

271

6.4.1 Using AFFECT to Garner Sympathy

Exploration of AFFECT in Study 1 yielded several instances in which authors were able to evoke pathos in the reader, sometimes by using characters other than the narrator as sources of emotion to lend credibility. Results from this study served as a reminder of the role that AFFECT can play in evaluating a character, even in the absence of explicit JUDGMENT. AFFECT commonly plays a prominent role in more personal genres (e.g., letters to intimates) and in story genres

(especially anecdotes), as well as in political texts aimed to persuade (Westen, 2008). These genres are often explored in introductory language courses, making AFFECT perhaps a natural hook for overt instruction for less experienced L2 writers.

This project indicated that writers at higher levels tended to express emotion through a greater number of characters (particularly in the fairy tale task); less experienced writers took advantage of this meaning-making opportunity less often. Teachers interested in encouraging multivoicedness in their students could draw on Rothery’s (1996) teaching cycle to design an instructional sequence based on AFFECT in narratives. This teaching cycle involves four distinct stages—Negotiating Field, Deconstruction, Joint Reconstruction, and Independent

Construction—and has proven fruitful in genre-based instruction (Warren & Winkler, 2016).

Rothery’s model suggests approaching a new genre by first discussing students’ prior knowledge of the subject area and introducing them to the language essential to that content; this is the stage of Negotiating Field. Instructors might begin a unit on fairy tales, then, with discussion of stories students know from their own upbringings and those stories’ common themes and tropes.

The following stage of Deconstruction involves examination of model texts, with an emphasis on the goal of these texts and the lexicogrammatical features that allow their authors to

272

approach those goals; put another way, if a genre represents a “staged, goal-oriented activity”

(Martin, 1984, p. 25), engaging in Deconstruction means investigating those goals and stages as realized in certain examples of that genre. During this stage, instructors could present students with fairy tale texts that include expressions of emotion on the part of a range of characters.

Again, the goal in Deconstruction is to “engage with the text on a meta-level by trying to discover its communicative goal and the stages and linguistic features that allow the text to fulfill this purpose” (Warren & Winkler, 2016, pp. 37–38); ideally, teachers would not prescriptively instruct students to use certain words in order to invoke set emotions, but would instead invite students to describe the emotional response that the text evokes at various points. They could then use probe questions such as those suggested in Macken–Horarik et al. (2018) to isolate evaluative tokens that contribute to that affective response. Students might consider which characters express emotion most often and most consistently; which characters they find themselves empathizing with; and whether they find that their inclination toward a character changes through the course of the narrative. In the case of the original Frau Holle text that Study

1 participants read before writing their own tales, students engaging with fairy tales might fill out a matrix resembling the one in Figure 20, either explicating examples provided by the instructor or finding their own.

In the stages of Joint Reconstruction and Independent Construction, students and instructors (and then students alone) create new texts using the insights gleaned in the previous two stages. The Joint Reconstruction stage represents an opportunity to examine the effect that different lexicogrammatical choices can have on the impact of the narrative as a whole; this is one stage when the instructor might invite students to ‘twiddle the knobs,’ assessing the impact of different AFFECT choices on otherwise similar settings. (An activity that allows for this sort of

273

reflection—on changes in wording that lead to changes in meaning—integrates well into many pedagogical cycles, as in Ryshina–Pankova’s [2021] description based on Swaffar, Arens, &

Byrnes [1991] of a potential teaching cycle on the anecdote story genre.) As one activity in this stage, groups of students might complete a “Mad Lib”-type text in which the plot is preset but students can determine the characters’ emotional responses. The instructor could provide an excerpt from a familiar text, and the class could discuss how the impact of the passage would change if the characters expressed different emotions. The following example is again from the

Frau Holle Text:

Als sie vor das Haus der Frau Holle kam, fürchtete sie sich nicht, und fing gleich mit ihrem Dienst an. Am ersten Tag war sie fleißig und folgte der Frau Holle, wenn sie ihr etwas sagte, denn sie dachte an das viele Gold, das sie ihr schenken würde. Am zweiten Tag aber fing sie schon an zu faulenzen, am dritten noch mehr, da wollte sie morgens gar nicht aufstehen. Sie machte auch das Bett der Frau Holle nicht und schüttelte es nicht, dass die Federn aufflogen. Das gefiel Frau Holle überhaupt nicht und sie wollte den Dienst der Faulen beenden. Die Faule war damit zufrieden und meinte, nun würde der Goldregen kommen.

‘As she arrived at the house of Frau Holle, she had no fear and immediately began with her duties. On the first day she was hard-working and obeyed Frau Holle whenever she spoke to her, because she thought of the gold that she would bestow on her. However, on the second day she didn’t even want to get up. She didn’t make Frau Holle’s bed either, nor shake it so that the feathers flew. Frau Holle wasn’t at all happy with this and wanted to end the service of the lazy sister. The lazy one was happy with this and thought that the rain of gold would now arrive.’

Pechmarie’s reaction to Frau Holle’s house, her lack of fear both contrasts with her sister’s anxiety and meshes with an ongoing characterization of Pechmarie as someone who pays little attention to others. If the author had instead characterized Pechmarie with a token of

AFFECT: in/security instead (as some of the student writers do), the reader might be better disposed to sympathize with her. Each of the underlined tokens above might offer an avenue for

274

students to discuss the impact of an authorial choice by proposing other options the author might have employed.

Figure 20. Potential Matrix for Instructional Activity on AFFECT.

In the final Independent Construction stage, students work on their own to produce an instance of the genre they have been exploring. The instructional sequence described here

275

envisions an instructor interested in developing their students’ ability to convey evaluation and evoke a positive emotional response through AFFECT; this overarching goal should factor into the writing task students are set at the end of the unit. Instructors might adapt the task from Study 1 and include more explicit instructions for students to rewrite a familiar story so that the hero is perceived as the villain, or vice versa.

6.4.2 JUDGMENT in Historical Narratives

Intermediate and advanced curricular levels of language instruction often include discussions of consequential periods in German history through the lens of eyewitness accounts

(e.g., the unit on Civil Courage and Opposition in Nazi-Germany and Today described in

Ryshina–Pankova [2016] or the units on the German Democratic Republic described in Eigler

[2001]). One goal of the GUGD (and many other language departments) is to foster students’ awareness of the varied experiences undergone by the many groups that make up the German- speaking realm (Cunningham, Pfeiffer, & Ryshina–Pankova, 2018) — or at least students’ recognition that there was no one experience common to all residents of, for example, East

Germany. Students working with autobiographies, epistolary writing, or other narratives from such a time period might exploit the insights from the JUDGMENT section of Chapter 4 to explore how eyewitnesses evaluated the various actors and groups with whom they lived. In East

Germany, for instance, citizens differed widely in their devotion to the state; while many served as vocal members of the Socialist Unity Party, others expressed lukewarm feelings about the party while working as unofficial informants for the Ministry for State Security (Stasi); some dissidents denounced the regime and fled to the West, while others agitated to reform the state from within. The testimony of firsthand observers is often ambivalent, complicated by ties of friendship or family to members of any or all of these groups.

276

To explore these nuances, students might analyze a model text in terms of the targets of

JUDGMENT to see which actors are commended or criticized most often. Such an activity could form the Deconstruction stage in an instructional unit on historical narratives. As in O’Donnell’s

(2014) analysis of a political speech, students could also then track the attributes that these characters possess, then assess which attributes are most often praised or condemned, to gauge which attributes a character most values. This approach might help students avoid framing the time period under discussion in a dichotomous way, viewing one group of participants as heroes and the other as villains; by characterizing an author based on their values instead of their group membership, students can better appreciate the multiplicity of perspectives that make up the cultural conversation at any point in time.

The Drei Freunde text discussed in Chapter 2 (Humann, 1991) provides an opportunity to illustrate how such a lesson might begin. While the author is fairly direct in his criticism of the

East German authorities, he characterizes Eberhardt (the ostensible villain) not only by identifying him with the state, but also by juxtaposing him and his ilk with other young men who had keine Lust (. . .), diesem verlogenen Staat zu dienen ‘no desire to serve this false regime’:

Junge Männer, die keine Lust hatten, diesem verlogenen Staat zu dienen, hatten nun keine Möglichkeit mehr, in die BRD auszuweichen. Aber es gab auch junge Männer, die glaubten, daß das Ehrenkleid, wie die Uniform blasphemisch genannt wurde, anzulegen, ihre patriotische Pflicht sei. So verpflichtete sich Eberhardt, wie einer meiner Freunde hieß, den Diktatoren in der DDR für 20 Jahre, also quasi ein Leben lang, zu dienen. Im Jahre l968 wurde er beauftragt, mit der Waffe in der Hand jeden wehrlosen DDR Bürger am Verlassen des Landes zu hindern. Eberhardt, wenn er auf Urlaub war und wir ein Glas Bier zusammen tranken, erzählte gern davon, was er bei der Armee für Heldentaten vollbrachte. Er versäumte keine Gelegenheit darauf zu verweisen, daß er Geheimnisträger sei, prahlte aber andererseits damit, daß Eingeweihte wie er wissen, wie man die Sicherheitsanlagen überwinden konnte.

‘Young men who had no desire to serve this false regime, now had no opportunity to flee to the FRG. But there were also young men who believed that the attire of honor, as the uniform was blasphemously called, was (subj.) their patriotic duty. Eberhardt, one of my friends, felt just so duty-bound to serve the dictators in the GRD for twenty years — in other words, about a lifetime. In the year 1968 he was charged to prevent all wehrlose GRD citizens from leaving the country, with a weapon in his hand. Eberhardt, whenever he was on

277

leave and we drank a glass of beer together, would gladly talk about the heroic deeds (- JUDGMENT: propriety) he accomplished in the army. He lost no opportunity to point out that he was (subj.) (-JUDGMENT: propriety) a keeper of secrets, but boasted (-JUDGMENT: veracity) on the other hand that insiders like him knew how to bypass the security checkpoints.’

