from being faithful to what we see in our data. In fact, once a construct has
been invoked for child data, it may turn out to have a place in a complete
theory of adult behaviour; after all, linguistics still does not have a theoretical
apparatus to deal with the difference between active and passive vocabulary
or partial knowledge of a syntactic form. In the research tradition of Charles
Ferguson (Ferguson & Farwell , p. ), a better approach is to ‘try to
understand ’ development in itself in order to improve theory,
even if this requires new constructs for the latter.’
An equally important issue is S&G’s apparent acceptance of the equation
of ‘innate constraints’ with ‘innate knowledge’. In fact, a major difference
between emergentists and people who prefer to postulate a high degree of
innate knowledge of language is that emergentists rely on innate processing
mechanisms and their inferred computational biases as a primary source of
constraints. Hearing, vision, memory, storage, retrieval, and other properties
of the brain and body are the raw material of the mind; hearing and memory
are already being structured by experience in late pregnancy, and this
interaction continues after birth. ‘Innate processing mechanisms’ seems a
redundant phrase, but it does have to be made explicit in order to be properly
emphasized: the structure and initial operating mode of the body (including
the brain, motor, and sensory organs) is something that is innate but is not
knowledge.
The other source of constraints on language is language itself. What we are
called upon to learn has been filtered through the minds and bodies of
countless generations of learners. S&G ask: ‘How did the input become so
regular?’ – i.e., as regular as it is. The answer is that it has been kept within
learnable bounds by the limitations of human information retrieval, memory,
perception, and learning – including second-language\second-dialect learn-
ing. (There’s a speculation that languages rarely learned by outsiders – small
isolated languages – tolerate more morphological irregularities than those
which have a constant influx of late-acquirers. Consider Icelandic!)
Here’s a small example to show that there are constant and countervailing
pressures on language, arising from how people learn it. The Oxford English
Dictionary shows that the meaning of the word enormity has slipped back
and forth among ‘irregularity’, ‘great size’ and ‘great evil’, and that
enormous has also had the adjectival forms of all three of these meanings
over the last several hundred years. However, sometimes – as recently –
enormous was dominated by one of these meanings, while enormity
denoted a different one. It seems that the transparency of the -ity mor-
phological structure pushes the two words towards semantic agreement,
while different contexts of use pull learners to differentiate their meanings.
The noun is currently moving back towards the adjective; the OED
says that the use of enormity to mean ‘great size’ instead of ‘great evil’ is
Bodroux
