develops out of, or is at least partly an elaboration of aspects of, utterance-activity. This
probably sounds at least slightly unorthodox: On a more standard conception, anything
deserving the name of (spoken) language is a different thing ‘in principle’ from the rest
of behaviour.
One simple argument for the standard conception might point out that to do justice to our
intuition (if we have one) that written and spoken language are in some fundamental
sense the same, we should regard the text-like, or digital, aspects of utterance-activity as
language proper, and the remaining twitches, whoops, smiles, wavings and so forth as
something else.
Our view, in contrast, is that we get to do things with words (and enable words to do
things to us) by means of behaviour in which the wordy and non-wordy are closely
integrated, and by going through a developmental period where we do many of the things
eventually done with words without them. We maintain that utterance-activity is the
arena in which what is standardly regarded as language gets started, and that both the
development and ongoing functioning of word-based language are made needlessly
mysterious if utterance-activity is sidelined.
We anticipate at least two major objections to our continuity proposal. Briefly, the first
points out that powerful and sophisticated models of language treat language as digital,
and suggests that the most likely reason these approaches are so powerful is that
language is in fact digital. If this objection is correct, then what we are doing is urging a
retrograde step, where apparently secure results are rendered doubtful. The second
objection notes that if utterance-activity includes (as it does) affective display, then it
includes signals that aren’t arbitrary (e.g. Ekman 1972), whereas we all ‘know’ that
language consists of tokens which are conventionally, arbitrarily, connected up to each
other and the world. This second objection asserts that we’re throwing our net too
widely, and running all the risks attendant on ignoring an important partition in the data.
We don’t propose to argue directly against either objection, merely suggest how at least
one response to each could get started. In the case of the first, note that the power of a
theoretical approach is not by itself a compelling argument for the truth of its
assumptions. The success of physical astronomy based on the assumption that planets are
point masses does not make it more likely that planets are in fact point masses, or that
they truly lack colours or interesting differences in material composition, it shows that
you can get a lot done by treating them that way.
In the case of the second objection, we note that what counts as arbitrary is a matter of
degree, and partly dependent on theoretical perspective.1 We, now, can’t do much about
the association between, say, smiling and feeling good. Plausibly, natural selection could
have latched onto some different patterns of facial motion and gone on to build
connections between those and social and affective states. So smiling could be non-
arbitrary to us, but arbitrary from the perspective of one interested in the evolution of
patterns of affective signalling in humans. Even supposedly paradigmatic examples of the
arbitrary baptism of some referent with a neologism are, of course, constrained by
contextual considerations such as what words are already ‘taken’, what phonemes are
available to the community in question, what phonetic transitions are easier than others,
what the neologism might sound like, etc.
The insistence on viewing language as a formal system of arbitrary elements involves
playing up what we call the ‘abstraction amenable’ aspects of language at the expense of
others. One particularly famous instance of this tendency to focus on the abstraction
amenable, or digital, aspects of language is, of course, Turing’ s (1950) proposal for an
empirical reformulation of the question ‘can machines think?’ Turing regarded it as a
virtue of his approach that it had ‘the advantage of drawing a fairly sharp line between
2 / 23