The passage evaluates four characters or groups — the GDR state, those opposed to the

GDR, those who supported it, and Eberhardt specifically. Each character is associated with at least one other by logico-semantic relationships, such as the conjunction aber (juxtaposing the two groups of men) and the adverb so (linking Eberhardt with other supporters). Instructors working with this text could direct students to identify the various characters who are judged as well as the characters with whom they are compared or contrasted, using a matrix such as the one in Figure 21:

JUDGMENT Judgment Token Compared to / Judgment Target Contrasted with Type whom? Young keine Lust hatten, diesem in opposition to +propriety dissidents verlogenen Staat zu dienen young pro-regime (invoked) men Young pro- glaubten, daß das Ehrenkleid in opposition to -propriety regime men (. . .) ihre patriotische Pflicht sei young dissidents +tenacity East German verlogen in alignment with -propriety government Diktatoren young pro-regime men Eberhardt erzählte gern davon, was er bei der in alignment with -propriety Armee für Heldentaten vollbrachte young pro-regime (invoked) versäumte keine Gelegenheit darauf men zu verweisen, daß er Geheimnisträger sei, prahlte aber andererseits…

Figure 21. Potential Matrix for Instructional Activity on JUDGMENT.

278

Students can then use the linkages between characters to draw conclusions about the kinds of behaviors the author favors. In the case of GDR supporters, the phrase glaubten, daß das Ehrenkleid (. . .) ihre patriotische Pflicht sei ‘believed that the attire of honor (. . .) was

(subj.) their patriotic duty’ could be read as JUDGMENT of positive tenacity. Patriotism’s association with these characters, who are otherwise cast in a negative light, suggests in turn that the author himself has a dim view of this kind of patriotic duty.

Using JUDGMENT in the Deconstruction phase in order to trace linkages between characters can help students identify the broader values that the author holds, and thus be better equipped to speak to these values in their own texts.

6.4.5 APPRECIATION and Assessment

In her study of student writing, Derewianka (2007) suggested that more experienced students express themselves using APPRECIATION more often in cases when less experienced students might use AFFECT, reflecting an ‘institutionalization of feeling’ (Martin & White, 2005) whereby writers project an evaluation onto the community at large. As writers compose texts with more of a public orientation, a reaction such as I was upset by the fireworks (-AFFECT: dis/satisfaction) or the fireworks were upsetting to me (-COVERT AFFECT) might be reconceptualized as the fireworks were disruptive (-APPRECIATION: valuation), implying a broader community affected by phenomenon being evaluated.

Texts in Study 2 seemed to bear out an analogous pattern; lower-proficiency writers who used APPRECIATION did so primarily to describe the effect of a situation on the narrator personally (via APPRECIATION: impact), while higher-proficiency described those effects in terms of ramifications in the community (using APPRECIATION: valuation). This finding has implications for best practices in the area of assessment (see below for a more detailed

279

discussion). When presenting grading rubrics and offering feedback on student writing, instructors should offer concrete examples of lexicogrammatical usage that characterize a successful writing task. When instructors’ expectations are not communicated clearly, or instructors offer vague, generalized feedback, or instructors’ actual internal grading criteria do not match their published rubrics, students may miss opportunities to compose texts that achieve their communicative goals (Lancaster, 2011). In an articulated curriculum such as that of the

GUGD, in which students are prompted to write with an increasing awareness of a broad audience, use of APPRECIATION offers one concrete suggestion instructors could offer when giving feedback on student writing. Instructors can list examples of APPRECIATION: valuation

(after having discussed their use in model texts) among a number of word-level resources that we would expect to see in a text with an interpersonally distant readership.

6.4.4 Implicit Evaluation

Implicit evaluation can be particularly difficult for L2 students to master. As Myskow and Ono (2018) point out, “invoked tokens can be realized by a wide variety of language features, making it challenging for teachers to illuminate for students the specific ways implicit realizations (. . .) may invoke explicit judgments” [p. 58]. In Chapter 4, I discussed how learners at various levels drew on broader meaning-making options by using juxtapositions, syntactically connecting more and less explicit evaluations, and matching descriptions of an inner emotional state with physical manifestations of that state.

Some of these means of employing provoked evaluation proved within the reach of some

Level I and Level II writers, which suggests that they could be productively framed in an instructional context. Examples from Level I texts suggest that beginners can adopt logico- semantic connection as an evaluative technique. In Example (7), also discussed previously, a

280

Level II writer used a list of other positive attributes to indicate that they also saw konservativ as a praiseworthy trait.

(7) Sie ist schlau, konservativ, und sie sieht gut aus. ‘She is clever, conservative, and good-looking’

In an additional instance of juxtaposition, the Level I writer in Example (8) uses his location as a means of implying a negative emotional state:

(8) Meinen Freunden waren alles ins Schwimmbad und haben Spaß gemacht, aber ich war in die Bibliothek My friends were all in the pool having fun (+AFFECT: un/happiness: misery/cheer; inscribed), but I was in the library (-AFFECT: un/happiness: misery/cheer; invoked)’

The narrator describes their friends explicitly as enjoying themselves; aber in this sentence indicates that the second half of the sentence is meant to be interpreted in the opposite light.

These implicit evaluations, pertaining to descriptions of friends and everyday activities such as pool parties, could easily be found in texts representative of introductory genres—e.g., letters to friends or children’s stories—and the identification of such emotions in introductory texts serves as important stepping stone both to in-class discussions of evaluation generally, and to the use of implicit evaluation in students’ own texts (Kuhn, in press). Instructors interested in working with beginners on implicit evaluation might select a text in which the author compares their situation to that of their friends.

In Chapter 5, I analyzed the case of one L1 writer who effectively described their classmates using a succession of evaluative adjectives; while some of these, such as ehrgeizig

‘ambitious’ might have indicated praise in other contexts, their inclusion in a list of other condemnations made it clear that they were meant as criticism. These instances of semantically connecting multiple evaluative tokens brings to mind a closely related evaluative technique, in

281

which one word in a list is “syntactically related to two or more words, with only one of which it seems logically connected” (New World Dictionary of the American Language, Second College

Edition, 1980). Known as zeugma, this is a figure of speech that can lead to an arresting dissonance. While there were no examples of this in the learner texts, the following example may provide an illustration: The first three seasons of the TV show Buffy the Vampire Slayer featured a particularly autocratic high school principal who disapproved of most typical adolescent behavior. In one episode, he mentioned a list of Halloween-related behaviors he found offensive:

“Tossing eggs, keying cars, bobbing for apples…” (Ellsworth & Green, 1997). Apple-bobbing’s inclusion in this list makes it clear that the character finds this sort of levity just as objectionable as other destructive behaviors; the dissonance serves both as a moment of comic relief and as a way to implicitly highlight the principal’s affinity for order.

This evaluative technique can be used to humorous effect and might offer one avenue for drawing students’ attention to examples of this kind of implicit evaluation in their own’ and others’ writing. Students and instructor could jointly analyze a text containing instances of semantically connected evaluations, some of which involve a zeugma; students might then engage in a creative writing project to describe a character using potentially ambiguous attributes whose semantic connections make the overall evaluation clear. This assignment might contribute to a longer writing task in which students produce a letter to a friend, a book review, or another instance of a genre in which humor is common.

6.4.5 Insights Related to Assessment

A healthy cycle of instruction does not end with the conclusion of a lesson; it also includes a phase of assessment to gauge students’ progress and guide the process of curricular renewal. Meaningful assessment requires that teachers be able to call on operationalizable

282

constructs of the competences they expect to see in their students (Byrnes, Maxim, & Norris,

2010), to avoid mismatches between stated instructor rubrics and the implicit expectations those instructors may use in their evaluations (Lancaster, 2011).

Within the instructional context of the GUGD, recent curricular development efforts have focused on the assessment of humanities learning goals, in the interest of drawing “transparent connections between FL study and humanistic inquiry” (Ryshina–Pankova, 2015b, p. 221) as one means of countering the language–literature dichotomy splitting many university foreign language departments. A report on the department’s Humanities Learning Assessment Project argues that FL study can and must be conceptualized as an integral part of humanistic scholarship, in that language learning involves grappling with texts that “enable exploration of human action, intention, and reflection in a particular cultural realm” (Ryshina–Pankova, 2015b, p. 222). In the course of the project, a task force of instructors revised curricular materials, including assessment rubrics and prototypical writing tasks, to reflect the following goals for students’ development as humanistic thinkers: “(a) the learning of cultural content, (b) understanding of cultural content through multiple perspectives within the German speaking area, and (c) reflective stance towards various viewpoints on content including students’ own perspectives as learners at an American university” (p. 7). In the course of their work on the project, the instructors clarified their own sometimes-implicit understandings of the textual elements that indicated students’ developing mastery of those areas. Determining the extent to which students demonstrated their achievement of these goals proved a subjective process; rating student essays according to their awareness of cultural content, multiple perspectives, and reflective stance yielded some discrepancies. While reflective stance was the most common source of interrater disagreements (38.2% of cases), multiple perspectives accounted for just

283

slightly fewer instances (34%). The analysis described in this thesis, appropriately scaled down, represents one means of bringing greater objectivity to holistic assessments such as the humanities learning goals of the GUGD.

Many of the components of an APPRAISAL analysis bear in particular on the expression of multiple perspectives, in several ways: (a) In the course of coding for AFFECT, it is useful to mark the source of each token; a text that is more balanced in its sources of AFFECT is more likely to include a diversity of perspectives; (b) shifting patterns of ATTITUDE use within a text may indicate different voices within that text (Coffin, 2002, 2006); and (c) implicit evaluation often depends on a shared understanding of values and norms with which both writer and reader must be familiar.

Put another way, running an APPRAISAL analysis dovetails nicely with an assessment approach that foregrounds humanities learning goals. Tracking the sources and targets of

APPRAISAL tokens is useful for an analysis of AFFECT and JUDGMENT generally; it also provides a ‘gut check’ of the extent to which an author draws on multiple perspectives in their writing. An

APPRAISAL analysis also involves consideration of implicit evaluative strategies, such as judging individual actors as a means of praising particular traits or worldviews.

This last point may be most relevant to a humanities-oriented assessment of learner texts.

Awareness of multiple perspectives, of course, reflects much more than the ability to simply write a story with multiple speaking characters. Students should be able to understand how “the deeper assumptions, emotions, interests, and ideologies as well as means of expression (e.g., music vs. lexicogrammar of emotion) that lie behind [particular issues], actively shape[] our knowledge” (Ryshina–Pankova, 2015b, p. 11); APPRAISAL analysis offers one avenue of examining how these deeper assumptions are conveyed.

284

The particular field of study for which Ryshina–Pankova’s (2015b) goal statement was written is twentieth century German history. In this context, demonstrating awareness of multiple perspectives in contemporary accounts of life in East and West Germany requires both deploying background knowledge of the authors’ likely shared assumptions and analyzing the linguistic tools the authors use in order to make both explicit and implicit evaluations of their lived experiences. GUGD students do this though an analysis of the Drei Freunde text analyzed in

Chapter 2; each of the characters in this text displays a different attitude toward the East

Germany government, which is reflected to an extent in the author’s evaluation of himself and his friends.

The increased use of Invoked Appraisals among students at higher curricular levels of

Study 2 can be read as an indirect but encouraging sign that these students are better able to express multiple perspectives — not merely by putting words in more characters’ mouths, but by evaluating different voices’ values through reference to individual characters. Taken together, these considerations suggest that components of APPRAISAL — number of sources of AFFECT, patterns of ATTITUDE in different stages, and invoked evaluation that refers to community values

— could be built into a humanities learning assessment rubric, in the hope that these concrete examples would make such a rubric more objective and applicable to a wider range of curricular contexts.

285

6.5 Future Research

The process and results of this project suggest avenues for further inquiry into both the methodology of APPRAISAL analysis and its potential applications to assessment and program evaluation.

In terms of methodology, the coding process highlighted categories within APPRAISAL whose borders could be further clarified. As Benítez–Castro and Hidalgo–Tenorio (2019) argue and interrater reliability results from this project bear out, consistently differentiating between

AFFECT: un/happiness and AFFECT: dis/satisfaction often proves difficult, perhaps because different raters disagree on whether certain expressions of emotion are directed toward an unstated but implicit goal. Benítez–Castro and Hidalgo–Tenorio (2019) proposed a realignment of the APPRAISAL scheme by appealing to evidence from psychological studies rather than corpus analysis to justify categorical boundaries. Benítez–Castro & Hidalgo–Tenorio (2019) is not the only recent study to take this interdisciplinary tack; Goetze (2018) likewise drew on existing psychological constructs to design a framework for teacher anxiety. This project’s interrater discrepancies in AFFECT and JUDGMENT in particular suggest that there is room in future

APPRAISAL research for further such studies that draw on psychological research to justify updated category boundaries. It would be valuable to discover whether psychological studies on human perception of others’ capabilities could clarify or elide the distinction between

JUDGMENT: capacity and JUDGMENT: tenacity. Such research might also reveal benefits to modifying the APPRAISAL scheme based on the specific language being analyzed, in the event that cross-linguistic differences in the way information is presented (Slobin, 1996) affect, for instance, how explicitly writers usually identify a source of AFFECT. (In languages that more

286

often identify an agent even when using the passive voice, COVERT AFFECT might be less common, or might be realized in a different way.)

Benítez–Castro and Hidalgo–Tenorio (2019) have also proposed an ambitious restructuring of the entire ATTITUDE system that invites studies applying it to genres in which emotion is particularly prevalent.

In terms of assessment, this analysis explored ways in which students at several levels used evaluation that indexes social values, and expressed those evaluations through multiple voices, in ways that indicated their awareness of multiple perspectives. In section 6.2.5, I argued that students’ use of polyphony represented one metric by which their achievement of humanities learning goals (Cunningham, Pfeiffer, & Ryshina–Pankova, 2018) could be assessed. This observation could be expanded further into a project that explicitly links APPRAISAL analysis to the assessment practices that inform a department’s curriculum development and renewal.

Section 6.1 raised the concern that texts for this project were drawn from students who had common instructional experience within the same departmental curriculum. This could open results to the objection that GUGD students’ facility with evaluative language might not be representative of other L2 German students with similar experience with language study (e.g., if overt instruction in evaluative language is common in GUGD courses). To the extent that these results contribute to GUGD’s own assessment and curricular renewal efforts, however, that focus on a singular department could be considered a strength.

Effective program evaluation that can support curricular innovation and weigh the effectiveness of instructional practices requires assessment practices that include transparent, operationalizable descriptions of successful student work (Byrnes, Maxim, & Norris, 2010).

When student-facing assessment rubrics do not reflect actual grading practices, it is infeasible to

287

use those assessments to evaluate the strengths and weaknesses of the curriculum. It is also crucial to explicitly state shared departmental understandings of how students manifest in their writing the aspects of literacy that instructors aim to foster; one benefit of the intradepartmental conversations described in Cunningham, Pfeiffer, & Ryshina–Pankova (2018) was the opportunity to clarify each instructor’s understanding of concepts such as reflective stance and how those concepts might be realized linguistically.

There is precedent for using APPRAISAL as a tool for assessment in the context of program evaluation; Lancaster (2011) demonstrated one way in which APPRAISAL analysis can expose discrepancies between instructors’ professed expectations and the criteria they tend to actually use when grading. In a similar vein, the GUGD or another department which prioritizes humanities learning goals could incorporate APPRAISAL into its assessment cycle to gauge both students’ progress in achieving those goals and instructors’ agreement on how successful progress toward those goals manifests at the word and sentence level.

6.6 Final Thoughts

Evaluative language ultimately draws its force from humans’ nature as social animals who “are constantly communicating their evaluations: those about themselves, the actions of other people and the circumstances of their interactions with others. When these evaluations are good they may lead to praise, and when bad to criticism and blame. Evaluation acts as both social motivation and restraint; in fact, as Darwin underlines, it is one of the forces holding society together” (Partington, 2017, p. 192). At a smaller scale, language learners writing fairy tales and narratives about the meaning of friendship are engaged in this same conversation, using evaluation in order to define and redefine their identities relative to their audience(s). If this

288

project offers language teachers additional insight as they support language learners in these fundamentally human pursuits, I consider it time well spent.

289

APPENDIX A: STUDY PROMPTS

Study 1

Students who agreed to participate in Study 1 received an e-mail containing scheduling information and the Frau Holle text. When participants arrived for the writing session, they received and signed the task prompt shown below.

Pre-Session E-mail

Thank you for agreeing to participate in my research study.

Attached to this e-mail you will find the text of a German fairy tale (or Märchen). This is a version of one of the Märchen collected by the brothers Jacob and Wilhelm Grimm. Before attending the upcoming research session, please read the Märchen to get the gist of the story.

You may read through the Märchen as many times as you wish. Do not worry if you do not understand every word.

Please read the Märchen before arriving at the research session on [TIME] in [PLACE.]

I appreciate your participation. Please don’t hesitate to get in touch if you have any questions.

Task Prompt

Thank you for agreeing to participate in this research study.

290

By now, you should have read the Märchen (Frau Holle) which I e-mailed to you. You will find a copy of the Märchen stapled to this prompt as well.

Having read the Märchen, please use GoogleDocs to write your own version of the story in which you take on the perspective of one of the characters (of your choice).

You are invited to include orientation, complicating action, and resolution stages (as is typical of narratives) and to present your character’s inner thoughts and attitudes. You may wish to do some or all of the following in your narrative:

— explore your character’s relationships with other characters in the story

— express how your character feels about the events that take place

— address your audience either explicitly or implicitly.

Remember to take on the perspective, thoughts, and attitudes of your chosen character rather than simply summarizing the action in the story.

This is not a writing task and will not be assessed for errors in grammatical details such as gender articles, adjective endings, and verb conjugations; do not worry if you cannot remember the exact word you’d like to use or the precisely correct way to express a particular sentiment.

That said, please do not use any external aids (e.g., a dictionary, dict.cc, Linguee) while writing your narrative; we are interested in your ability to use language at this point in your studies.

291

Please feel free to take as much time as you like to write your narrative. Once you have finished, please save your document using the format “Narrative_[Your GUID]” and share it with me at [email protected].

Thank you once again for taking the time to participate. If you would like to be contacted with a report on the findings from this survey, please add your e-mail address here:

______

After you have signed and dated this form, please begin writing your narrative.

I, ______[name] have read and understood this prompt, as well as the informed consent form, and all of my questions about this study have been answered by the investigator.

Signature: ______

Date: ______

292

Study 2

Task Prompt

Thank you for agreeing to participate in this research study. This study is being done in order to elicit patterns of language realization at various levels of German proficiency. Exploring these patterns is an important step toward being able to effectively teach language learners to engage their audience in the ways they mean to.

Once you have finished reading these instructions, please open a new GoogleDoc on your laptop and begin writing a story in German in response to the following prompt:

Write a letter to either your parents or some more distant relatives about a time at Georgetown when you were in a difficult situation and a friend helped you, and describe your reaction to those events.

There is no particular required length for this story; it can be as long or as short as it needs to be in order for you to relate the events of the story and your reaction to them.

You are invited to include typical stages of a narrative (such as orientation, complicating action, and resolution stages) and to present your inner thoughts and attitudes. You may wish to do some or all of the following in your story:

— explore your relationships with the other people you describe

— express how you feel about the difficulty that you faced

293

— describe the character qualities of your friend

— address your audience either explicitly or implicitly.

You are encouraged to describe your perspectives, thoughts, and attitudes in addition to summarizing the action in the story.

That said, please do not use any external aids (e.g., a dictionary, dict.cc, Linguee) while writing your narrative; we are interested in your ability to use language at this point in your studies.

Please feel free to take as much time as you like to write your narrative (up to 45 minutes). Once you have finished, please save your document using the format “Narrative_[Your netID]” and share it with me at [email protected].

Thank you once again for taking the time to participate. If you would like to be contacted with a report on the findings from this survey, please add your name, e-mail address, and signature here:

Name: ______

Email address: ______

Signature: ______

294

APPENDIX B: MÄRCHEN TEXT

Students in Study 1 received this Märchen text via email before they arrived for the writing session. Translations given here in brackets were provided as MS Word comments.

Frau Holle

Eine Witwe [widow] hatte zwei Töchter, davon war die eine schön und fleißig [hardworking], die andere hässlich und faul. Sie hatte aber die hässliche und faule, weil sie ihre richtige Tochter war, viel lieber, und die andere musste alle Arbeit machen und das Aschenputtel [Cinderella] im

Haus sein. Das arme Mädchen musste sich täglich auf die große Straße neben einem Brunnen

[fountain] setzen und musste so viel spinnen [to spin (wool)], dass ihm18 das Blut aus den

Fingern spritzte [to spurt].

