language as a tool system mark h. bickhard [email protected]
TRANSCRIPT
Language as a Tool SystemAbstract
It is common knowledge that language can be used as a tool to accomplish tasks.
But utterances are taken to be constituted as encodings of mental contents.
I argue that this encoding view is in error, and that language is directly constituted as a tool system.
In particular, utterances are tools for engaging the common understandings that constitute social realities.
Doing Things with Words J.L. Austin was an early investigator of
actions performed with words, e.g., assertions, commands, promises, etc. Searle is one contemporary who has
followed in this path Others also focused on actions using
words, such as the later Wittgenstein’s likening of language to a tool box
Actions with Encodings All of these approaches have in common the
assumption that the actions are performed with encoded propositions
These propositions represent states of affairs that are asserted to exist, commanded to be brought into existence, promised, and so on.
The encoded representations are used in the service of the actions
Problems with Encodings But such encoding assumptions
encounter serious problems The problem is not that encodings do
not exist — they clearly do But they cannot serve primary epistemic
functions
Encoded Representations Consider how actual encodings work:
Example: Morse code “...” encodes “S”
Example: Neutrino counts encode properties of fusion processes in the sun
Encodings change the form of representation This can be useful:
“…” can be sent over telegraph wires while “s” cannot Neutrinos can be counted on earth, while fusion in the
sun is not accessible
Encodings are Derivative Encodings must borrow their
representational content They are derivative representations, not
primary Some agent must know — represent —
both ends of the encoding relationship, and the relationship itself.
Encodings Cannot be the Basic Form of Epistemic Access Consequence: encodings cannot cross epistemic
boundaries — they cannot be the primary form of representation for perception or for language
Not mind to world in perception Nor world to mind in language
Because both ends must already be represented, the epistemic boundary would have to be already crossed in order for the encoding to exist
There is no way in which such non-derivative encodings could be defined or learned
Circularities Circularities result from assuming that
encodings can serve as their own ground, can provide their own representational content: Incoherence Piaget’s copy argument Classical radical skeptical argument
Encodingism Nevertheless, the assumption that
representation is encoding is common Encodingism: The assumption that (all)
representation is encoding Representation, including mental
representation, is presumed to be constituted in some kind of encoding correspondence causal, nomological, informational, conventional
Fatal Problems Such encodingism assumptions, however,
encounter myriads of fatal problems: Circularities Too many correspondences Possibility of error Possibility of system detectable error Possibility of emergence
Innatism is not a solution
What Could Language Be If Not Encodings? Apperception
Other side of visually perceived object Animal in forest on basis of sound
Modifies representation of environment Adds, subtracts, transforms
Apperceptive maintenance of representation of environment around you, including extended environment Anticipations of what could happen, what would be
possible
Apperceptions of Utterances Utterances induce apperceptions of
representations Apperceptions add, subtract, transform Utterances operate on, transform,
representations Utterances interact
With what?
The Locus of Language What is the locus of language interactions?
Obvious candidate: other minds Partially correct, but this overlooks the sociality of
language What about, e.g., the difference between arranging for
someone to see that X is the case and telling them that X is the case?
Both have mind as object — interactive locus — but the first is not language
In the second case, the result is a transformation of the common understanding between the utterer and the audience: both end up knowing that X was asserted, and knowing that both know that X was asserted, etc.
Situation Conventions Commonalities of understanding —
which I call situation conventions, conventions about the current social situation — constitute social realities E.g, lecture situation, birthday party, formal
meeting, etc.
Utterances as Operators Utterances transform situation
conventions, in multiple ways, at multiple scales Macro: e.g., call a meeting to order Micro: e.g., change the manner in which a
pronoun will be resolved in a conversation
What About Minds? Situation conventions are constituted in
relations of commonality and consistency among the representations of social participants So, transforming situation conventions
does transform minds, but indirectly Lifting a coffee cup is interacting with the atoms
in the cup, but indirectly
Productivity Signals can transform social
understandings Language provides a means for
constructing unbounded ranges of operators on social understandings
Language is productive
Language as a Tool System Language is a conventional recursive
tool kit for constructing conventional tools (utterances) for interacting with situation conventions
Context Dependence One strong consequence: Language is
inherently context dependent Utterances transform the context in which
they are uttered, and therefore are sensitive to that context.
Operators are sensitive to their arguments Unless they are akin to constant functions
Ubiquitous Context Dependencies
Not just for pronouns, indexicals, etc. Even proper names
This is obvious, but has tended to be overlooked in favor of the model of language as encoded Names
Recursive context dependencies: Partee: “The man who gave his paycheck to his
wife was wiser than the man who gave it to his mistress.”
Not co-referential
What Happened to the Propositions?
Utterances are interactions with representational systems
They are not representational themselves
Representations are generated by utterances, not encoded by them
There are no encoded propositions wrapped up in utterances
Many Further Consequences An interactive model of language has many
further consequences Syntax, Semantics, Pragmatics cannot be defined
in standard ways These do not constitute a theory neutral division of the
field of study Functional constraints on syntax generates UG Provides an approach to creative language
Utterances do not encode transformations either Apperception can yield problem solving tasks