I have here summarised a brief but, for me, helpful email exchange I have had today with Harry Collins starting with a query he had about the final line of my previous blog entry on his group’s new thinking about interactional and contributory expertise. I had said that it seemed to have undermined what seemed interesting to me in the original account: that there was a fundamental kind of non-practical but still tacit knowledge. But Harry pointed out that:
“There are still people who have tacit knowledge in virtue of their language alone and nothing from [practice]; some of those are special interactional experts. The point is that there are lot more non-special interactional experts than we thought: these have something from the practice but most of it from the language.”
Now the point I was trying to make was that one difference between the new and the old picture was that on the old picture, interactional expertise (IE) seemed to be a further development from contributory expertise (CE) (with the latter serving as a paradigm of tacit knowledge for a reader like me). So IE seemed like an interesting extension to (a kind of optional extra to) CE. But on the new picture, with language at the base of everything, IE isn’t an extra but is instead presupposed by CE. (And that isn’t really affected by the existence of logical limit case of Special Interactional Experts (SIEs) who have the language without any of the practices.) So whilst I like the new picture because it places (if I follow) language (or conceptually structured ability) at the heart of tacit knowledge, interactional expertise as a phenomenon looks less interesting.
Harry replied by pointing out that the ubiquity of interactional expertise shouldn’t make it any the less interesting:
“I don’t see IE as less interesting because it turns up everywhere. We just hadn’t noticed that most of what was going on was IE rather than CE and that seems very interesting to me. It seems interesting that though language is practice-based it, as it were, contains the practice on which it is based so to that an understanding of that practice can be acquired by individuals via acquisition of the language alone. And we now realise it has to be that way if we are going to have society as we know it.
Sociology, and policy-wise, the ‘special IE’ people are still specially interesting over and above this, of course, because they are normally excluded from decision-making as they have no obvious practical credentials.
All that has changed outside of the SIEs is that we now discover that even those with practical credentials are using their practical experience hardly at all, or not at all, when they make judgments. The whole programme of research has been going this way from the beginning when IE was first thought of as an exception, then when it was realised that managers have it, then when it was realised that it was central to the division of labour and this is just the ‘gestalt switch’ was inevitable as things went along in this way.”
This reply helped me focus my worry. The most obvious way in which “though language is practice-based it, as it were, contains the practice on which it is based” is an intellectualist account of the priority of language over practice. It would be true, eg, of the platonist account that Wittgenstein undermines (and which Collins summarised in the introduction to Changing Order). Language would contain practice by fully describing it. (Of course to speak is to take part in a practice so there would still be an element of truth to language also being practice-based.)
Thus my worry was that the trajectory of Collin’s research group seems to be from those who put practice over autonomous platonist codification (Wittgenstein, Ryle, the early Heidegger on some interpretations) towards those who play up knowledge-that as the basis of knowledge-how (very influentially these days: Timothy Williamson in Oxford, but in a kind of a way also Jerry Fodor in his AI-inspired approach to the mind). Amongst those latter people, the idea that there may be SEIs without any CE wouldn’t come as any surprise: they (the SEIs) are just those people who master a language. Others - mere artisans, as it were - may also acquire some additional CE but it now seems to be an inessential add-on since language itself already contained the practices in question.
What would thus look to have gone missing would be the idea that there is a fundamental contextual element to practical expertise, an element not contained in what those who merely master general language learn. As a long standing fan of Changing Order, that’s a qualm.
Fortunately Collins has, apparently, not given way to an intellectualist dismissal of the fundamental role of tacit knowledge, replying:
“Well Tim, I am not a philosopher but the key idea might be the following. You say:‘Language would contain practice by fully describing it.’I say, absolutely not! The practice is not contained in what people can say, it is contained in their saying it. Language, as one might say, is a practice and it is that practice that contains the other kind of practice (the understanding of physical practice). It is quite hard to know how to say that well. But I am still completely with Changing Order in that the visible formal descriptive content of language cannot capture practice and that still means there are no recipes for experiments and so forth.As far as Changing Order is concerned I would now like to redo my TEA-laser study, say, and work out how much of the learning that went on when laser-builders actually visited other laser builders took place in the talk and how much my practical manipulation in the lab. Probably impossible to pull apart in practice but pull-able apart in principle.”
The first distinction in this reply suggests a distinction with which analytic philosophers have not, in the main, been very interested but might be that between ‘langue’ and ‘parole’.
I can also see how, especially perhaps with that distinction in mind, the further exploration of the distinction between manipulation and talk would be intriguing. But I wonder what Collins could find there. He said that the ‘visible formal descriptive content of language cannot capture practice’. What difference does moving to consider the particular moves that might be made in say the Cavendish coffee room (as opposed to the moves that were grammatically licensed) make? Here’s my worry: leave a tape recorder running to capture these utterances and write them down. What difference is there now between these and an instruction manual written for a Wiley science textbook in formal descriptive language?
Where one might have a difference is (to echo Changing Order) in the laboratory itself with an expert saying to a student: the connections should look like that! This isn’t transferable in merely conventional linguistic terms because we need to see the connection to know what the concept of ‘like that’ means. I’m a disciple of a particular view of a bit of philosophy (McDowellian) which says that this is itself a concept (so not all concepts are fully linguistically codified). This also fits with what Wittgenstein says about samples being part of language. So I’d say that this was consistent with your general ‘language is at the heart of expertise’ view (with language construed as conceptual activity governing both words and samples). But it doesn’t sit happily on either side of your distinction between talk and practical manipulation. So I’m not sure that that’s a helpful distinction.
