So we must approach the problem from both a practical and a conceptual
perspective, and find a way to make them coincide. For Bakhtin, utterance
unit boundaries can be concretely identified by turn-taking. For Vygotsky,
word meaning is the irreducible unit of analysis, beneath which the
functional integration of word and meaning ceases to exist.
If conversation is, indeed,
that irreducible functional unit, then what is the smallest concrete form
conversation can take? It cannot be defined as a turn at talk, for some
turns can be quite extensive, such as a monologue consisting of multiple
sentences.
But if an individual utterance is defined in terms of a single word (at
minimum)
or a single sentence (at maximum), and these linguistic structures are
shown to have the functional properties of conversation (i.e., they participate
in an initiation-response structure with other utterance units or practical
activity),
then this unit would meet both the practical and conceptual criteria we
have been discussing.
Of course, this does not address all of the problems associated with the
analysis of private speech utterances, for there is still the knotty
problem of
*who is conversing with whom*! But that moves the problem down a different
path, which is a whole other topic.