In her paper on the good language learner, Rubin (1975) identified aptitude, motivation, and opportunity as three factors that account for differential success in language learning. With respect to aptitude, she argued that aptitude tests predict success but do not provide sufficient information to guide pedagogical decision making. Instead, she advocated focusing on the strategies used by good language learners. The celebration of the thirtieth anniversary of Rubin's paper provides an excellent vantage point to view the latest developments in language aptitude research and to revisit the issue of how the concept of aptitude can inform pedagogy.

What is aptitude?

The concept of aptitude current at the time Rubin wrote her paper was that of John Carroll (1962, 1981) who saw foreign language aptitude as a stable cognitive characteristic of those individuals who have a knack or talent for learning other languages. It is defined in terms of speed in language learning. The most widely used measure of aptitude today is still the Modern Language Aptitude Test – MLAT (Carroll and Sapon, 1958) which was designed in the post World War II period to select learners who would be best able to profit from short intensive language courses. It has proved to be a consistent predictor of learning outcomes in a wide variety of contexts (for instance, Carroll, 1981; Gardner and Lambert, 1972). However, despite its predictive utility, the MLAT later came to be devalued by teaching practitioners because of its association with the audiolingual method.

Consider the experiences of a grade-12 student who has studied French for a total of 13 years in Alberta, the first few years of which were in a total immersion program where all instruction was in French. She travels to Montreal for the first time. To her great shock, she finds that she is unable to use her French very much because the bilingual speakers she comes into contact with switch to English the minute they hear her speak. She comments, “I quickly realized … that my French is not the same as Quebecois French” (Haynes, 2001). This experience of communicative failure in the second language (L2) puts a human face on the research findings that characterize the oral production of French immersion students as being “non-nativelike.” Immersion researchers have studied the effectiveness of different types of pedagogical interventions designed to overcome the limitations of immersion instruction. These interventions have included enriching the input learners are exposed to (Harley, 1989b), drawing learners' attention to non-salient features of the L2 (Lyster, 1994), increasing the amount of student output (Kowal & Swain, 1997), and providing unambiguous feedback on learners' non-targetlike utterances (Lyster, 2004). This chapter examines the issue of the non-nativelike quality of immersion students' L2 production from the perspective of the language practice that occurs in early immersion classrooms. In this discussion, we highlight the value of using cognitive skill-learning theory (Anderson, 1983) as a framework for sequencing instructional activities aimed at improving the formal accuracy of the otherwise fluent speech of immersion students.

Introduction: a paradox in SLA research

Individuals who attempt to learn a foreign language differ dramatically in their rates of acquisition and in their ultimate attainment. This is perhaps the clearest fact about SLA that we currently have, and of course it is particularly striking in relation to the relative uniformity of learning rate and ultimate success in L1 acquisition. Bley-Vroman (1989), among other scholars, has argued that these differences demonstrate that the underlying L1 and L2 acquisition processes must be fundamentally different. Whether or not L1 and L2 acquisition develop in completely different ways, it is important for both theoretical and practical reasons to attempt to specify and understand the sources that contribute to differences among second language (L2) learners.

The paradox is that despite this importance, L2 research into the sources of individual differences (IDs) has lagged far behind research seeking universal principles. With the exception of the social psychologically oriented work on motivation and other affective variables pursued by Robert Gardner (see Gardner & MacIntyre, 1992, 1993, for an overview), no major research program focusing on IDs and integrating them within a model of SLA has yet been developed. This is partly due to limitations inherent in the correlational research designs favored by earlier studies, and the difficulty of finding reliable and valid measures of both the learner traits and the L2 learning outcomes (Larsen-Freeman & Long, 1991; Lightbown & Spada, 1993).