Consider the sentence: 'Every farmer who owns a donkey beats it.' Why does this sentence pose a problem for the standard treatment of pronouns as either bound variables or directly referential expressions?
AThe sentence is grammatically ill-formed, so standard semantics simply doesn't apply
B'It' cannot be a bound variable (nothing in the syntax binds it within the right scope) and cannot be a directly referential expression (there is no particular donkey being referred to), leaving the co-variation inference pattern unexplained
CFirst-order logic lacks quantifiers that can range over animals, so the sentence is undefinable
DThe problem is pragmatic, not semantic — context supplies the missing donkey referent
This is the classic 'donkey sentence.' If 'it' were a bound variable, something in the syntax would need to bind it — but 'a donkey' is inside a relative clause and cannot bind 'it' in the main clause. If 'it' were directly referential (like a name), it would pick out a specific donkey — but the sentence makes a universal claim about all farmer-donkey pairs, not about any particular donkey. The inference pattern (each farmer beats their own donkey) falls through the cracks of both standard analyses, motivating dynamic approaches.
Question 2 Multiple Choice
In dynamic semantics, what is the fundamental difference in how sentence meanings are characterized compared to classical truth-conditional semantics?
ADynamic semantics assigns truth values to individual words rather than to whole sentences
BSentence meanings are relations between input context states and output context states — they transform the discourse context rather than having static truth conditions
CDynamic semantics abandons truth conditions entirely and replaces them with speech act types
DSentence meanings are evaluated relative to the speaker's intentions rather than to discourse context
In classical truth-conditional semantics, a sentence's meaning is a set of possible worlds (the worlds in which the sentence is true) — a static object. In dynamic semantics, meaning is a program or transition function: a sentence takes an input information state, processes it, and yields an updated output state. Indefinites introduce new discourse referents; subsequent sentences can access those referents via pronouns. This shift from static propositions to context-update functions is the central theoretical innovation.
Question 3 True / False
In dynamic semantics, indefinite noun phrases like 'a farmer' function by picking out a specific individual that is already present in the discourse context.
TTrue
FFalse
Answer: False
It is precisely the other way around: indefinite noun phrases *introduce* new discourse referents into the context. Pronouns are what *access* referents that are already available. This asymmetry is fundamental to the dynamic approach. 'A donkey entered the field' introduces a donkey referent; 'It brayed' picks up that referent. If indefinites merely picked out pre-existing referents, dynamic semantics would offer no advantage over classical approaches.
Question 4 True / False
The failure of standard bound-variable and directly-referential analyses to handle donkey sentences was a key motivation for developing dynamic approaches to natural language meaning.
TTrue
FFalse
Answer: True
Donkey sentences were a historically important trigger for developing dynamic semantics (particularly Discourse Representation Theory by Kamp, and Dynamic Predicate Logic by Groenendijk and Stokhof). The bound-variable analysis requires an unattested syntactic binding relationship; the directly-referential analysis requires a specific donkey referent that doesn't exist in a universally quantified statement. Dynamic semantics handles both by making quantifiers introduce referents that can be accessed within their scope — including from subsequent sentences.
Question 5 Short Answer
How does dynamic semantics differ from static truth-conditional semantics in its account of what a sentence means, and why does this difference matter for handling pronouns?
Think about your answer, then reveal below.
Model answer: In static truth-conditional semantics, a sentence's meaning is a fixed truth condition — a proposition that is true or false at possible worlds, independent of what has been said before. In dynamic semantics, a sentence's meaning is a context-update function: a relation that maps an input context (a set of available discourse referents and information about them) to an output context. This matters for pronouns because pronouns depend on what referents have been introduced in prior sentences. By making sentence meaning a context-transformation rather than a static proposition, dynamic semantics can formally represent how later sentences inherit referents introduced by earlier ones.
The practical payoff is that dynamic semantics can compositionally derive the reading of donkey sentences and cross-sentential anaphora within a unified framework, without stipulating special rules for each case. The shift is analogous to treating program meaning as what a subroutine *does* to memory state, rather than as a static value the subroutine *holds*.