Peter Anderson
In this paper I discuss the theory of "recasts", a type of corrective feedback closely studied in the last two decades by SLA researchers working from a cognitive-interactionist approach; I also present a model for the pedagogical application of form-focused recasts.
A recast is a reformulation of a wrong utterance that preserves meanings and is employed in order to solve a problem in communication. Form-focused recasts, however, are embedded in controlled classroom contexts, in which there is little need to negotiate meaning, and have a pedagogical aim: the correction is more implicit and can be embedded in an active, task-based, communication (Ellis 2010, Plonsky & Kim 2016). Recent studies show that certain kinds of form-focused recasts (oral or written) when supported by learner attention strategies, can lead to significant gains in acquisition among students (McDonough 2005). The "Comprehensible Outputs", whether oral or written, that result from these kinds of recasts meet the active production requirements in the new Standards. Since recasts seem to be the most frequent kind of corrective feedback in L2 classrooms (Choi & Li 2012) and are therefore likely already used effectively in oral "active Latin" classrooms built on communicative models, in the pragmatic section of my paper I focus on the potential of form-focused recasts in written active Latin.
Recent research demonstrates the value of cognitive-interactionist approaches to L2 learning. A focus on the entire cognitive-interaction cycle (input, interaction, output) is presenting empirical evidence for the positive (and negative effects) of focus on form strategies, especially those based on corrective feedback (Loewen 2012). While corrective feedback might be seen as a useful interaction, such interaction alone seems less beneficial than corrective feedback with modified output (McDonough 2005), partly because instances of corrective feedback alone vary in effectiveness relative to a very wide range of factors (Ellis 2006, Mackey 2006, Jeon 2007, Loewen 2012, Brown 2016). Gurzynski-Weiss & Baralt 2015 demonstrates that partial modified output after feedback (isolated repetition and repetition only of the element corrected) was the single best tool for developing student skill in "noticing" linguistic structures. I use a partial modified output exercise to model a form-focused recasting strategy, using corrections of morphosyntax (recasts are also useful for lexical and phonological errors). For instance, to promote acquisition of final clauses we might:
1. elicit or input (inaccurate) example, e.g. Caesar Brutum mittit ut [pacem petit].
2. (teacher) offer corrective feedback as recast, e.g. ut pacem [petat]
3. (students) generate partial modified output using a variable lexis of verbs only: e.g. ut pacem [perdat]; e.g. ut pacem [servet]; e.g. ut pacem [cohibeat]; or generate partial modified output with variable lexis of the whole clause: e.g. libellos legat, det, obtineat.
4. (teacher) offer corrective feedback to elicit correct utterance as needed.
5. elicit or input (inaccurate) example using new tense, e.g. Caesar Brutum misit ut [pacem petat].
6. (teacher) offer corrective feedback as recast, e.g. ut pacem [peteret]
7. (students) generate partial modified output using a now familiar lexis but other tenses: e.g. ut pacem [perderet]; e.g. ut pacem [servaret]; e.g. ut pacem [cohiberet]; or e.g. libellos legeret, daret, obtineret.
8. (teacher) offer corrective feedback to elicit correct utterance as needed.
The syntax and lexis on which this method can be imposed is very wide, but it follows a consistent pattern that supports deeper acquisition of structures (syntactical or lexical) to which students have already been introduced: input / error; interaction / correction; output / noticing / reinforcement through repetition. In oral, primarily communicative classrooms, errors are likely already prevalent and do not need to be elicited (although a specific error might need to be prompted). In classrooms that have not primarily adopted a communicative model, the same effect and opportunity can be recreated in controlled exercises.