In reply to these thoughts, Harry said:
You are getting there but still haven’t quite got it.
I would not find any interest in what was in the Cavendish conversation – what was recorded on the tape that I could listen to – I would be interested in how long the novice spent in conversation with the others about technical matters.
Think of this as an analogy. When a child learns a natural language like English the child learns, for example: Verb in the middle not at the end. But nothing you the philosopher/sociologist would hear on the tape would correspond to ‘verb in the middle not at the end’ (unless you already knew grammar and could do some meta-analysis of the conversation). What you, as a child, would be learning would be how to do it (put the verb in the middle) and you would have learned how to do it without being aware that you had learned or even that you now knew it. You would just have learned to speak.
I visualise the practical knowledge you learn in becoming fluent in a practised-based language as being like this grammar: it isn’t spoken, yet it is contained. The reason it can be contained is that the language is formed by the physical practices in the first place. It is in the silences, as one might say (or in the form of the words, the nuances, the balance of the vocabulary for this and that, and I don’t know what else). You get none of this from books -- you must have fluency.
Is that better?
I will have to think a little more about how this analogy works between learning verb placement from utterances and practical knowledge from utterances. Of course, things are made much more difficult because of the well known problem of both codifying the patterns of correct usage children learn (in, for example, transitions between questions and statements) and the apparent paucity of evidence for merely inductive generalisation (which is why Chomsky thinks grammar is innate). But if that could be put aside for the moment (and I am going to assume that there will be no equivalent of Chomskian innativist explanation in the other part of the original analogy: getting science practice from science utterances), what is the sense in which correct verb order is or isn’t in the utterances?
Well it is not contained in a meta-linguistic grammatical instruction (the one which is also so hard to formulate). So it’s not there in that sense. And it might seem now that that metalinguistic rule has merely a shadowy presence in the actual utterances. But there is something that is simply there in the utterances a child hears: correct verb placement. Contra Harry’s reply it seems to me that it is spoken. The child hears the correct pattern in the utterances they hear, a pattern which we might now explain (in the absence of the metalinguistic rule) through the use of the same examples the child hears. So that doesn’t seem too mysterious after all (pace the major Chomskian issues).
The other complication in the analogy is that the child is learning grammar, not moves then to be made within grammar. So they are learning the possibilities of what was earlier called the ‘visible formal descriptive content of language’ which was, it was suggested, not up to the job of capturing practice. It’s not clear to me whether we can apply this in the case of the apprentice scientists learning from the utterances, their gaps, foci and pregnant silences, to say that they too are learning a ‘grammar’ of scientific practice, a grammar which, for the analogy to help, could not be captured in a textbook account of the basics of science which might also have characteristic gaps, concentrations and odd silences.
I’ll have to think more on this. I’m not unaware, of course, of the irony that I’m sitting in my philosophical study talking theoretically about tacit knowledge to a sociologist who actually investigates it practically!
PS: on the following day.
Yep, I'm happy with that as a reproduction of our email conversation.
It's your blog so you should have the last word but let me just say one more thing -- whether you put it up or not is up to you.
Note that neither the child nor the philosopher/sociologist (who is not already dealing with the abstract notion of grammar) in the analogy, ever `knows' that the verb is in the middle. The philosopher/sociologist who merely listens to the conversation will not possess the knowledge that the verb goes in the middle. The child only comes by that knowledge by actually learning to speak. That is why there is such a stress on participatory methods in sociology: you come to know not by observing but by participating. And you generally do not know what you know. But having learned to participate you can use this `unknown' tacit knowledge to make good judgements even if you don't know how you make them.
The idea of learning a language to the degree of fluency required by the interactional expert is, then, congruent with idea of learning a physical practice like bicycle-balancing. You come to know (that is not know) how to do it by practicing not by observing and when you have acquired the practice you cannot describe the contents of what it is you have acquired. Language is the same -- language is in this sense `a practice.' The idea of interactional expertise is that in learning (not observing) a physical-practice-based language you also learn to understand the practices on which the language was based. You don't learn them explicitly, you learn them after the fashion of acquiring tacit knowledge.
This exchange is useful to me because, perhaps because of the tradition I have always worked in, it has always felt good enough to me to say things like `language is a practice, and contains practice -- geddit?' I always assume that those around me will be satisfied with that. I am now being forced to spell it out a bit more. I am now `thinking on my feet' but one way fluency can contain knowledge is through the un-self-conscious acquisition of the frequency with which something is said in the community. For example, if I want to know the importance of, say, some source of noise in an apparatus -- is it just a nuisance or is it fatal? -- I might pick it up from how often it is mentioned in conversation. I might also pick it up from whether it causes people to joke about it in a certain way (which they do in the case of serious things), or joke about it in another way (which they do about silly things), the reverence with which it is mentioned and how all these things balance out. Of course, in lived language the balance of these kinds of things is changing all the time which is why the interactional expert's abilities to make good practical judgments diminish rapidly as soon as they are no longer in touch with the language community.
I am not here trying to develop a calculus of conversation or a recipe for the tacit knowledge in language, just indicating one way in which it is possible for there to be knowledge in fluency that is not in the descriptive content of what is said. But I would no more think about trying to work this out fully than I would want to try and work out what it is in the muscles, brain and nerve pathways that enable a person to balance on a bike. All I want to do is show how it can be that fluency in a language can contain much more than is said.
PS: For my first thoughts on his recent book on tacit and explicit knowledge see this.