Nun passierte es, dass die Spule [spool (of thread)] einmal ganz blutig war, da wollte es die

Spule in dem Brunnen abwaschen: die Spule sprang ihm aber aus der Hand und fiel hinab. Es weinte, lief zur Stiefmutter [stepmother] und erzählte ihr das Unglück. Sie schimpfte mit ihr aber so heftig und war so unbarmherzig, dass sie sprach: „Hast du die Spule hinunterfallen lassen, so hole sie auch wieder herauf.“ Da ging das Mädchen zu dem Brunnen zurück und wusste nicht, was es machen sollte: Und in seiner Herzensangst sprang es in den Brunnen hinein, um die Spule zu holen. Es verlor die Besinnung [consciousness], und als es erwachte und wieder zu sich selber kam, war es auf einer schönen Wiese [plain], wo die Sonne schien und viele tausend Blumen standen. Auf dieser Wiese ging es fort und kam zu einem Backofen, der war voller Brot; das

Brot aber rief: „Ach, zieh mich raus, zieh mich raus, sonst verbrenn ich: Ich bin schon längst

18 NB: Since das Mädchen is considered neuter, the girl will be referred to as es and ihm throughout the tale.

295

ausgebacken [baked long enough].“ Da trat es näher und holte mit dem Brotschieber [baker’s peel; https://i.ebayimg.com/images/i/261989973187-0-1/s-l1000.jpg] alles nacheinander heraus.

Danach ging es weiter und kam zu einem Baum, der hing voll Äpfel, und rief ihm zu: „Ach schüttel [to shake] mich, schüttel mich, wir Äpfel sind alle reif.“ Da schüttelte es den Baum, dass die Äpfel fielen, als regneten sie, und schüttelte, bis keiner mehr oben war. Als es alle in einen

Haufen [heap] zusammengelegt hatte, ging es wieder weiter.

Endlich kam es zu einem kleinen Haus, daraus guckte eine alte Frau, weil sie aber so große

Zähne hatte, bekam das Mädchen Angst, und es wollte weglaufen. Die alte Frau aber rief ihm nach: „Was fürchtest du dich, liebes Kind? Bleib bei mir; wenn du alle Arbeit im Haus ordentlich tun willst, so soll es dir gut gehen. Du musst nur Acht geben [pay attention], dass du mein Bett gut machst und es fleißig aufschüttelst [to shake up / to fluff], dass die Federn

[feathers] fliegen, dann schneit es in der Welt. Ich bin die Frau Holle.“ Weil die Alte ihm so gut zusprach, so willigte [einwilligen: to acquiesce] das Mädchen ein und begab sich in ihren Dienst.

Es besorgte [to see to] auch alles zu ihrer Zufriedenheit und schüttelte ihr das Bett immer gewaltig [powerfully] auf, dass die Federn wie Schneeflocken umherflogen. Dafür hatte es auch ein gutes Leben bei ihr, kein böses Wort, und alle Tage Gekochtes und Gebratenes [cooked and roasted (food)]. Nun war es eine Zeit lang bei der Frau Holle, da wurde es traurig und wusste anfangs selbst nicht, was ihm fehlte [to be lacking]. Endlich merkte es, dass es Heimweh

[homesickness] war; ob es ihm hier gleich viel tausendmal besser ging als zu Hause, so hatte es doch ein Verlangen [desire] dahin. Endlich sagte es zu ihr: „Ich habe zu Hause kein einfaches

Leben, aber wenn es mir auch noch so gut hier unten geht, so kann ich doch nicht länger bleiben, ich muss wieder hinauf zu den Meinigen.“ Die Frau Holle sagte: „Es gefällt mir [to be pleasing],

296

dass du wieder nach Hause möchtest, und weil du mir so treu [devotedly] gedient hast, so will ich dich selbst wieder hinaufbringen.“

Sie nahm es darauf bei der Hand und führte es vor ein großes Tor [gate]. Das Tor öffnete sich, und wie das Mädchen gerade darunter stand, fiel ein gewaltiger Goldregen [rain of gold], und alles Gold blieb an ihm hängen, so dass es über und über davon bedeckt [covered/coated] war.

„Das sollst du haben, weil du so fleißig [hardworking] gewesen bist“, sprach die Frau Holle und gab ihm auch die Spule wieder, die ihm in den Brunnen gefallen war. Darauf wurde das Tor verschlossen, und das Mädchen befand sich oben auf der Welt, nicht weit vom Haus seiner

Mutter. Und als es in den Hof kam, saß der Hahn auf dem Brunnen und rief:

„Kikeriki [cock crowing sound] unsere goldene Jungfrau ist wieder hie.“

Da ging es hinein zu seiner Mutter, und weil es so mit Gold bedeckt ankam, wurde es von ihr und der Schwester gut aufgenommen.

Das Mädchen erzählte alles, was ihm passiert war, und als die Mutter hörte, wie es zu dem großen Reichtum gekommen war, wollte sie der andern hässlichen und faulen Tochter gerne dasselbe Glück verschaffen [to provide]. Sie musste sich an den Brunnen setzen und spinnen.

Und damit ihre Spule blutig wurde, stach sie sich in die Finger und stieß sich die Hand in die

Dornenhecke [hedge of thorns]. Dann warf sie die Spule in den Brunnen und sprang selber hinein. Sie kam, wie die andere, auf die schöne Wiese und ging auf demselben Weg weiter. Als sie zu dem Backofen kam, schrie das Brot wieder: „Ach, zieh mich raus, zieh mich raus, sonst

297

verbrenn ich, ich bin schon längst ausgebacken.“ Die Faule aber antwortete: „Als ob ich Lust hätte [as if I wanted…], mich schmutzig zu machen!“, und ging fort.

Bald kam sie zu dem Apfelbaum, der rief: „Ach, schüttel mich, schüttel mich, wir Äpfel sind alle miteinander reif.“ Sie antwortete aber: „es könnte mir einer auf den Kopf fallen!“ und ging weiter.

Als sie vor das Haus der Frau Holle kam, fürchtete sie sich [to be afraid] nicht, und fing gleich mit ihrem Dienst an. Am ersten Tag war sie fleißig und folgte der Frau Holle, wenn sie ihr etwas sagte, denn sie dachte an das viele Gold, das sie ihr schenken würde. Am zweiten Tag aber fing sie schon an zu faulenzen [to be lazy], am dritten noch mehr, da wollte sie morgens gar nicht aufstehen. Sie machte auch das Bett der Frau Holle nicht und schüttelte es nicht, dass die Federn aufflogen. Das gefiel Frau Holle überhaupt nicht und sie wollte den Dienst der Faulen beenden.

Die Faule war damit zufrieden und meinte, nun würde der Goldregen kommen. Die Frau Holle führte sie auch zu dem Tor, als sie aber darunter stand, wurde statt des Goldes ein großer Kessel

[cauldron] voll Pech [pitch] ausgeschüttet. „Das ist zur Belohnung deiner Dienste“, sagte die

Frau Holle und schloss das Tor zu. Da kam die Faule heim, aber sie war ganz mit Pech bedeckt, und der Hahn auf dem Brunnen, als er sie sah, rief:

„Kikeriki, unsere schmutzige Jungfrau ist wieder hie.“

Das Pech aber blieb fest an ihr hängen und wollte, solange sie lebte, nicht abgehen...

298

REFERENCES

Achugar, M., & Oteíza, T. (2009). “In whatever language people feel comfortable”: Conflicting language ideologies in the US Southwest border. Text & Talk, 29(4), 371–391.

Adendorff, R., & de Klerk, V. (2006). The role of APPRAISAL resources in discussing AIDS. Corpus Linguistics and World Englishes, 1, 70–88. Adendorff, R., & Pienaar, K. (2013). ’Busty babes and passionate pleasures’: A systemic functional linguistic analysis of sex worker discourse in a South African city. In G. O’Grady, T. Bartlett, & L. Fontaine (Eds.), Choice in language: Applications in text analysis (pp. 31–54). Sheffield, UK: Equinox. Aull, L. L., & Lancaster, Z. (2014). Linguistic markers of stance in early and advanced academic writing: A corpus-based comparison. Written Communication, 31(2), 151–183. https://doi.org/10.1177/0741088314527055 Bakhtin, M. M. (1981). The dialogic imagination: Four essays by M. M. Bakhtin. (M. Holquist, Ed.). Austin, TX: University of Texas Press. Bakhtin, M. M. (1986). The problem of speech genres. In C. Emerson & M. Holquist (Eds.), Speech genres and other late essays. (pp. 60–102). University of Austin Press. Banks, D. (2017). A systemic functional grammar of French. London: Routledge. Barletta, N., Mizuno, J., & Moss, G. (2013). The use of appraisal resources in the construction of second language teacher-researcher identity. In G. O’Grady, T. Bartlett, & L. Fontaine (Eds.), Choice in language: Applications in text analysis. Sheffield, UK: Equinox Publishing Ltd. Barton, E. L. (1993). Evidentials, argumentation, and epistemological stance. College English, 55(7), 745–769. Bednarek, M. (2006). Evaluation in media discourse: Analysis of a newspaper corpus. New York: Continuum. Bednarek, M. (2007). Polyphony in Appraisal: Typological and topological perspectives. Linguistics & the Human Sciences, 3(2), 107–136. https://doi.org/10.1558/lhs.v3i2.107 Bednarek, M. (2008). Emotion talk across corpora. Basingstoke, UK: Palgrave Macmillan. Bednarek, M. (2009a). Dimensions of evaluation: Cognitive and linguistic perspectives. Pragmatics & Cognition, 17(1), 147–174. Bednarek, M. (2009b). Language patterns and attitude. Functions of Language, 16(2), 165–192. Belcher, D. D. (1997). An argument for nonadversarial argumentation: On the relevance of the feminist critique of academic discourse to L2 writing pedagogy. Journal of Second Language Writing, 6(1), 1–21. https://doi.org/10.1016/S1060-3743(97)90003-5

