‘Language is not a natural kind’

An interview with Martin Stokhof

1 March 2016

Martin Stokhof is professor in philosophy of language at the University of Amsterdam. He studied philosophy and linguistics, and wrote a joint dissertation with Jeroen Groenendijk on questions. His early research was on questions and dynamic semantics; more recently he has published on the philosophy of semantics and on Wittgenstein. In December, a festive event was held on the occasion of the upcoming retirement of Stokhof, Jeroen Groenendijk and Frank Veltman.

‘The retirement festivity was a bit curious, because I am not retiring yet: I still have two years to go. But Jeroen Groenendijk, with whom I have worked intensively, and Frank Veltman, also a co-author and long-time colleague, are retiring in the spring. Because our joint work is best known in the semantics community, the idea was to have the event at the Amsterdam Colloquium, which took place in December. A collection of essays was written for the event, it was a lot of fun reading them. We certainly did not deserve al the praise that has been bestowed upon us, but that is part of how such events go. I am grateful for all the nice things that were said.’

‘Retirement is, unfortunately, compulsory. I know that some people have tried to argue that this is a matter of age discrimination, but I think the European Court has decided that it is up to the various nations to determine whether they want to do that or not. Of course, there are lots of ways to continue doing what you like doing. Even though you are formally no longer employed by the university, you can still be a guest lecturer. And you still have five years in which you can act as a supervisor of PhD students. Since I do not really have hobbies that I want to have more time for, that is what I will do.’

‘The work I did together with Jeroen Groenendijk was on questions, and later on dynamic semantics, which is in the centre of formal semantics, and a bit on pragmatics. At some point I wanted to change directions. At the same time, my administrative load became much more substantial. I was scientific director of the ILLC for more than five years, and chairman of the Humanities Council of NWO for six years. I also wanted to keep teaching, so the time for research became much less. Combined with this change of directions in my own work, Jeroen and I sort of parted ways, at least scientifically. When we started our careers, we were doing more or less the same things on a daily basis, a lot of them together. It took a couple of years for it to become clear that we were going in somewhat different directions. Then we slowly disentangled. Fortunately, our personal relationship has not suffered from that.’

‘When I started moving away from formal semantics, I got interested in Wittgenstein, so I wrote a book and a few papers about his work. At some point, I started coming back to semantics, but more from a meta-theoretical perspective. I asked myself why semantics is what it is. That led to the question of what kind of intellectual enterprise semantics is, for example whether it is a straightforward empirical science, as most people would think. Then you start to explore, you look back at the origins, and you see what kind of principles have motivated the whole enterprise. You see that there is lots of influence from logic, from philosophy, and from linguistics. Semantics of course deals with natural language, and in that sense it is an empirical science. But it also relies on assumptions and concepts from logic and philosophy, which are not empirical in the same way. This investigation has occupied me for the last ten years or so.’

‘There are not that many people working on what I call ‘philosophy of semantics’. I think that is a pity, although I am certainly not saying that everybody in semantics should constantly ask themselves what they are doing. Then nothing gets done! But it is good to take a step back every now and then, and think about what assumptions you have made. That is useful for two reasons. First, we see an enormous diversity in semantics and linguistics. I am, unfortunately, old enough to have been trained in really dominant paradigms. Generative Grammar was what linguistics was all about – at least it seemed like that for a long time – and formal semantics was also pretty homogeneous. Nowadays there is more diversification in terms of methods that people employ, which is a good development. However, it raises the important question of how to combine the results you get from different methodological paradigms. The answer to that is not obvious. For example, if you have a strictly intuition-based methodology, you must subscribe to something like a competence-performance distinction. If you study actual data, then you are studying performance and not competence, and it is not obvious that you can compare the results of both methodologies. The second reason is the connection with other sciences. Despite the success of linguistics, it does have difficulties leading to strong predictions that can be tested in, for instance, cognitive neuroscience. This must have something to do with the way in which linguistics has originally constructed its object of study. Traditional grammar-oriented linguistic description fails to interact in a really productive way with cognitive neuroscience, for example. Of course, interactions do exist, but they are not as straightforward as you might think they are. I think this has to do with a certain lack of clarity about methodological issues.’

‘Two years ago I published a paper together with Michiel van Lambalgen, on the distinction between what we call abstraction and idealisation. In any scientific enterprise, you do not deal with the raw phenomenon, but you must construct an object of study, by focusing on certain aspects and neglecting others. Otherwise you simply never leave the stage of collecting data. Once theory starts, you deal with a certain constructed reality. What we claim in our paper is that in the natural sciences this constructed reality usually takes the form of abstraction. You leave out certain quantitative aspects. For example, you neglect friction, in order to describe the movements of objects in a plane. This leads to predictions that can be verified or falsified with further experiments and observation. But linguistics is somewhat different than the natural sciences. We studied for example the way in which the competence-performance distinction has been introduced in Generative Grammar. In this case, you do not simply leave out a quantitative aspect of a phenomenon, but you idealize away from a qualitative feature, saying for example that the ideal speaker or listener is someone who does not have any memory limitations. The next question, which we are still working on, is: why is it that what looks like the same intellectual movement, has one result if you make it in one area, but has a completely different result in the other area. That, we claim, has to do with the ontological diversity of what language is all about. The slogan of that paper, which we are still writing, is ‘language is not a natural kind’. So language is not ontologically homogeneous, unlike for example matter or energy, or even living matter.’

‘The best phrase that I know to describe what language is, actually comes from Wittgenstein. He talks about “the motley of language”: it is a patchwork, a mix of different things. Language has something to do with very basic cognitive mechanisms that we as humans have, and some of which we may share with other animals. But it is also a social art, which we acquire in a social setting and which functions in a social setting. Language functions in a social setting in ways that are simply not captured by the idea of a homogeneous set of well formed expressions, with phonological and semantic representations that are described by a grammar. Communication succeeds using linguistic material that does not satisfy any grammatical description, and even with incomplete understanding, you can still have successful communication. To account for the entire range of phenomena related to language, you have to deal with lots of things of different natures. Linguistics is therefore only a part of what is needed to understand the phenomenon language in its full breadth and depth. This makes it all the more interesting, I would say.’

‘I have never thought about whether there is a leading question in my research. But if I am forced to think about it, I would choose something that plays a role not just in this philosophy of semantics work, but that also explains why I find the work of Wittgenstein and similar authors so fascinating. Our understanding of the world and of ourselves takes place not just at different levels of understanding, but also in different modes. There is the scientific mode of understanding, which some people think is the only mode that will ever teach us anything worthwhile. There is also is an everyday mode of understanding, an everyday way of coping with problems and questions. Then there are ways of understanding that you find in the arts, for example. And, viewed from a certain perspective, you find different modes of understanding in religion and related areas.

You can argue that it is impossible to come up with a comprehensive account of the ways in which we deal with ourselves and with reality, because that is also constantly changing. There will never be a complete description that you can finalize at some point: our self-understanding changes according to what we find out about ourselves and about reality, and that also brings shifts in other modes of understanding. How these modes of understanding interact, and what are the often quite subtle differences between those modes of understanding, is something that fascinates me.’


by Gisela Govaart, March 2014

Published by  SMART Cognitive Science