outside agent; the state is established solely because of the dynamical
interactions among individual elements of the system: the critical state is
self-organized. Self-organized criticality is so far the only known general
mechanism to generate complexity.
(1996: 1–2)Bak’s ideas have been applied to many differentfields, ranging from eco-
nomics to epidemics, and visual attention to landscape formation. While the
mathematics of SOC are well beyond the scope of the present contribution,
some of the thinking on SOC seems to be applicable to the study of lan-
guage development and language use, and in particular to CS. It can be
argued that in a CS setting, the individual’s language system tends to be
attracted to a critical state in which the transition from one state (speaking
language A, La) to another state (speaking language B, Lb) appears to be
highly unpredictable, but not random. Many different sociolinguistic and
psycholinguistic factors interact in the selection of the language of a specific
utterance or part of that. But even if we accept that this line of thinking
supports Sankoff’s assumption on the unpredictability of CS, we still want
tofind out what might have led to a specific set of switches. Even if we
cannot predict specific CSs, we may learn more about the phenomenon by
looking at it“retrodictively”, to use Larsen-Freeman’s (2009) term. That is,
we may be able to look back at what happened and point out what factors
may have played a role in a specific instance, but we cannot claim that the
same set of factors will lead to the same outcome next time they occur.
The concept of SOC forces us to distinguish cause and reason carefully:
the single grain of sand can be the cause of an avalanche, but it is not the
reason why avalanches take place. Steepness of the slope of the pile, adhesive
forces between grains of sand and structural characteristics of the pile are all
important reasons for the kind of adjustments that avalanches are. Thefinal
grain of sand triggers the avalanche. Likewise, a set of psycholinguistic (such
as availability of elements) and sociolinguistic factors (such as being in a set-
ting in which CS is the conversational norm) make people switch code, but
they do not necessarily define why a particular CS takes place at a given
moment and what form it takes. It should be pointed out here that the lan-
guage system of a bilingual is always in a critical state: depending on the
setting, only one of the languages may be used, but in settings in which CS is
either the norm or there are no restrictions on the use of more than one
language, the systems will move into a state in which a switch is about to
take place. Various factors may interact to make a language system critical
and very sensitive to minor changes that might lead to a phase change or
code switch. There may be many phenomena that trigger a CS. Given the
critical state of the system a minor change in one of its subsystems may cause
a switch to a smaller or larger chunk in the other language. A sound from
the other language, the thought of an event specific to a particular language
setting, the occurrence of a word, sound, gesture or construction from the
96 Trends III