1 A Mixed-Methods Approach to the Aboutness of Writing Center Talk
To begin, here is a brief exercise in estimation.
The Writing Center Directory lists over 1720 writing centers in the United States alone. The list includes entries for writing centers in universities, colleges, seminaries, community colleges, technical colleges, and military institutes. It also includes some entries for writing centers in high schools and junior high and middle schools, as well as entries for community writing centers. In addition, the directory includes entries for over 360 more writing centers around the world, including writing centers in Australia, Columbia, Estonia, France, India, Japan, Kuwait, Lebanon, Mongolia, and Vietnam (Writing Center Directory 2015). If each of the 2,080 writing centers listed in the directory employs two tutors each hour of an eight-hour day for five days each week, each center generates 80 hours of talk about writing every week. Each week, then, all the writing centers listed in the directory together generate about 166,400 hours of talk about writing every week. And that means each academic year, fall and spring semesters, these writing centers together generate 4,999,200 hoursânearly 5 million hoursâof talk about writing. Thatâs over 570 years of conference talk each academic year.
Even this ballpark (and rather conservative) estimate shows the amount of talk that tutors and student writers produce is vast. But hereâs a question to ponder: What is all of this writing center talk about?
Until now, no one has addressed this questionâa question that gets at what makes writing center talk different from other kinds of talk and thus constructs its unique efficacy. In this book, The Aboutness of Writing Center Talk: A Corpus-Driven and Discourse Analysis (Aboutness), I report a detailed answer to this question. In addition, I argue for and use a mixed-methods approach to empirically examine talk between student writers and tutors in order to describe and analyze the aboutness of writing center discourse (GoĆșdĆș-Roszkowski 2011; Phillips 1989)âthe characteristics that constitute the content of one-to-one talk in writing centers. Then, I examine those linguistic features in context to understand how tutors and student writers used them. The mixed-methods approach that I describe and employ here combines corpus-driven analysis and discourse analysis to provide a rich, micro- and macrolevel view of writing center talk. In short, I have two main goals in this book: (1) to analyze the aboutness of writing center talk and (2) to show how the two research methods operate together to produce a robust and rigorous analysis of spoken discourse. To accomplish the first goal, I analyzed a specialized corpus of 47 writing center conferences to identify and examine words and sequences of words that revealed what writing center discourse is about. Then, I examined tutorsâ and student writersâ key words and word sequences in context to understand how they functioned in the task-oriented talk that tutors and student writers co-constructed. To accomplish the second goal, I explain the theory and research results that underlie the mixed-methods approach and reveal its benefits.
Aboutness
As the term implies, âaboutnessâ refers to the content of a text or a large collection of textsâa corpus. The concept informs a range of disciplines, particularly information and library science (Hutchins 1978; Woolwine, Ferguson, Joly, Pickup, and Udma 2011). Recently, research on blogs and other web content has explored the extent to which social tagging can capture the aboutness of those online texts (Kehoe and Gee 2011, 2012). In linguistic terms, aboutness stems from lexical choices and patterns and the meanings that they create. Indeed, in relation to meaning, the Oxford English Dictionary cites Joachimâs (1906) philosophical treatise on the coherence theory of truth as the first use of the term âaboutnessâ (Rondeau 2014). The linguistic choices that determine aboutness fluctuate according to contextâwhy and where the discourse occurs. For example, the aboutness of a corpus of face-to-face conversations among friends will differ from a corpus of operating-room discourse among doctors and nurses. The aboutness of the corpus of operating-room discourse would consist of more medical terminology than the corpus of everyday conversation. Similarly, the aboutness of writing center talk would differ from the aboutness of operating-room discourse and from everyday conversation.
Understanding the aboutness of a text or corpus (such as a corpus of writing center transcripts) by analyzing âlarge scale regularities,â says Phillips (1989), can help reveal a listenerâs or readerâs âpsychological perception of subject matterâ (6â7, see also GoĆșdĆș-Roszkowski 2011). In the field of writing center studies, analyzing aboutness to understand the linguistic features and patterns that comprise writing center talk can provide insight into what tutors and student writers discussâand what they do not. It also can provide insight into how they interact. For example, analyzing aboutness can provide insight into the stance, the attitude, or the level of certainty (Biber, Conrad, and Cortes 2004, 384), that tutors and student writers express. In contrastive studies, analyzing the aboutness of writing center talk can reveal how it differs from other types of instructional discourse and thus illuminate the linguistic characteristics that make it efficacious for some purposes and not others.