299

Belcher, D. D. (2013). The scope of L2 writing: Why we need a wider lens. Journal of Second Language Writing, 22(4), 438–439. https://doi.org/10.1016/j.jslw.2013.08.006 Benítez–Castro, M., & Hidalgo–Tenorio, E. (2019). Rethinking Martin & White’s affect taxonomy: A psychologically-inspired approach to the linguistic expression of emotion. In J. L. Mackenzie & L. Alba–Juez (Eds.), Emotion in discourse (pp. 301–331). Philadelphia/Amsterdam: John Benjamins. Biber, D., & Zhang, M. (2018). Expressing evaluation without grammatical stance: Informational persuasion on the web. Corpora, 13(1), 97–123. https://doi.org/10.3366/cor.2018.0137 Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. London: Longman. Boye, K. (2012). Modality. In C. Chapelle (Ed.), The encyclopedia of applied linguistics. Hoboken, NJ: John Wiley & Sons, Inc. https://doi.org/10.1002/9781405198431.wbeal0769 Bybee, J., & Fleischman, S. (1995). Modality in grammar and discourse: An introductory essay. In J. Bybee & S. Fleischman (Eds.), Modality in grammar and discourse (pp. 1–14). Philadelphia/Amsterdam: John Benjamins. Byrnes, H. (2001). Reconsidering graduate students’ education as teachers: “it takes a department!” Modern Language Journal, 85(4), 512–530. Byrnes, H. (2002). Toward academic-level foreign language abilities: Reconsidering foundational assumptions, expanding pedagogical options. In B. L. Leaver & B. Shekhtman (Eds.), Developing professional-level language proficiency (pp. 34–58). Cambridge: Cambridge University Press. Byrnes, H. (2006). What kind of resource is language and why does it matter for advanced language learning? In H. Byrnes (Ed.), Advanced language learning: The contribution of Halliday and Vygotsky (pp. 1–28). London: Continuum. Byrnes, H. (2009). Emergent L2 German writing ability in a curricular context: A longitudinal study of grammatical metaphor. Linguistics and Education, 20, 50–66. https://doi.org/10.1016/j.linged.2009.01.005 Byrnes, H., & Kord, S. (2001). Developing literacy and literacy competence: Challenges for foreign language departments. In SLA and the literature classroom: Fostering dialogues (pp. 35–73). Byrnes, H., Maxim, H., & Norris, J. (2010). Monograph issue: Realizing advanced foreign language writing development in collegiate education: Curricular design, pedagogy, assessment. Modern Language Journal, 94. Caldwell, D. (2009). ’Working your words’: Appraisal in the AFL post-match interview. Australian Review of Applied Linguistics, 32(2), 1–17.

300

Caldwell, D. (2014). The interpersonal voice: Applying appraisal to the rap and sung voice. Social Semiotics, 24(1), 40–55. https://doi.org/10.1080/10350330.2013.827357 Carretero, M., & Taboada, M. (2014). Graduation within the scope of Attitude in English and Spanish consumer reviews of books and movies. In G. Thompson & L. Alba–Juez (Eds.), Evaluation in context (pp. 221–239). Philadelphia/Amsterdam: John Benjamins. Chafe, W. (1986). Evidentiality in English conversation and academic writing. In W. Chafe & J. Nichols (Eds.), Evidentiality: The linguistic coding of epistemology (pp. 261–272). New York: Ablex. Chang, P., & Schleppegrell, M. (2016). Explicit learning of authorial stance-taking by L2 doctoral students. Journal of Writing Research, 8(1), 49–80. Chin, H. B., Taib, F., & Mei, C. C. Y. (2017). Constructing interpersonal meaning via teacher talk in Chinese as a Second Language classrooms: An Appraisal analysis. Journal of Modern Languages, 27, 24–47. Christie, F. (1997). Curriculum macrogenres as forms of initiation into a culture. In F. Christie & J. R. Martin (Eds.), Genre and institutions: Social processes in the workplace and school (pp. 134–160). London: Continuum. Christie, F. (1999). Genre theory and ESL teaching: A systemic functional perspective. TESOL Quarterly, 33(4), 759–763. Coates, J. (1987). Epistemic modality and spoken discourse. Transactions of the Philological Society, 85(1), 110–131. Coffin, C. (1998). Reconstruals of the past: Settlement or invasion? The role of JUDGEMENT analysis. Paper presented at the Annual Meeting of the American Association for Applied Linguistics (Seattle, WA, March 14–17). Coffin, C. (2002). The Voices of History: Theorising the interpersonal semantics of historical discourses. Text, 22(4), 503–528. Coffin, C. (2006). Learning the language of school history: The role of linguistics in mapping the writing demands of the secondary school curriculum. Journal of Curriculum Studies, 38(4), 413–429. Coffin, C., & O’Halloran, K. (2005). Finding the global groove: Theorising and analysing dynamic reader positioning using APPRAISAL, corpus, and a concordancer. Critical Discourse Studies, 2(2), 112–132. Cope, B., & Kalantzis, M. (2009). "Multiliteracies”: New literacies, new learning. Pedagogies: An International Journal, 4(3), 164–195. Cunningham, D. J., Pfeiffer, P. C., & Ryshina–Pankova, M. (2018). Assessing humanities learning in an integrated undergraduate German curriculum. Die Unterrichtspraxis/Teaching German, 51(2), 144–155. https://doi.org/10.1111/tger.12070

301

Danbin, W. (2015). The relationship between voice development and the quality of EFL writing: A study on first-person pronouns in the revision process. Chinese Journal of Applied Linguistics, 28(1), 110–127. https://doi.org/10.1515/cjal-2015-0006 Davis, J. M., Kim, A., McKay, T., Niu, M., Son, Y., & Venezia, F. (2016). Useful evaluation in language programs. In Georgetown University Round Table on Languages and Linguistics: Useful assessment and evaluation in language education (March 11–13, Washington, DC). Dawkins, R. (2011). The tyranny of the discontinuous mind. New Statesman, 19, 54–57. Dennett, D. C. (1995). Darwin’s dangerous idea. New York: Simon & Schuster. Dennett, D. C. (2014). Intuition pumps and other tools for thinking. New York: W. W. Norton & Company. Derewianka, B. (2007). Using appraisal theory to track interpersonal development in adolescent academic writing. In A. McCabe, M. O’Donnell, & R. Whittaker (Eds.), Advantages in language and education (pp. 142–165). London: Continuum. Don, A. (2016). “It is hard to mesh all this”: Invoking attitude, persona and argument organisation. Functional Linguistics, 3(1). https://doi.org/10.1186/s40554-016-0033-1 Dueñas, P. M. (2012). Hyland, Ken. In C. A. Chapelle (Ed.), The encyclopedia of applied linguistics. Oxford: Blackwell Publishing Ltd. https://doi.org/10.1002/9781405198431.wbeal0518 Eigler, F. (2001). Language and content articulation at college level. Die Unterrichtspraxis/Teaching German, 34(2), 107–118. Eggins, S. (2004). An introduction to systemic functional linguistics (2nd ed.). London: Bloomsbury. Elbow, P. (1981). Writing with power. New York: Oxford University Press. Elbow, P. (1994). What do we mean when we talk about voice in texts? In K. B. Yancey (Ed.), Voices on voice: Perspectives, definitions, inquiry (Vol. 1, pp. 1–35). Urbana, IL: National Council of Teachers of English. Ellsworth, C. (Writer), & Green, B. S. (Director). (1997). Halloween [Television series episode]. In J. Whedon (Executive producer), Buffy the Vampire Slayer. Los Angeles, CA: Mutant Enemy Productions. Ewbank, M. P., Cummins, R., Tablan, V., Bateup, S., Catarino, A., Martin, A. J., & Blackwell, A. D. (2020). Quantifying the association between psychotherapy content and clinical outcomes using deep learning. JAMA Psychiatry, 77(1), 35–43. https://doi.org/10.1001/jamapsychiatry.2019.2664 Fairclough, N. (2001). Language and power. London: Longman.

302

Flowerdew, J., & Wang, S. H. (2015). Identity in academic discourse. Annual Review of Applied Linguistics, 35, 81–99. https://doi.org/10.1017/S026719051400021X Fløttum, K., Dahl, T., Kinn, T., Müller Gjesdal, A., & Vold, E. T. (2007). Cultural identities and academic voices. In K. Fløttum (Ed.), Language and discipline perspectives on academic discourse (pp. 14–39). Cambridge: Cambridge Scholars Publishing.

Fuoli, M. (2015). A step-wise method for annotating APPRAISAL. Functions of Language, 25(2), 1–26. Gales, T. (2011). Identifying interpersonal stance in threatening discourse: An appraisal analysis. Discourse Studies, 13(1), 27–46. https://doi.org/10.1177/1461445610387735 Gee, J. P. (1998). What is literacy? In V. Zamel & R. Spack (Eds.), Negotiating academic literacies: Teaching and learning across languages and cultures (pp. 51–59). Mahwah, NJ: Lawrence Erlbaum. Georgetown University (2020). Curriculum. Department of German. Retrieved June 19, 2020 from https://german.georgetown.edu/undergraduate/curriculum/ Gleason, J. (2014). Meaning-based scoring: A Systemic Functional Linguistics model for automated test tasks. Hispania, 97(4), 666–688. https://doi.org/10.1353/hpn.2014.0115 Goetze, J. (2018). Linking cognition and emotion: An appraisal study of foreign language teacher anxiety (Unpublished doctoral dissertation). Georgetown University, Washington, DC. Gray, B., & Biber, D. (2012). Current conceptions of stance. In K. Hyland & C. Sancho Guinda (Eds.), Stance and voice in written academic genres (pp. 15–48). Basingstoke, UK: Palgrave Macmillan. Gray, B., & Biber, D. (2015). Stance markers. In K. Aijmer & C. Rühlemann (Eds.), Corpus pragmatics: A handbook (pp. 219–248). Cambridge: Cambridge University Press. Gruber, H. (2004). Scholar or consultant? Author-roles of student writers in German business writing. In L. Ravelli & R. A. Ellis (Eds.), Analysing academic writing: Contextualized frameworks (pp. 45–65). London: Continuum. Hall, J. K., Cheng, A., & Carlson, M. T. (2006). Reconceptualizing multicompetence as a theory of language knowledge. Applied Linguistics, 27(2), 220–240. https://doi.org/10.1093/applin/aml013 Hallgren, K. A. (2012). Computing inter-rater reliability for observational data: An overview and tutorial. Tutorials in Quantitative Methods for Psychology, 8(1), 23–34. Halliday, M. A. K. (1984). Language as code and language as behavior: A systemic-functional interpretation of the nature and ontogenesis of dialogue. In R. P. Fawcett (Ed.), The semiotics of culture and language: Language as social semiotic (Vol. 1, pp. 3–35). London: Bloomsbury.

