Skilja Blog

Complex Predicates

by | Apr 25, 2013 | Uncategorized

Editor’s note: This is a guest post from Prof. Dr. Jürgen Lenerz from University of Cologne

As we have seen (read the previous post here), predicates such as be red, be sleeping, snores, be a horse etc. may be conceived of as functions from individuals in truth values. What about more complex predicates as those which turn up in the following sentences: John loves Mary, Peter is bigger than Chris, John gives money to the poor, Mary put the necklace around her neck etc.? These are clearly predicates which define a relation between two or three individuals, hence 2- or 3-place-predicates. There are several ways to deal with them. The preferable way seems to be to consider them as “stacked” 1-place-predicates. This is a property of human languages not shared by any known other system of communication. (An exception may be the structure of DNA.) We say that human language is “recursive”. That means that we may apply an operation to its own output, thus (in principle) to infinity. This is the core property of the syntax of human languages, and it mirrors in a way the semantic structure of complex predicates. Thus syntax is a (mental) structure of the combination of words. From this structure, the semantics of a complex utterance can be computed. This is called the principle of compositionality, which was assumed by Frege: The meaning of complex forms derives from the meanings of its simple parts and the manner of their composition.

In this respect, a complex predicate like”(John) loves (Mary)” may be structured as followed: There is a 1-place predicate of the form LOVE (Mary) which in itself is a 1-place predicate for the subject, John: [[ LOVE (Mary)] ( John) ]. The most deeply embedded predicate [ LOVE (y)] is true for all individuals (y) which are being loved (by someone), and [[ LOVE (Mary) ] (x) ] is the set of all individuals (x) which have the property of loving Mary.

In a similar vein, a 3-place-predicate like John gives money to the poor may be decomposed into something like [[[ GIVE (money)] (to the poor)] (John) ].

Note that this decomposition reflects the fact that GIVE is a predicate which consists of a basic state (John owns or possesses the money), a change of this state, i.e. a process (later on, the poor will possess the money) and someone who Causes this process, namely John. You may look this up in an earlier post of mine called “What do we talk about?”.

Completely in the vein of compositionality, the internal semantic structure of these predicates is reflected in the syntactic structure in which they occur. This is in fact true of most linguistic expressions, but there are some complications and (apparent) deviations – otherwise linguists wouldn’t have much to work on. A very prominent example is given in the apparently similar sentences

John is easy to please. (It is easy (for someone) to please John)
John is eager to please. (John is eager to please someone).

It is with these cases and others that the so-called transformational grammar is concerned.


Prof. Dr. Jürgen Lenerz, born 1945, was a professor for German Linguistics at the University of Cologne from 1985 until his retirement in 2011. His main interests are in the interaction of syntax, semantics, intonation and information structure in natural languages