The principle of compositionality in modern semantics

Semantics is the discipline in which the web of relations between language and the world is analysed. In the tradition of analytic philosophy such relations between the expressions of a language and extralinguistic entities - such as individuals, their properties and relations to other individuals, states of affairs, situations, truth values and so on - are called semantic relations. Examples of such relations are: the meaning-relation, the relations of reference, expressing, designation etc. Those entities which bear one or more of these relations to the linguistic expressions are considered to be semantic values of these expressions. A desirable semantic feature of a language L is productivity or finite learnability, that is, that a competent user of L is capable of understanding the meaning of new expressions without having been acquainted with them before. That is why many proponents of the tradition of analytic philosophy such as Gottlob Frege, Rudolf Carnap, Richard Montague, Jon Barwise, John Perry and others have-based the formation of their semantic theories on a certain metatheoretic -principle, namely the principle of compositionality of meaning. It states, that the meaning of a complex expression is uniquely determined by the meanings of its parts. It plays a prominent role, when one asks what kinds of entities semantic values are and what ontological status these entities have. Situation-semantics was developed by Barwise and Perry as an alleged improvement on the usual modeltheoretic semantics for intensional languages as exemplified in Montague`s possible-world-semantics with its reliance on eternal sentences as the primary bearer of meanings. These semantic conceptions (supplemented with their corresponding ontological theories) give opposing answers to the above questions. Moreover, there are further differences between Montague- and situation-semantics to observe: the notions of meaning and compositionality presupposed in the principle of compositionality of meaning by the two approaches are different. Differences can also be observed in the applications of this principle as a basis in the formation of these semantic theories, as well as in its methodological status as a falsifiable versus unfalsifiable principle. In this project different ways of making the following items (i)-(iv) more precise will be investigated: (i) the notion of meaning and further semantic concepts like the notions of reference, intension, designation, (ii) the notion of compositionality, (iii) the principle of compositionality of meaning and (iv) further metatheoretic principles concerning the semantic relations such as substitutivity priniciples. We compare the two conceptions with respect to these items. In order to explain the innovative aspects of situation-semantics our basic questions are: What are the differences between the two conceptions? Are these differences justified? What are the reasons for the development of a new semantics by Barwise and Perry? Is it possible to realize their project within the limits of Montague-semantics? If not, why not? In view of these problems it is desirable to make the items (i)-(iv) more precise and in particular to compare the effects which the metatheoretic principles in items (iii) and (iv) have on the formation and ontological foundations of the competing semantic conceptions (which calls for a situation- theory and a theory of worlds respectively). Finally, we are going to investigate the logical connections between different types of metatheoretic principles, namely those which state properties of the semantic relations and those which determine the construction of the semantic values.

A competent language user has not only learned the meaning of finitely many linguistic ex-pressions of a lexicon but has also mastered the rules how other linguistic expressions are compounded in a grammatically correct manner out of those expressions. Suppose now that he gets confronted with such a compound linguistic expression and, furthermore, that this ex-pression is new to him, that is, he has never heard it before nor even seen. He will neverthe-less understand the meaning of this new expression. Our competent language user has the amazing ability to understand the meaning of potentially infinite many compound linguistic expressions. How is this possible? Usually, the property of the finite learnability of a language is explained by means of the principle of compositionality. It says, that the meaning of a com-pound linguistic expression depends only on the meaning of its linguistic parts and the gram-matical rules according to which these linguistic parts are compounded. In this project three possibilities to understand the principle of compositionality were investi-gated: It can be understood (i) as a compositional thesis (in the sense that the meaning itself is composed out of the meaning of the parts), (ii) as a substitutional thesis or (iii) as a functional thesis. The focus of our investigation were mainly on the discussion of the substitutional the-sis, after which the meaning of a compound linguistic expression doesn`t change, if one re-places in it linguistic parts by other expressions which mean the same. If one understands now by the meaning of a linguistic expression that to what it refers - that is, its referent - and by to mean` to refer to`, then a close connection can be drawn between the compositionality of meaning - in the sense of (ii) - and the extensionality of a declarative statement (in the fol-lowing shorter: statement`). According to this view, an extensional statement can be under-stood as a compound linguistic expression whose referent doesn`t change, if one replaces in it linguistic parts by other expressions which refer to the same. If one supposes, furthermore, that statements refer to truth-values, then an extensional statement can be understood in a dif-ferent way, namely as a compound expression whose truth-value doesn`t change, if one re-places in it linguistic parts by other expressions which refer to the same. We investigate - by way of these two concepts of extensionality - the question what it means that a language is extensional. Furthermore, in our formalization of the slingshot-arguments, the rôle is investi-gated which the principle of compositionality plays as a guiding principle in the search for the entities that statements refer to.