303

Halliday, M. A. K. (1985). Introduction to functional grammar. London: Arnold. Halliday, M. A. K. (1993). Towards a language-based theory of learning. Linguistics and Education, 5(2), 93–116. https://doi.org/10.1016/0898-5898(93)90026-7 Halliday, M. A. K. (2002a). Grammar and daily life: Concurrence and complementarity (1998). In J. Webster (Ed.), On grammar (pp. 369–383). London: Continuum. Halliday, M. A. K. (2002b). Spoken and written modes of meaning (1987). In J. Webster (Ed.), On grammar (pp. 323–351). London: Continuum. Halliday, M. A. K., & Matthiessen, C. M. I. M. (2004). An introduction to functional grammar. New York: Routledge. Hamilton, H., & Cohen, A. (2004). Creating a playworld: Motivating learners to take chances in a second language. In J. Frodesen & C. Holten (Eds.), The power of context in language learning and teaching (pp. 237–247). Boston, MA: Thomson Heinle. Hamilton, H. E., Crane, C., & Bartoshesky, A. (2005). Doing foreign language: Bringing Concordia Language Villages into language classrooms. Pearson/Merrill/Prentice Hall. Hanauer, D. I. (2015). Measuring voice in poetry written by second language learners. Written Communication, 32(1), 66–86. https://doi.org/10.1177/0741088314563023 Hao, J., & Humphrey, S. (2012). The role of “coupling” in biological experimental reports. Linguistics and the Human Sciences, 5(2), 169–194. https://doi.org/10.1558/lhs.v5i2.169 Harwood, N. (2005). ‘Nowhere has anyone attempted... In this article I aim to do just that’: A corpus-based study of self-promotional I and we in academic writing across four disciplines. Journal of Pragmatics, 37(8), 1207–1231. https://doi.org/10.1016/j.pragma.2005.01.012 Hashemi, M. R., & Babaii, E. (2013). Mixed methods research: Toward new research designs in applied linguistics methods research. Modern Language Journal, 94(4), 828–852. Heffer, C. (2007). Judgement in court: Evaluating participants in courtroom discourse. Language and the Law: International Outlooks. Lodz Studies in Language Series, 16, 145–179. Helms–Park, R., & Stapleton, P. (2003). Questioning the importance of individualized voice in undergraduate L2 argumentative writing: An empirical study with pedagogical implications. Journal of Second Language Writing, 12(3), 245–265. https://doi.org/10.1016/j.jslw.2003.08.001 Hinkel, E. (2005). Hedging, inflating, and persuading in L2 academic writing. Applied Language Learning, 15(1–2), 29–53. Hirvela, A., & Belcher, D. (2001). Coming back to voice: The multiple voices and identities of mature multilingual writers. Journal of Second Language Writing, 10, 83–106.

304

Hong, J. J.–W. (2012). Lemke, Jay. In C. A. Chapelle (Ed.), The encyclopedia of applied Linguistics. Oxford: Blackwell Publishing Ltd. https://doi.org/10.1002/9781405198431.wbeal0682 Hood, S. (2004). Managing attitude in undergraduate writing: A focus on the introductions to research reports. In Analysing academic writing: Contextualized frameworks (pp. 24–44). London: Continuum. Hood, S. (2008). Texturing interpersonal meanings in academic argument: Pulses and prosodies of value. In G. Forey & G. Thompson (Eds.), Text type and texture (pp. 216–233). London: Equinox. Hood, S. (2010). Appraising research: Evaluation in academic writing. Basingstoke, UK: Palgrave Macmillan. Hood, S., & Martin, J. R. (2007). Invoking attitude: The play of graduation in appraising discourse. In R. Hasan, C. M. I. M. Matthiessen, & J. J. Webster (Eds.), Continuing discourse on language: A functional perspective (pp. 739–764). London: Equinox Publishing Ltd. Hu, C., & Li, X. (2015). Epistemic modality in the argumentative essays of Chinese EFL learners. English Language Teaching, 8(6), 20–31. https://doi.org/10.5539/elt.v8n6p20 Huan, C. (2015). Journalistic engagement patterns and power relations: Corpus evidence from Chinese and Australian hard news reporting. Discourse & Communication, 18(1), 1–20. https://doi.org/10.1177/1750481315611239 Humann, K. (Ed.) (1991). “Denk ich an Deutschland—”: Menschen erzählen von ihren Hoffnungen und Ängsten. Hamburg: Luchterhand Literaturverlag. Hunston, S. (1993). Evaluation and ideology in scientific writing. In M. Ghadessy (Ed.), Register analysis: Theory and practice (pp. 57–71). London: Pinter. Hunston, S., & Thompson, G. (2000). Evaluation in text: Authorial stance and the construction of discourse. Oxford: Oxford University Press. Hyland, K. (2001). Bringing in the reader: Addressee features in academic articles. Written Communication, 18(4), 549–574. Hyland, K. (2005a). Metadiscourse: Exploring interaction in writing. New York: Continuum. Hyland, K. (2005b). Representing readers in writing: Student and expert practices. Linguistics and Education, 16(4), 363–377. https://doi.org/10.1016/j.linged.2006.05.002 Hyland, K. (2005c). Stance and engagement: A model of interaction in academic discourse. Discourse Studies, 7, 173–192. https://doi.org/10.1177/1461445605050365 Hyland, K. (2010). Metadiscourse: Mapping interactions in academic writing. Nordic Journal of English Studies, 9(2), 125–143.

305

Hyland, K. (2012). Undergraduate understandings: Stance and voice in final year reports. In K. Hyland & C. Sancho Guinda (Eds.), Stance and voice in written academic genres (pp. 134– 149). Basingstoke, UK: Palgrave Macmillan. Hyland, K. (2015). Metadiscourse. In K. Tracy, T. Sandel, & C. Ilie (Eds.), The international encyclopedia of language and social interaction. Hyland, K., & Milton, J. (1997). Qualification and certainty in L1 and L2 students’ writing. Journal of Second Language Writing, 6(2), 183–205. Hyland, K., & Tse, P. (2004). Metadiscourse in academic writing: A reappraisal. Applied Linguistics, 25(2), 156–177. https://doi.org/10.1093/applin/25.2.156 Isaac, A. (2012). Modelling voice as Appraisal and Involvement resources: The portrayal of textual identities and interpersonal relationships in the written stylistic analyses of non- native speaker, international undergraduates (Unpublished doctoral dissertation). University of Canberra, Canberra. Ivanič, R., & Camps, D. (2001). I am how I sound: Voice as self-representation in L2 writing. Journal of Second Language Writing, 10(1–2), 3–33. https://doi.org/10.1016/S1060- 3743(01)00034-0 Ivanič, R., & Simpson, J. (1992). Who’s who in academic writing? In N. Fairclough (Ed.), Critical language awareness (pp. 141–173). London: Longman. Jakaza, E. (2013). Appraisal and evaluation in Zimbabwean parliamentary discourse and its representation in newspaper articles (Unpublished doctoral dissertation). Stellenbosch University, Stellenbosch, South Africa. Jeffery, J. V. (2011). Subjectivity, intentionality, and manufactured moves: Teachers’ perceptions of voice in the evaluation of secondary students’ writing. Research in the Teaching of English, 92–127. Kern, R. (1995). Redefining the boundaries of foreign language literacy. In C. Kramsch (Ed.), Redefining the boundaries of language study (pp. 61–98). Boston, MA: Heinle & Heinle. Kern, R. (2002). Reconciling the language–literature split through literacy. ADFL Bulletin, 33(3), 20–24. Kim, E. Y. A., & Brown, L. (2014). Negotiating pragmatic competence in computer mediated communication: The case of Korean address terms. CALICO Journal, 31(3), 264–284. https://doi.org/10.11139/cj.31.3.264-284 Kuhn, K. (in press). A Systemic Functional Linguistics approach to instructional design for L2 German: Connecting language and content. Die Unterrichtspraxis/Teaching German. Kuo, C. (1999). The use of personal pronouns: Role relationships in scientific journal articles. English for Specific Purposes, 18(2), 121–138. https://doi.org/10.1016/S0889- 4906(97)00058-6

306

Labov, W. (1972). Language in the inner city: Studies in the black English vernacular. Philadelphia: University of Pennsylvania Press. Labov, W. (2008). Oral narratives of personal experience. In P. C. Hogan (Ed.), Cambridge encyclopedia of the language sciences (pp. 546–548). Cambridge: Cambridge University Press. Labov, W., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal experience. In J. Helm (Ed.), Essays in the verbal and visual arts (pp. 12–44). Washington, DC: University of Washington Press. Lakoff, G. (2002). Moral politics. Chicago: University of Chicago Press. Lancaster, Z. (2011). Interpersonal stance in L1 and L2 students’ argumentative writing in economics: Implications for faculty development in WAC/WID programs. Across the Disciplines, 8(4), 1–24. Retrieved from http://wac.colostate.edu/atd/ell/lancaster.cfm Lancaster, C. I. (2012). Stance and reader positioning in upper-level student writing in political theory and economics (Unpublished doctoral dissertation). University of Michigan, Ann Arbor, MI. Lancaster, Z. (2013). Tracking interpersonal style: The use of functional language analysis in college writing instruction. In M. Duncan & S. M. Vanguri (Eds.), The centrality of style (pp. 199–220). Parlor Press. Lancaster, Z. (2014). Exploring valued patterns of stance in upper-level student writing in the disciplines. Written Communication, 31(1), 27–57. https://doi.org/10.1177/0741088313515170 Lancaster, Z. (2016). Expressing stance in undergraduate writing: Discipline-specific and general qualities. Journal of English for Academic Purposes, 23, 16–30. https://doi.org/10.1016/j.jeap.2016.05.006

Lee, S. H. (2007). An application of multiple coding for the analysis of ATTITUDE in an academic argument. Linguistics & the Human Sciences, 3(2), 165–190. https://doi.org/10.1558/lhs.v3i2.165 Lee, S. H. (2015). Evaluative stances in persuasive essays by undergraduate students: Focusing on APPRECIATION resources. Text and Talk, 35(1), 49–76. https://doi.org/10.1515/text-2014- 0029 Lemke, J. L. (1989). Semantics and social values. Word, 40(1–2), 37–50. https://doi.org/10.1080/00437956.1989.11435795 Lemke, J. L. (1992). Interpersonal meaning in discourse: Value orientations. In M. Davies & L. Ravelli (Eds.), Advances in systemic linguistics (pp. 82–104). Lonton: Pinter. Liu, F. (2017). Strategies for affiliation in media editorials: Persuading and aligning readers (Unpublished doctoral dissertation). University of Technology Sydney, Sydney.

