(Waningeet al.2014). For language learning a student may be motivated at
the year scale because she wants a good job for which knowledge of that
language is mandatory. On the months scale she may be motivated by
the prospect of the end of year examinations. On the week scale it may be
the tests she has to do and on the day scale it may have to do with inter-
actions with the teacher. The factors on these timescales interact and so
form a CDS.
In research on CDST a distinction has been made between dynamical
systems and dynamic systems. The former refer to the quantitative and
mathematical approach to CDST, while the latter is oriented more on the
qualitative and social aspect of CDST. The application of the mathematical
tools of dynamical systems theory requires large numbers of data points. For
instance, Van Ordenet al.(2003) had informants repeat the same word 1,100
times in order to have enough power for the fractal analysis. For most of the
research in AL such numbers are hard to reach.
One of the key concepts in CDST is the fractal nature of many processes
that develop over time. The concept of fractals was developed by Mandelbrot.
It refers to the self-similarity of phenomena at different scales of granularity.
A fractal is“a rough or fragmented geometric shape that can be split into
parts, each of which is (at least approximately) a reduced-size copy of the
whole”(Mandelbrot 1982: 23). The best-known example of fractals is the
shape of the coast of Britain that Mandelbrot pointed to. By changing the unit
of analysis or the scale of measurement shapes are found that are similar at
different grades of granularity.
Development can also be fractal in nature with similar patterns at different
timescales. For the study of motivation fractals are interesting, since we may
assume that there is similarity in patterns on different timescales (days/weeks/
months).“In dynamic terms, the timescales may be fractal, or have self
similarity at many levels of observation”(Thelen and Smith 1998: 277). But
for a mathematical approach to fractals, very large numbers of data points
are needed, which is problematic for measuring motivation, since we cannot
measure levels of motivation onfiner timescales (minutes/hours). There is a
risk that such frequent assessments actually have an impact on the phe-
nomena they intend to measure, because subjects will object to being asked
to reveal their motivation so frequently. An interesting alternative is the
use of physiological measures that are related to emotions. MacIntyre (2014)
studied a group of learners of French in Canada and gathered online physio-
logical data (event-related skin resistance and heart rate) while subjects car-
ried out certain verbal tasks in their L2. After they had completed the task,
the subjects were asked to reflect on the patterns the measurements revealed.
For instance, when there was a spike in the physiological data, the subjects
were asked what might have caused that change. Quite often the subjects
reported word-finding problems at these moments. So this technique allows for
registering changes in emotional state at the millisecond level with reflections
on the minute level.
102 Trends III