My research is motivated by a striking asymmetry in linguistic meaning. Some dimensions of meaning are grammatically regimented. The functional elements of language encode relatively rigid distinctions related to individuation, quantity, event structure, temporality, and modality, just to name a few. Other aspects of interpretation are remarkably flexible, emerging through context, inference, and our knowledge of the world. Why does grammar privilege particular dimensions of meaning, and what can this distinction tell us about the representations available to the mind?
I approach these questions by studying grammatically relevant aspects of sentence interpretation from both representational and processing perspectives. Linguistic theories specify the representations and opeartions that grammatical distinctions require, while psycholinguistic theories specify the processes that retrieve, construct and draw inferences over those representations. Their interaction provides a way to ask not only what linguistic meanings look like, but why the cognitive system represents and uses meaning in the ways it does.
Which aspects of meaning are grammatically structured, and what does that structure tell us about our mental representations?
Some dimensions of meaning are remarkably sensitive to grammatical structure. Distinctions involving individuals and kinds, quantity, events and states, thematic relations, time, and modality are not simply matters of what words and phrases mean. Grammatical structure shapes how these meanings are represented and composed. Much of my formal work investigates the particular dimensions of meaning that are grammatically privileged, and asks what their grammatical organization can tell us about the representations that are available to the mind.
A theory of linguistic representation is also a commitment about what the processing system has to construct and manipulate. Representational distinctions motivated by linguistic theory should have consequences for the processes that construct and use interpretations, and evidence about those processes can, in turn, constrain the representations that linguistic theories posit. This interaction between representation and process is central to my approach to grammatically relevant aspects of sentence interpretation.
What does the grammatical representation of thematic relations commit us to about the processes that construct sentence meaning?
How is the kind/object distinction shaped by grammatical representation?
What does the way events are grammatically encoded tell us about how events themselves are represented in the mind?
How does our cognitive system access knowledge to guide expectations about language before the relevant input arrives?
Language comprehension unfolds over time, but comprehenders do not simply wait for each word before acting. Our knowledge of language, context, and the world allows us to form expectations about how upcoming language is likely to proceed. These expectations provide a window onto the knowledge that supports comprehension: what kinds of information can guide prediction, how uncertainty is represented, and how expectations at different levels of representation relate to one another.
My work approaches these questions by treating prediction as a problem of probabilistic inference while asking what cognitively plausible algorithms could actually implement those inferences. Rather than simply demonstrating whether comprehenders predict, my work asks how probabilistic knowledge is represented and how cognitive processes access and make use of that knowledge during real-time comprehension. This provides a way to investigate how stored knowledge, contextual information, and online inference jointly shape the interpretation of language as it unfolds.
How does comprehension construct meaning that goes beyond what is explicitly encoded?
Understanding utterances routinely involves more than composing the meanings explicitly provided by words and grammatical structure. Speakers choose to make certain meanings and their alternatives available, and comprehenders consequently draw inferences and implications, establish relations of coherence, and enrich interpretations in response to context. The question is how these less rigidly encoded aspects of meaning are represented and constrained: what licenses an implicit interpretation, what alternatives are considered, and how an evolving context shapes the relations comprehenders establish among meanings.
My work approaches these questions by investigating how comprehenders generate, maintain, and use information that is not straightforwardly given by the linguistic signal. Alternatives, implicatures, and coherence relations provide different cases in which interpretation depends on information beyond what is explicitly encoded. Studying them together offers a way to ask how flexible, context-sensitive inference interacts with the more structured representations supplied by language.
What is the relationship between the processes that build interpretations and the representational products they ultimately leave behind?
The incremental nature of language comprehension means that linguistic interpretations are continuously constructed and revised as linguistic input unfolds. But the representations that form during comprehension and persist afterwards need not simply correspond to the final linguistic analysis. Provisional interpretations can linger after reanalysis, representations can be interfered with during encoding and retrieval, implicitly constructed information can become part of what is remembered, all while prior knowledge shapes the representations that comprehenders construct. I am interested in what these effects reveal about the relationship between the moment-to-moment processes of comprehension and their representational products.
My work investigates both how information in memory is retrieved and used during comprehension, and how the processes involved in comprehension determine what is ultimately encoded and retained. From this viewpoint, memory is more than a resource that language processing draws upon. Instead, this research provides a way to investigate how the history of processing shapes an interpretation, what aspects of that interpretation persist, and how the resulting representations become available for subsequent interpretation and inference over both shorter and longer stretches of time.