winograd is one of the founders of the HCI lab at stanford and has worked on trying to make language understood to computers.
approach through design (the interaction between creating and understanding), and mostly about developing a theory of language
the computer is a device for creating, manipulating and transmitting symbolic (i.e. linguistic) objects.
“every questioning grows out of a tradition” -> what is the tradition of this thesis? (also, meaning is relative to what is understood through the tradition), the analytical tradition
combination of hermeneutics with speech act theory (meaningful acts in situations of shared activity)
mathematical language is, by definition, context-free.
Meaning is created by an active reading, in which the linguistic form triggers interpretation, rather than conveying information.
That which is not obvious is made manifest through language
austin has *five fundamental illocutionary points**:
(each of these points can have different forms)
theoretical language (rational) vs. practical (situated)
Language and cognition are fundamentally social (p.61).
structural coupling (skills) with a consensual domain (piece of source code) is how an autopoeitic structure emerges (self-sustained attention? aesthetic experience?) ????
there are three basic kinds of grounding:
summarization of concerns
The essence of computation lies in the correspondence between the manipulation of formal tokens and the attributions of a meaning to those tokens as representing elements in worlds of some kind. (p.74)
pre-understanding and background are important for meaning-making (fluency). knowledge is always the result of interpretation, which in turn depends on the entire previous experience of the interpreter and on situadness in a tradition.
their approach to language is not that language is transmission of information, but rather that is a form of human social action, with mutual orientation (but i don’t see those as mutually exclusive)
focusing on the fundamental issues of language and rationality that are the background for designing and programming computers.
The first and most obvious point is that whenever someone writes a program, it is a program about something.
(called the subject domain; interesting how it switched to being called problem-domain in software engineering)
the programmer has a systematic correspondence between the symbols contained in storage cells and how they represent objects and relationships in the subject domain.
Success in programming depends on designing a representation and set of operations that are both veridical [produce results that are correct] and effective [they are usable].
most important parts of a formal system (e.g. a programming language):
the problem is that representation is in the mind of the beholder: in no way do programming languages depend on the fact that they represent something (at least, sth too concrete; because they do have pointers, registers, etc., which represent something tangible).
cf. Newell & Simon, 1976, “Computer Science as Empirical Enquiry” -> for the ability of computers to refer to themselves
computers have levels of representation (abstraction?)
the subject domain is always the next higher level itself (p. 89)
some issues, though: