Analysing Learner Language

Language As Form and Language As Formative

Long, long ago, when one could turn out a single volume on a second language acquisition topic and still claim to have created a definitive work, I remember a well-known professor brandishing a copy of the recently published and then much admired Second Language Classrooms by Craig Chaudron and complaining that it was a ?bad book?. Although it included almost all of the studies relevant to what we were talking about, he said, it simply piled them up like action sequences in a Hollywood blockbuster, and by the end of each chapter one seems to have lost the plot. Things have only gotten worse, both in Hollywood and in applied linguistics. Indeed, if it were not for the prolific writing of Rod Ellis, not only the plot but most of the actual studies might escape us. The price that Ellis? readers have traditionally had to pay for the encyclopedic scope of his volumes is that the chapters may occasionally appear to have the baggy shape of the kind of three paragraph essays I once had to teach Chinese university sophomores:
?X has many advantages? (examples in no particular order)
?However, X also has some disadvantages? (more examples, more disorder) ?Therefore, further study of X is necessary?
where ?X? might be television, the internet, or a socio-cultural approach to task-based instructed second language acquisition.

Mercifully, this book is not like that. Instead, Ellis and Barkhuizen begin with the observation that a scientific approach to second language acquisition has to reconcile two contradictory phenomena: apparently all-inclusive universal stages, and apparently endless individual variation. With this observation, the chapters begin to fall into place: error analysis, obligatory occasion analysis, frequency analysis, and measures of accuracy, fluency, and complexity suggest universal stages, while conversation analysis, sociocultural methods, qualitative methods generally, critical approaches and metaphor analysis suggest inimitable variations. This ordering of the chapters yields far more than a rag-bag of advantages and disadvantages; it gives us brief history of the mainstream of second language acquisition, beginning with the rather Pyrrhic early struggles to establish a universal order of morpheme acquisition, continuing with a concern with larger, less repeatable, and less orderable units such as utterances, culminating with a focus on unique speech events, uniquely situated in non-repeatable moments of discourse. In this way, Ellis and Barkhuizen succeed in creating chapters that interact with each other. For example, the advantages of obligatory occasion analysis arise directly as a response to the disadvantages of error analysis (because obligatory occasion analysis provides some way of accounting for learners who think ?when in doubt, leave it out?, while error analysis does not). On a larger time scale, the advantages of looking at learner language as content compensate for the disadvantages of looking at as mere form, and the advantages of considering what learners have to say arise from the disadvantages of merely looking at how they say it. A kind of historical narrative emerges, not of neat stages supplanting each other (though one is always tempted to summarize it in this way) but of different schools of thought contending, and eventually finding differentiated niches by providing rather different answers to extremely different problems.

Perhaps the days when a treatment of the core issues of second language acquisition could be objective are past; the authors admit they ?cannot promise to guard entirely against their own biases? (p. 3), and their text bears this out in ways both big and small. At the big end, we have ?obligatory occasion analysis? and ?frequency analysis?, which are little more than data-treatment techniques, given whole chapters to themselves (and even then these whole chapters spill over into chapters such as ?Analysing accuracy, complexity, and fluency?), while all ?sociocultural methods? have to share a chapter, as do all ?critical approaches?. At the small end, we have a not-so-little table on pp. 10-11 where the interpretative and critical research paradigms are compared to the normative. In the interpretative and critical research columns we find expressions like ?thick description? and ?discourses? in inverted commas which make it very clear that the authors themselves would not use quite these words in quite this way. In the normative column we see that the purpose of normative research is to test an explicit theory of L2 acquisition objectively and that normative studies measure learning outcomes, but we look in vain for those ironical little punctuation marks. Especially unsatisfying is the section on Vygotsky. One of the few clear links that the authors manage to make with the issues tying the rest of the book together is, as the authors themselves admit, highly suspect?the attempt to repackage the irreducibly social zone of proximal development of Vygotsky as the utterly cognitivist ?i + 1? notion of Krashen (p. 234). Then, once again, we lose the plot in a pile of studies, and the much anticipated ?final comment? to this chapter ends somewhat lamely?with little more than a list of advantages and disadvantages, and a call for further studies.

By the end of the volume, this bias towards the normative study begins to seem a gratuitous frustration of the overall design. Since the normative chapters are mostly at the beginning of the book (and tend to be situated chronologically near the beginning of the historical narrative of second language acquisition), it inevitably makes the book front-loaded (and backward looking). More importantly, it disrupts the way in which one approach to learner language improves on another and leads to a third. For example, although the penultimate chapter by Michael Barlow, which deals with learner corpora, begins on p. 335 with a bracing restatement of the book?s central theme, viz., that language may be treated as ?expression? or as content, the chapter itself is a clear throwback to the concerns of early chapters or even earlier. Amusingly, the sample study used to exemplify the author?s software on pp. 350-356, which compares learner productions with a database apparently drawn from the Times newspaper, is a pure example of the Comparative Fallacy criticized on pp. 61, 93, and 121, and finally disposed of on the pages following hard on Barlow?s chapter (p. 360). My grad students will gather from this study, correctly, that corpus software has made a vast number of theoretically unmotivated lines of research very easy to do.

This kind of anticlimax is entirely unnecessary, for the book is, at bottom, not only well-intended but very well conceived. Let us return briefly to that long table on pp. 10 and 11. When we read along the rows, across the columns, it?s quite easy to see how the interpretative paradigm solves certain problems that the normative one fails even to identify (for example, the problem that there doesn?t seem to be a single homogenous population of learners which we can make predictions about from a random sample), and the critical paradigm provides a clear program for teachers where the interpretative one fails to. For example, reading across the first ?theoretical orientation? row, we notice that the authors do use ironical punctuation in the normative column when they say that in its theoretical orientation normative research seeks ?ultimate truth?. Next, we see that the interpretive paradigm brings this down to earth by seeing ?reality as subjective and relative?, and the critical one corrects the relativism of this by providing ?a clear value position?. The kind of reading is equally possible in the row marked ?research purpose?: the normative paradigm only tests one dependent variable at a time, while the interpretative allows the proliferation and interaction of variables. The critical paradigm allows us to privilege some of these varia...