Dynamics Greg Kobele Semantics II Winter 2015 1 Introduction Treating pronouns as denoting algorithms for selecting individuals in a context has enabled us to deal with the context-sensitive nature of lingusitic expressions. However, from our current vantage point, there are a number of phenomena which suggest that the context in which a pronoun is interpreted can be modified by a prior utterance. In other words, we want to develop a system in which sentences not only depend on contexts for their meaning, but also update the context passed on to subsequent sentences in a discourse. To motivate this further, consider the sentences below. 1. A boy smiled. 2. He was happy. Sentence 2 in isolation (i.e. in an empty context) is deviant; there is no individual that the pronoun can refer to. However, if immediately preceded by sentence 1, sentence 2 is perfectly interpretable. This is so regardless of the context in which the discourse 1;2 is uttered. Indeed, in an out of the blue context the discourse 1;2 is interpreted as equivalent to A boy smiled and was happy. Following de Groote [2], we will treat discourses as the fundamental units of speech, denoting context-sensitive propositions (functions from contexts to truth values). Sentences then are discourse modifiers; they build new discourses from old. A sentence will denote a function from context-sensitive propositions to context-sensitive propositions. For the moment, we simplify things by ignoring (!) context-sensitivity. We think of T (the proposition true) as a trivial discourse; it is one in which nothing has been said. This trivial discourse will play an important role later on. Discourses are built up, somewhat counter-intuitively on this perspective, from end to start; if [[T ]] is a sentence meaning (a function from discourses to discourses), then [[T ]] (T) is a discourse as well. This process can be iterated; if [[S]] is a sentence meaning, then [[S]] ([[T ]] (T)) is also a discourse, representing the result of first uttering sentence S and then sentence T . Of course, real discourses proceed forwards, not backwards; we do not know what the rest of the discourse is when we hear its first sentence. We can see, 1 however, that, for any discourse D, the result of uttering first S, then T , followed by D will be [[S]] ([[T ]] (D)). Thus, uttering first S then T in an out of the blue context will denote the following discourse modifier. B [[S]] [[T ]] = λx. [[S]] ([[T ]] x) We will interpret initial segments of a discourse in terms of function composition; more poetically, we interpret the period ‘.’ as composing sentences with one another. As was the case when we enriched our semantic values to allow for contextsensitivity, we wish here to have a conservative extension of our framework, which will allow us to ignore dynamic aspects of meaning (as for example when computing the way the meaning of a sentence is related to the meanings of its constituent parts) without affecting the resulting denotation. We do this, as before, by defining operations Dynα and Nydα , for each type α, which map inherently non-dynamic meanings to dynamic ones. The relation between non-dynamic types and dynamic ones is a type homomorphism, which can be represented as follows (writing αd for the type which results from α): Case ed td (αβ)d Output e tt (αd β d ) In order to actually define these operations, we need to decide how sentences influence the meanings of discourses. Although there is a rich literature on just this topic [1, 3], we will drastically simplify things by assuming that a sentence is an intersective discourse modifier. Individuals (things of type e) are unaffected by dynamicization. Dyne (a) = a Dynt (φ) = λψ.φ ∧ ψ Dynαβ (f ) = λA.Dynβ (f (Nydα A)) The above defines how to wrap an inherently static expression in the trappings of dynamism; the result can then be used wherever a truly dynamic expression is expected. The dynamicization of an individual of type e to be itself, the dynamicization of a static proposition of type t to be the function which conjoins it with an upcoming discourse, and the dynamicization of a function of type αβ to be a function from dynamic arguments of type αd to the dynamicization of the result of applying the function to the statification of its argument. 2 Nyde (A) = A Nydt (Φ) = Φ (T) Nydαβ (F ) = λa.Nydβ (F (Dynα a)) Given a dynamic term, we can also turn it into a static one. The fundamental observation is that a dynamic proposition can be turned into a static one simply by applying it to a static proposition. References [1] Nicholas Asher and Alex Lascarides. Logics of Conversation. Cambridge University Press, 2003. [2] Philippe de Groote. Towards a montagovian account of dynamics. In Masayuki Gibson and Jonathan Howell, editors, Proceedings of SALT 16, pages 1–16, 2006. [3] William C. Mann and Sandra A. Thompson. Rhetorical structure theory: Toward a functional theory of text organization. Text, 8(3):243–281, 1988. 3
© Copyright 2026 Paperzz