307

Liu, S. (2014). Engaging with dialogic alternatives in ESL argumentative essays: Systemic Functional Linguistic and teacher perspectives (PhD Thesis). McGill University, Montréal, QC. Liu, X. (2013). Evaluation in Chinese university EFL students’ English argumentative writing: An APPRAISAL study. Electronic Journal of Foreign Language Teaching, 10(1), 40–53. Liu, Y., & Zhou, H. (2014). Reporting and stance in second language academic writing. Chinese Journal of Applied Linguistics, 37(4), 483–497. https://doi.org/10.1515/cjal-2014-0029

Macken–Horarik, M. (2003a). APPRAISAL and the special instructiveness of narrative. Text, 23(2), 285–312. https://doi.org/10.1515/text.2003.012 Macken–Horarik, M. (2003b). Envoi: Intractable issues in appraisal analysis? Text, 23(2), 313– 319. Macken–Horarik, M., & Isaac, A. (2014). Appraising Appraisal. In G. Thompson & L. Alba– Juez (Eds.), Evaluation in context (pp. 67–93). Philadelphia/Amsterdam: John Benjamins. Macken–Horarik, M., Love, K., Sandiford, C., & Unsworth, L. (2018). Functional grammatics. Abingdon, UK: Routledge. Macken–Horarik, M., & Morgan, W. (2011). Towards a metalanguage adequate to linguistic achievement in post-structuralism and English: Reflections on voicing in the writing of secondary students. Linguistics and Education, 22(2), 133–149. https://doi.org/10.1016/j.linged.2010.11.003 Martin, J. R. (1984). Language, register and genre. In F. Christie (Ed.), Children writing: Reader (pp. 21–29). Geelong, Victoria, Australia: Deakin University Press. Martin, J. R. (1997). Analyzing genre: Functional parameters. In F. Christie & J. R. Martin (Eds.), Genre and institutions: Social processes in the workplace and school (pp. 3–39). London: Continuum. https://doi.org/10.1007/s00268-012-1581-5 Martin, J. R. (2004). Mourning: How we get aligned. Discourse & Society, 15(2–3), 321–344. Martin, J. R. (2014). Evolving systemic functional linguistics: Beyond the clause. Functional Linguistics, 1(3). https://doi.org/10.1186/2196-419X-1-3 Martin, J. R., & Rose, D. (2003). Working with discourse: Meaning beyond the clause. London/New York: Continuum. Martin, J. R., & White, P. R. R. (2005). The language of evaluation: Appraisal in English. Basingstoke, UK: Palgrave Macmillan. Matsuda, P. K. (2001). Voice in Japanese written discourse: Implications for second language writing. Journal of Second Language Writing, 10(1–2), 35–53. https://doi.org/10.1016/S1060-3743(00)00036-9

308

Matsuda, P. K. (2015). Identity in written discourse. Annual Review of Applied Linguistics, 35(2015), 140–159. https://doi.org/10.1017/S0267190514000178 Matsuda, P. K., & Tardy, C. M. (2007). Voice in academic writing: The rhetorical construction of author identity in blind manuscript review. English for Academic Purposes, 26, 235–249. https://doi.org/10.1016/j.esp.2006.10.001 Matsuda, P. K., & Tardy, C. M. (2008). Continuing the conversation on voice in academic writing. English for Specific Purposes, 27(1), 100–105. https://doi.org/10.1016/j.esp.2007.04.002 Mei, W. S. (2007). The use of engagement resources in high- and low-rated undergraduate geography essays. Journal of English for Academic Purposes, 6, 254–271. https://doi.org/10.1016/j.jeap.2007.09.006

Miller, D. R. (2004). ‘Truth, justice and the American way’: The APPRAISAL system of JUDGEMENT in the U.S. House debate on the impeachment of the President, 1998. In P. Bayley (Ed.), Cross-cultural perspectives on parliamentary discourse (pp. 271–300). Philadelphia/Amsterdam: John Benjamins. Mirallas, C. A. (2017). The writing of abstracts and SRAs by EFL students: Analysis of productions and assessment of SFL genre pedagogy (Unpublished MA thesis). Universidad Nacional de Córdoba, Córdoba, Spain. Mohan, B., & Slater, T. (2004). The evaluation of casual discourse and language as a resource for meaning. In J. Foley (Ed.), Language, education and discourse: Functional approaches (pp. 257–269). London: A&C Black. Morton, T., & Llinares, A. (2018). Students’ use of evaluative language in L2 English to talk and write about history in a bilingual education programme. International Journal of Bilingual Education and Bilingualism, 21(4), 496–508. https://doi.org/10.1080/13670050.2016.1192101 Mozgalina, A., & Ryshina–Pankova, M. V. (2015). Meeting the challenges of curriculum construction and change: Revision and validity evaluation of a placement test. Modern Language Journal, 99(2), 346–370. https://doi.org/10.1111/modl.12217 Multon, K. D. (2010). Interrater reliability. In N. J. Salkind (Ed.), Encyclopedia of research design (pp. 626–628). Thousand Oaks, CA: SAGE Publications. Myskow, G., & Ono, M. (2018). A matter of facts: L2 writers’ use of evidence and evaluation in biographical essays. Journal of Second Language Writing, 41, 55–70. https://doi.org/10.1016/j.jslw.2018.08.002 The New London Group (1996). A Pedagogy of multiliteracies: Designing social futures. Harvard Educational Review, 66(1), 60–93.

309

Ngo, T. T. B., & Unsworth, L. (2011). Vietnamese person reference system as an appraisal resource. In Q. Lê & T. Le (Eds.), Linguistic diversity and cultural identity: A global perspective (pp. 169–188). Hauppauge, NY: Nova Science Publishers. Ngo, T. T. B., & Unsworth, L. (2015). Reworking the appraisal framework in ESL research: Refining attitude resources. Functional Linguistics, 2(1), 1–24. https://doi.org/10.1186/s40554-015-0013-x O’Donnell, M. (2008). The UAM CorpusTool: Software for corpus annotation and exploration. In C. M. Bretones Callejas et al. (Eds.), Applied linguistics now: Understanding language and mind / La lingüística aplicada hoy: Comprendiendo el lenguaje y la mente (pp. 1433– 1447). Almería, Spain: Universidad de Almería. O’Donnell, M. (2012). Appraisal analysis and the computer. Revista Canaria de Estudios Ingleses, 65, 115–130. O’Donnell, M. (2014). Exploring identity through Appraisal analysis: A corpus annotation methodology. Linguistics and the Human Sciences, 9(1), 95–116. https://doi.org/10.1558/lhs.v9i1.95 Okimoto, T. G., & Brescoll, V. L. (2010). The price of power: Power seeking and backlash against female politicians. Personality and Social Psychology Bulletin, 36(7), 923–936. Ortega, L. (2015). Syntactic complexity in L2 writing: Progress and expansion. Journal of Second Language Writing, 29, 82–94. https://doi.org/10.1016/j.jslw.2015.06.008 Oteíza, T. (2017). The appraisal framework and discourse analysis. In T. Bartlett & G. O’Grady (Eds.), The Routledge handbook of systemic functional linguistics (pp. 457–472). London: Routledge. Paesani, K. (2016). Investigating connections among reading, writing, and language development: A multiliteracies perspective. Reading in a Foreign Language, 28(2), 266– 289.

Page, R. E. (2003). An analysis of APPRAISAL in childbirth narratives with special consideration of gender and storytelling style. Text, 23(2), 211–237. https://doi.org/10.1515/text.2003.009 Partington, A. (2017). Evaluative clash, evaluative cohesion and how we actually read evaluation in texts. Journal of Pragmatics, 117, 190–203. https://doi.org/10.1016/j.pragma.2017.06.008 Patja Howell, J. (2019, November 23). The week that was on impeachment [audio podcast]. Retrieved from https://www.lawfareblog.com/lawfare-podcast-week-was-impeachment Pessoa, S. (2017). How SFL and explicit language instruction can enhance the teaching of argumentation in the disciplines. Journal of Second Language Writing, 36(May), 77–78. https://doi.org/10.1016/j.jslw.2017.05.004

310

Polio, C. (2017). Thinking allowed. Second language writing development: A research agenda. Language Teaching, 50(2), 261–275. https://doi.org/10.1017/S0261444817000015 Polletta, F., & Callahan, J. (2017). Deep stories, nostalgia narratives, and fake news: Storytelling in the Trump era. American Journal of Cultural Sociology, 5(3), 392–408. https://doi.org/10.1057/s41290-017-0037-7 Poynton, C. (1989). Language and gender: Making the difference. Oxford: Oxford University Press. Põldvere, N., Fuoli, M., & Paradis, C. (2016). A study of dialogic expansion and contraction in spoken discourse using corpus and experimental techniques. Corpora, 11(2), 191–225. https://doi.org/10.3366/cor.2016.0092 Quam, J., & Hamilton, H. (2020). Authentic signs as portals to cultural practices within a residential German language immersion program: From part of the scenery to curricular resources. In G. Niedt & C. Seals (Eds.), Linguistic landscapes beyond the language classroom. London: Bloomsbury. Rothery, J. (1996). Making changes: Developing an educational linguistics. In R. Hasan and G. Williams (Eds.), Literacy in society (p. 86–123). New York: Longman, 1996. Rothery, J., & Stenglin, M. (1997). Entertaining and instructing: Exploring experience through story. In F. Christie & J. R. Martin (Eds.), Genre and institutions: Social processes in the workplace and school (pp. 231–263). London: Continuum.