Determining aboutness involves using quantitative measures, identifying, for example, frequently occurring sequences of words. However, to expand the focus from the tight focus on the microlevel of particular words and word sequences to a broader macrolevel, many researchers supplement the quantitative analysis with qualitative analysis, particularly discourse analysis. In Aboutness, I followed a mixed-methods approach to analyzing writing center talk.
Mixed-Methods Analysis
As noted above, I used two analytical methods to examine a specialized corpus of writing center talk. First, I used corpus analysis, the quantitative method, to describe the aboutness of the conferences. It identified the frequently occurring and key words and the most frequently occurring word sequences, called lexical bundles. A corpus is a collection of representative samples of language from a particular situation, for example, writing center conferences (Biber and Conrad 2009; Biber Conrad, and Reppen 1998; OâKeefee, McCarthy, and Carter 2007). The intention is for the corpus to represent the language situation (for example, all of the talk in writing centers located in the United States) so that the researcher can then generalize from those findings. Hence, a corpus must include the range of linguistic variation that occurs in the language situation and, as a result, must be quite large. The present studyâs corpus, a specialized corpus of transcripts from 47 writing center conferences, contained 157,665 wordsânot large enough to make generalizations but large enough to understand the specialized corpusâs aboutness.
As evident from its focus on single words and lexical bundles, corpus analysis reveals a corpusâs microstructure. With Anthonyâs (2014a) AntConc 3.4.3 concordance application, I compiled tutorsâ and student writersâ most frequently occurring words, type/token ratios, key words (words occurring statistically more frequently in a study corpus than in a reference corpus) and words collocating with (occurring in the environment of) writing-related key words, and frequently occurring four-word lexical bundles. After using corpus analysis to identify these linguistic features, I analyzed those features in context using discourse analysis, the studyâs qualitative method. Discourse analysis of spoken language focuses on the macrolevel connections to identify how speakers co-construct their interaction on a moment-to-moment basis. Discourse analysis revealed how tutors and student writers used the aboutness-revealing linguistic features.
An Example
To get a sense of how a mixed-methods approach operates and how it differs from the kinds of research on writing center talk that have come before, I present below a short excerpt of talk from the opening stage (see Mackiewicz and Thompson 2015, 63â65) of a conference between a writing center tutor (T47) and a student writer (S47). During this opening stage of the conference, T47 began by reading aloud segments of the assignment sheet that delineated guiding questions. Then, T47 and S47 discussed a potential thesis statement that would make a claim about the relationship between two main characters from Alice Walkerâs book The Color Purple:
T47: I guess with all of these. Ok. âHow does Celie view herself before Shug- Shugâs effect on Celie? And then what is it about Shug?â Ok. So first, we need to come up with a topic sentence.
T47: Um. Yeah. A thesis statement.
S47: Yeah. Just write it down there.
T47: What in general- [3 seconds] In general, like, the most general quest- um, statement you can come up with. Ha. What general effect does Celie- I mean, does Shug have on Celie?
S47: Um. [2 seconds] She makes her feel like a- She makes her feel like a person. Well, she makes her feel- I canât explain it. Um. [3 seconds]
S47: Yeah. She makes her feel alive. So-
T47: Makes her feel alive.
T47: Makes her feel important.
T47: Um. So I think we should start off at that idea. Um. And say something to the effect that- [2 seconds] âCelie- Before Shug wasâ A B C. And then âafter Shug,â you know, âshe felt important. She felt alive.â Does that make sense?
S47: Yeah. We pretty much got to write- Um. Tell, um, how she makes her feel from beginning to end.
Writing center researchers could take a variety of approaches to analyzing the excerpted writing center talk above. Some researchers might focus on the different roles that T47 enacted even during this brief exchange. They might note how T47, through her questions and advice, moved from enacting an instructor role to enacting a peer or collaborator role before returning to an instructor role with her suggestion for potential words (âbeforeâ and âafterâ) to structure a possible thesis statement. Other writing center researchers might home in on the ways that demographic variables such as sex (T47 is female, S47 is male), race (T47 and S47 are African-American), and age (T47 is late twenties, S47 is late teens) affected the interaction. They might also explore how those variables correlated with the conferenceâs outcomes, for example, the quality of the student writerâs paper and the result in terms of participantsâ satisfaction with the conference. These possible analyses focus on characterizing the conference participants, possibly in terms of the conference outcomes, whereas the present studyâs corpus analysis focused on characterizing the content of what tutors and student writers said and how that content distinguished the writing center talk. The discourse analysis that follows the corpus-driven analysis allows more focus on individuals, even though it too is concerned with language in that it focuses on how the individuals used the aboutness-revealing words and bundles of words.
Because one short excerpt of writing center talk on its own will say little concerning the aboutness of writing center talk, I explain how I carried out the corpus-driven analysis and mention some of the overall findings from the analysis of the entire 47-conference corpus that this excerpt illustrates. I used this procedure of identifying aboutness-revealing words and bundles and then showing how tutors and student writers used them to co-construct their task-oriented talk throughout the book.
I began the corpus-driven analysis by considering basic measures (chapter 4). I analyzed tutorsâ and student writersâ participation in conferences through their respective word counts. I also determined their most frequently occurring words and, particularly important in analyzing tutorsâ talk, their type/token ratios. Type/token ratios relate the proportion of unique word types to overall word (token) count. Type/token ratio is a rough measure of lexical variation and provides one way to gauge the difficulty of the vocabulary in spoken or written texts. Tutorsâ average type/token ratios, then, gave a rough sense of the difficulty student writers, particularly English-language learners, might have in understanding what tutors say. In addition, using Anthonyâs (2014b) AntWordProfiler, I determined the percentage of tutorsâ and student writersâ words in Westâs (1953) General Service List (GSL) and Coxheadâs (2000) Academic Word List (AWL) in order to get a rough sense of the lexical difficulty and thus comprehensibility of their talk.
The talk excerpted above illustrates a common finding in writing center research and in the present study: the tutor talked moreâhad greater volubilityâthan the student writer did. Therefore, the tutorâs speech contained more tokens than the student writerâs speech. The excerpt also shows the student writer responding to the tutor more than he initiated topics. This respondent role led to less volubility. In addition, the excerpt contains several words among the most frequently occurring words in the present studyâs corpus, including the function words (reference words) âIâ and âyouâ and (hesitation marker) âum.â
Although using basic measures such as word counts and word frequencies offers a sense of the content of tutorsâ and student writersâ talk and the differences between tutorsâ talk and student writersâ talk, these basic measures are not the best indicators of the aboutness of writing center talk; key words are a much better indicator of aboutness, and I examine the key words in the writing center corpus in chapters 5 and 6. Even though they may not be the most frequently occurring words in a particular corpus, key words occur statistically more frequently in comparison to their occurrence in another corpus (or corpora), often a corpus that represents spoken English as people use it more generally. As I discuss in chapter 3, I used subsections of three large corpora, the Manually Annotated Sub-Corpus (MASC), the Corpus of Contemporary American English (COCA), and the Michigan Corpus of Annotated Spoken English (MICASE) as the reference corpora. The excerpted talk above illustrates some key words emerging from the writing center corpus, including minimal responses âyeah,â âok,â and âuhhuhâ and the writing-related key words âthesisâ and âsentence.â In addition, rather than looking solely at the key words, I also analyzed the immediate linguistic environments of the writing-related key words, identifying the words that collocated (co-located) with them statistically frequently. For example, the word âstatementâ strongly collocated with the word âthesis.â In the talk excerpted above, the two words appear together (âUm. Yeah. A thesis statementâ).
Key words are not the only means of revealing the aboutness of a given corpus; analyzing frequently occurring word sequences, lexical bundles, also facilitates the process. A lexical bundle occurred in the talk excerpted above. T47âs question, âDoes that make sense?â was the fourth most frequent lexical bundle in tutorsâ talk. This lexical bundle, one of three in tutorsâ talk that were syntactically complete sentences, marked tutorsâ use of a motivational scaffolding strategy (see Mackiewicz and Thompson 2015), specifically, showing concern for student writers by checking on their u...