The study of SLA, as is true for much social science research, aims
broadly at answering questions of causality—for instance,
“Is one learning context more likely than another to promote
gains in second language learning?” Context-of-learning research
in the study of SLA, however, often involves observational, rather than
experimental, manipulation of subjects. As is well known, this
compromises the ability to draw causal inferences. There are some
interesting possibilities for overcoming this handicap, arising from
recent statistical theory. In particular, the so-called Rubin causal
model has some advantages over other approaches, such as path analysis,
that are more familiar to social scientists. The purpose of this
article is to relate ideas of causal inference to the studies presented
in this thematic issue and to suggest how they may be useful to future
research in SLA, particularly in context-of-learning studies.