Rothery, J., & Stenglin, M. (2000). Interpreting literature: The role of APPRAISAL. In L. Unsworth (Ed.), Researching language in schools and functional linguistic perspectives (pp. 222–244). London: Cassell. Rowling, J. K. (1998). Harry Potter and the sorcerer’s stone. New York: Scholastic. Rowling, J. K. (2007). Harry Potter and the deathly hallows. New York: Scholastic. Ryshina–Pankova, M. V. (2006). Constructing coherent and cohesive textual worlds in advanced foreign language learner writing (Unpublished doctoral dissertation). Georgetown University, Washington, DC. Ryshina–Pankova, M. V. (2010). Toward mastering the discourses of reasoning: Use of grammatical metaphor at advanced levels of foreign language acquisition. Modern Language Journal, 94(2), 181–197. https://doi.org/10.1111/j.1540-4781.2010.01016.x Ryshina–Pankova, M. V. (2011). Developmental changes in the use of interactional resources: Persuading the reader in FL book reviews. Journal of Second Language Writing, 20(4), 243–256. https://doi.org/10.1016/j.jslw.2011.05.011 Ryshina–Pankova, M. V. (2015a). A meaning-based approach to the study of complexity in L2 writing: The case of grammatical metaphor. Journal of Second Language Writing, 29, 51– 63. https://doi.org/10.1016/j.jslw.2015.06.005

311

Ryshina–Pankova, M. V. (2015b). Foreign language curriculum as a means of achieving humanities learning goals: Assessment of materials, pedagogy, and learner texts. In J. M. Norris & J. McE. Davis (Eds.), Student learning outcomes assessment in college foreign language programs (pp. 221–246). Honolulu, HI: National Foreign Language Resource Center at University of Hawaiʻi. Ryshina–Pankova, M. V. (2016). Scaffolding advanced literacy in the foreign language classroom: Implementing a genre-driven content-based approach. In L. Cammarata (Ed.), Content-based foreign language teaching: Curriculum and pedagogy for developing advanced thinking and literacy skills (pp. 51–76). New York: Routledge. Ryshina–Pankova, M. V. (2021). Memories of childhood: The genre of anecdote in German. In F. J. Troyan (Ed.), Genre in world language education: Contextualized assessment and learning. New York: Routledge. Ryshina–Pankova, M. V., & Byrnes, H. (2013). Writing as learning to know: Tracing knowledge construction in L2 German compositions. Journal of Second Language Writing, 22(2), 179– 197. https://doi.org/10.1016/j.jslw.2013.03.009 Schleppegrell, M. (2005). Technical writing in a second language: The role of grammatical metaphor. In L. J. Ravelli & R. A. Ellis (Eds.), Analysing academic writing: Contextualized frameworks (pp. 172–189). London: Continuum. Searle, J. R. (1965). What is a speech act? In M. Black (Ed.), Philosophy in America (pp. 221– 239). London: Allen & Unwin. Searle, J. R. (1969). Speech acts: An essay in the philosophy of language. Cambridge: Cambridge University Press. Shenhav, S. R. (2006). Political narratives and political reality. International Political Science Review, 27(3), 245–262. https://doi.org/10.1177/0192512106064474 Simon–Vandenbergen, A., White, P. R. R., & Aijmer, K. (2007). Presupposition and ‘taking-for- granted’ in mass communicated political argument: An illustration from British, Flemish and Swedish political colloquy. In A. Fetzer & G. E. Lauerbach (Eds.), Political discourse in the media: Cross-cultural perspectives (pp. 31–74). Philadelphia/Amsterdam: John Benjamins. Slobin, D. I. (1996). Two ways to travel: Verbs of motion in English and Spanish. In M. Shibatani & S. Thompson (Eds.), Grammatical constructions (pp. 195–219). Oxford: Clarendon Press. Stapleton, P. (2002). Critiquing voice as a viable pedagogical tool in L2 writing: Returning the spotlight to ideas. Journal of Second Language Writing, 11(3), 177–190. https://doi.org/10.1016/S1060-3743(02)00070-X

312

Stapleton, P., & Helms–Park, R. (2008). A response to Matsuda and Tardy’s “Voice in academic writing: The rhetorical construction of author identity in blind manuscript review.” English for Specific Purposes, 27(1), 94–99. https://doi.org/10.1016/j.esp.2007.04.001 Stillman, S. (2016, March 14). The List. The New Yorker. Retrieved from https://www.newyorker.com/magazine/2016/03/14/when-kids-are-accused-of-sex-crimes Stock, I., & Eik–Nes, N. L. (2016). Voice features in academic texts: A review of empirical studies. Journal of English for Academic Purposes, 24, 89–99. https://doi.org/10.1016/j.jeap.2015.12.006 Strunk, W., & White, E. B. (1959). The elements of style. New York: Macmillan. Stubbs, M. (1986). ‘A matter of prolonged field work’: Notes towards a modal grammar of English. Applied Linguistics, 7(1), 1–25. https://doi.org/10.1093/applin/7.1.1 Swaffar, J., Arens, K., & Byrnes, H. (1991). A procedural model for integrative reading. In Reading for meaning: An integrated approach to language learning (pp. 73–232). Upper Saddle River, NJ: Prentice Hall. Swain, E. (2007). Constructing an effective “voice” in academic discussion writing: An appraisal theory perspective. In A. McCabe, M. O’Donnell, & R. Whittaker (Eds.), Advances in language and education (pp. 166–184). London: Continuum. Swain, E. (2012). Analysing evaluation in political cartoons. Discourse, Context & Media, 1(2– 3), 82–94. https://doi.org/10.1016/j.dcm.2012.09.002 Swain, E. (2013). Choice in a ’partly free’ press: Evaluative key in Italian journalistic discourse. In G. O’Grady, T. Bartlett, & L. Fontaine (Eds.), Choice in language: Applications in text analysis (pp. 55–85). Sheffield, UK: Equinox Publishing Ltd. Tardy, C. M. (2016). Voice and identity. In R. M. Manchón & P. K. Matsuda (Eds.), Handbook of second and foreign language writing (pp. 349–363). Boston: de Gruyer Mouton. Tashakkori, A., & Teddlie, C. (2008). Advances in mixed methods research: Theories and applications. In M. M. Bergman (Ed.), Quality of inferences in mixed methods research: Calling for an integrative approach (pp. 101–119). Thousand Oaks, CA: SAGE Publications. Taverniers, M. (2008). Interpersonal grammatical metaphor as double scoping and double grounding. Word, 59(1–2), 83–109. https://doi.org/10.1080/00437956.2008.11432582 Taylor, J. M. (2003). Constructing right-wing extremism in the media: The role of evaluation in three news genres (Unpublished doctoral dissertation). Georgetown University, Washington, DC. Thaler, R. H., & Sunstein, C. R. (2003). Libertarian paternalism. American Economic Review, 93(2), 175–179.

313

Thomas, D. P., Thomas, A. A., & Moltow, D. T. (2015). Evaluative stance in high achieving Year 3 persuasive texts. Linguistics and Education, 30, 26–41. https://doi.org/10.1016/j.linged.2015.03.003 Thompson, G. (2001). Interaction in academic writing: Learning to argue with the reader. Applied Linguistics, 22(1), 58–78. https://doi.org/10.1093/applin/22.1.58 Thompson, G. (2013). Picking an argument: Politicians’ choice of persuasive strategies. In L. Fontaine, T. Bartlett, & G. O’Grady (Eds.), Systemic functional linguistics: Exploring choice (pp. 226–246). Cambridge: Cambridge University Press. Thompson, G. (2014a). Affect and emotion, target-value mismatches, and Russian dolls: Refining the appraisal model. In G. Thompson & L. Alba–Juez (Eds.), Evaluation in context (pp. 47–66). Philadelphia/Amsterdam: John Benjamins. Thompson, G. (2014b). Introducing functional grammar. New York: Routledge. Thompson, G., & Alba–Juez, L. (2014). Evaluation in context. Philadelphia/Amsterdam: John Benjamins. Thompson, G., & Thetela, P. (1995). The sound of one hand clapping: The management of interaction in written discourse. Text, 15(1), 103–128. https://doi.org/10.1515/text.1.1995.15.1.103 Tilakaratna, N. L., & Mahboob, A. (2013). Appraisal in the time of conflict: Coding evaluation through textual and contextual analysis. Linguistics and the Human Sciences, 8(1), 63–90. https://doi.org/10.1558/lhs.v8i1.63 Toolan, M. (2001). Narrative: A critical linguistic introduction (2nd ed.). Abingdon, UK: Routledge. Troyan, F. J. (2016). Learning to mean in Spanish writing: A case study of a genre-based pedagogy for standards-based writing instruction. Foreign Language Annals, 49(2), 317– 335. https://doi.org/10.1111/flan.12192 Urlaub, P. (2011). Developing literary reading skills through creative writing in German as a second language. Die Unterrichtspraxis/Teaching German, 44(2), 98–105. Vian, O. (2008). Appraisal system in Brazilian Portuguese: Resources for graduation. Odense Working Papers in Language and Communication, 29, 808–824. Warren, M., & Winkler, C. (2016). Developing multiliteracies through genre in the beginner German classroom. In Y. Kumagai, A. López–Sánchez, & S. Wu (Eds.), Multiliteracies in world language education (pp. 29–57). New York: Routledge. Westen, D. (2008). The political brain: The role of emotion in deciding the fate of the nation. New York: PublicAffairs.

314

White, P. R. R. (2003). News as history: Your daily gossip. Re/Reading the Past: Critical and Functional Perspectives on Time and Value, 8, 61–89. White, P. R. R. (2008). Modality as dialogue: A Bakhtinian reanalysis of epistemic stance. Word, 59(2), 143–167. White, P. R. R. (2015). Appraisal and the resources of intersubjective stance. Engagement: An overview. Retrieved July 4, 2020 at https://www.grammatics.com/appraisal/ Wigboldus, D., Spears, R., & Semin, G. (1999). Categorization, content and the context of communicative behaviour. In N. Ellemers, R. Spears, and B. Doosje (Eds.,), Social identity (pp. 147–163). Oxford/Malden: Blackwell. Yang, Y. (2013). A corpus-based study of interpersonal Grammatical Metaphor in spoken Chinese. Language Sciences, 38, 1–21. https://doi.org/10.1016/j.langsci.2012.12.003 Yasuda, S. (2017). Toward a framework for linking linguistic knowledge and writing expertise: Interplay between SFL-based genre pedagogy and task-based language teaching. TESOL Quarterly, 51(3), 576–606. https://doi.org/10.1002/tesq.383 Yoon, H. (2017). Textual voice elements and voice strength in EFL argumentative writing. Assessing Writing, 32, 72–84. https://doi.org/10.1016/j.asw.2017.02.002 Zappavigna, M. (2011). Ambient affiliation: A linguistic perspective on Twitter. New Media & Society, 13(5), 788–806. https://doi.org/10.1177/1461444810385097 Zappavigna, M., & Martin, J. R. (2018). Expressing feeling: Appraisal systems. In Discourse and diversionary justice: An analysis of youth justice conferencing (pp. 157–197). Basingstoke, UK: Palgrave Macmillan. Zappavigna, M., Cléirigh, C., Dwyer, P., & Martin, J. R. (2010). Visualizing appraisal prosody. In A. Mahboob & N. Knight (Eds.), Appliable linguistics: Texts, contexts, and meanings (pp. 150–167). London/New York: Continuum. Zipes, J. (1979). Breaking the magic spell: Radical theories of folk and fairy tales. New York: Routledge.

315