Curriculum vitae

Jan Koster was born in Delft, The Netherlands, on July 8th, 1945. He studied Dutch language and literature at the University of Amsterdam and got an MA in theoretical linguistics in 1972 (cum laude). He obtained his PhD in 1978, also at the University of Amsterdam (cum laude) with a dissertation titled Locality Principles in Syntax. In 1976, he was a visiting scientist at the Massachusetts Institute of Technology, where he studied with Noam Chomsky and the late Kenneth Hale, among others.

Koster was assistant professor at the University of Amsterdam (1972-1981) and Utrecht University (1974-1975) and participated as postdoc in a ZWO research project in cooperation with The Max-Planck-Institut für Psycholinguistik at Nijmegen. In 1981, he became associate professor at Tilburg University and in 1985 he was appointed full professor of linguistics and the philosophy of language at the University of Groningen.


Koster's ideas on syntax and semantics are "Chomskyan" in orientation but differ from the mainstream in a number of respects, particularly in their cultural, strictly-lexicalist character. In his overall theory of language, the capacity for language, even in its narrowest sense, is not seen as a matter of biology but as applied biology, i.e., a technology belonging not primarily to individuals but to their shared culture. Invented words rather than syntax are at the essence of language in this view, while recursive syntax is seen as a successful extension of the properties of the cultural objects in question ("words"). The combinatorial potential of words is as cultural and non-individual as the words it belongs to and therefore first and foremost public property that individuals seek to adopt from the day they are born into a community. Innateness is not at issue here, since all our cultural activities are constrained by our biology. Language only more so than, say, playing the piano.


Syntactic structures are not generated by lexicon-independent rules (like phrase structure rules or Merge) but as the spelling out of the contextual properties of lexical items ("valency"). The elements of spelled-out valency frames may share their properties, but only in strictly local configurations. For the hierarchical structures generated, this means that information can only be shared between a node and its sister node or a node and its immediately dominating mother node. Property sharing over longer distances can be achieved by iteration of strictly local sharing via mother nodes (that become a sister on the next cycle, etc.). Minimal cycles are called "triads" and consist of two combined sister elements (cf. binary branching) and an immediately dominating mother element that contains a subset of the combined properties of the daughter elements. Syntax, then, defines the strictly local conditions for property sharing on the basis of the hierarchical structures defined by the valency of lexical items. Merge, in this view, is at best a characterization of a pre-linguistic, not-yet-applied background capacity.

Lexical semantics

Koster's theory of lexical semantics can be characterized as "informational-interpretive." According to this theory, what is associated with words in the physical world (including the brain) is not meanings or concepts but complexes of coded and incomplete information that are partially represented in individual brains but ultimately belong to a shared culture. Meanings and concepts are only accessed or created and completed in this view via ultimate decoding, i.e., as the contextual and often selective interpretation of said information complexes by a living individual. This form of interpretation (of data structures) is central to the theory of meaning and, unlike what we study in the natural sciences, it involves an inalienable first-person dimension (like qualia and consciousness in general). We do not know of meaning other than in relation to participants in history. Ultimate interpretation usually involves conventional (hence cultural) elements but can also be innovative. Since interpretation is a free, agentive form of application of information complexes to contexts, it is an open-ended and often truly creative process that must perhaps be compatible with, but cannot be reduced to, biology and the brain sciences.

The extended mind

Koster's view of semantics and mental content in general is based on a rejection of mind/brain identity. The well-known notion of mind/brain defines the relationship between brain and mind both too narrowly and too broadly. The brain is too broad as a model of mind because much (perhaps most) of the brain has functions other than causing or processing the mental, for instance, sustaining respiration and other automated body functions. It is too narrow because the necessary information used to create mental content is distributed over memory structures both within some individual's brain and outside that brain, for instance, in other people's brains and in records of our symbolic culture, as in libraries and other (collections of) media.

As there is no principled distinction between the various brain-internal and -external memory structures (apart from mode of access), it is not possible, unlike what we can do with the brain, to construe "mind" as a strictly individual concept. This highlights the fact that humans are not isolated islands from a mental point of view but live in symbiosis with a shared culture and its symbolic records. This conception of humans as symbionts goes back to a substantially modified version of Karl Popper's world 3 concept and Merlin Donald's rejection of "the myth of the isolated mind," among others. More recently, the symbiotic view was advocated by David Chalmers and Andy Clark in their Extended Mind Thesis (EMT). If correct, the EMT has profound consequences for how we view language and mind.


| home | recent articles |

Last update: February 23, 2009