It is clear that learners make ample use of memorized bits and pieces of language. These memorized formulas, which can be found in many learner transcripts, have been seen under different theoretical lights over the history of interlanguage studies. In the 1970s and 1980s, L2 users’ initial reliance on memorized formulas was thought to be spurred by communicative and strategic motives that happened to promote learning. This was Lily Wong Fillmore’s (1979) conclusion in her study of five Mexican children whose families had just emigrated to California for farm work. She recorded the children weekly over a year as they interacted in the playroom of their school and, in doing so, she documented their gradual transformation from no English at all to being relatively functional in their daily dealings with English-speaking peers. All five children were driven by social needs and adopted a general ‘speak now, learn later’ approach (p. 215). They soon began using memorized material which ‘was learnable and memorable by virtue of being embedded in current, interest-holding activities over which the learners had already acquired some mastery, and from which they had already received social rewards’ (p. 211). These bits and pieces of language were used in an unanalysed fashion at first: Wait a minute, You know what?, Knock it off, It’s time to clean up,
No fair!, Gotcha. Among the children, six-year-old Nora proved to be a ‘spectacular
success as a language learner’ (p. 221) because she exhibited the fastest learning pace by far. Wong Fillmore discovered part of the reason was Nora’s formula-based analysis of the input, which she intently sought through abundant social interactions. Table 6.1 summarizes the attested history of Nora’s use of how do you
do dese (‘how do you do this’) over the school year. As can be seen, the formula
began to be combined with increasingly more varied slots and was successively analysed and broken down into grammar-like constituents. By the end of the study,
how-questions had emerged in Nora’s interlanguage, even though they still ‘lacked
the detailed refinements which required further analysis’ (p. 215).
More contemporary usage-based theories of L2 learning posit that the process of formula-based analysis is not only a springboard to communication and grammatical analysis at beginning stages, but the stuff of acquisition, in that it guides the majority of the acquisitional task (N. Ellis, 2008). In this view, language acquisition proceeds bottom up, from formulas to low-scope patterns, to constructions, in an implicit and inductive process that is only peripherally intentional or strategic. Ellis (2002a, 2002b) explicates the process as follows. The first step is the registration of formulas (also called items or exemplars), defined as
Formula-based learning: the stuff of acquisition 115
the pairing of a form and a meaning that is experienced in a particular language use event. As part of the processing of meaningful input, all experienced material is mandatorily tallied, and information about the frequencies, distributions and contexts of exemplars is implicitly encoded in memory upon each new encounter. Learners will encounter some items repeatedly and in contexts of meaning that are relevant to them; if the form–meaning pairings are frequent enough, and the formal and functional cues to how they work salient enough, they will extract information that cumulatively leads to generalizations from such experiences. In other words, repeated experience of the same formulas enables the abstracting of low-scope patterns, the second step of usage-based language learning. As Lieven and Tomasello (2008) explain with respect to L1 acquisition, low-scope patterns are often extracted around a single high-frequency word or chunk that is prototypical of the pattern, or an ‘island’ that helps learners get a quick fix on some generalization at first (e.g. Nora’s how do you do dese). Mechanisms of bootstrapping, or induction by categorization and generalization, enable the third step of gradual abstraction of the pattern into a construction or schema. As Lieven and Tomasello put it, ‘distributional analysis based on the relation between a form and (child-identified [or learner-identified]) functions, leads to linguistic representations developing internal structure’ and makes the inductive analysis progressively ‘less item-based and more schematic’ (p. 169).
It is important to note that L2 learners vary widely in their relative ability to make successful use of these processes. For example, we saw in Chapter 4 (section 4.1) that Wes was extremely adept at memorizing language and using it to great communicative effectiveness, but he never showed evidence of unpacking formulas (Schmidt, 1983). The same individual variation has been found in formal classroom learning. For example, in a study of wh-questions in French as a foreign language
Table 6.1
Nora’s use of How do you do dese over a school year
Time Formulaic part … … Slot variation part
Second school quarter How do you do dese
Third school quarter How do you do dese … … September por la mañana … flower power
… little tortillas … in English
Fourth school quarter How do you … … like to be a cookie cutter … make the flower
How … … did dese work
… do cut it … does this color is
conducted in the United Kingdom, Myles et al. (1999) documented how 11-year-old school students diligently memorized the question comment t’appelles-tu? (‘what’s your name?’) from their teachers and materials. Many began using it as a chunk when asking questions about a third entity: Comment t’appelles-tu le garçon? (with the intended meaning ‘what’s the boy’s name?’). Gradually, other versions appeared in their oral production that suggested some analysis, for example, without the post-verbal clitic tu: Comment t’appelles la fille? (meaning ‘what’s the girl’s name?’). Finally, the third person form of the verb with its pre-verbal clitic also emerged: comment s’appelle le garçon? However, only ten of the 16 students they investigated underwent this full process of analysis. At the end of their first two years of French in school, the remaining six had not gone far enough beyond the initial formula.
It would also be misguided to think that once formulas are analysed and give way to abstract constructions and schemata, they ‘disappear’ or dissolve themselves into this generalized knowledge. Instead, all levels of knowledge (formulas, low- scope patterns and constructions or schemata) ‘remain in the inventory … that speakers have created, to be drawn upon at various levels of abstraction’ (Lieven and Tomasello, 2008, p. 175). In fact, many language researchers are interested in describing the end result of such multi-layered repertoire of creative plus memorized language (e.g. Wray, 2002; Schmitt, 2004; Meunier and Granger, 2008). In addition, interesting efforts are beginning to be directed to support the L2 learning of formulas through instruction (Cortes, 2006; Fitzpatrick and Wray, 2006; Taguchi, 2007). These studies help us appreciate the central role that knowledge of formulas plays in the making of idiomatic and fluent L1 and L2 speakers.