Reference and salience

Team: Wojciech Rostworowski, Katarzyna Kuś, Bartosz Maćkiewicz

Students: Alicja Krasnowska, Marta Kasprzyk, Małgorzata Rębisz

Funding: National Science Centre, 2017/27/B/HS1/00130

“Theory of linguistic reference based on the notion of salience”

One important question in the philosophy of language is how words refer to objects in the world. In particular, what kind of relation between a singular term (an expression in the singular, such as the name “Mount Everest” or the description “the present Queen of England”) and an object or person makes that expression represent that thing or person?

In answering this question, most theorists assume that the link between a word and an object in the world is homogeneous, that is, reducible to one particular type of relation. For example, a term’s referent is supposed to be the object that uniquely satisfies certain descriptive conditions associated with the term by competent language users, or the object to which the speaker intends to refer. At the same time, it is often assumed that the nature of the reference relation depends on the type of expression, for instance, whether it is a description, a proper name or a context-dependent demonstrative. Sometimes two or more distinct mechanisms of reference are postulated within a given category of expressions.

The project “Theory of linguistic reference based on the notion of salience” contributes to this discussion by developing a theory according to which the reference relation is heterogeneous for every group of singular expressions and involves broadly understood salience factors. These include being within the speaker’s or interlocutors’ perceptual range, being demonstrated, occupying a particular position in discourse structure, or even satisfying a particular description. At the same time, the proposed theory is intended as a model of reference that is as universal as possible, applying, with some modifications, to different kinds of expressions used for singular reference.

In the project, we investigated various salience factors that may be relevant to reference, primarily using experimental methods. We focused on singular expressions in demonstrative and anaphoric use: complex demonstratives, such as “that black dog”, and pronouns. A central part of our strategy was to show that a demonstrative’s reference in context is not unconditionally determined by the speaker’s intention, contrary to a claim accepted by most theories in the philosophy of language. We used scenario-based questionnaires involving truth-value judgments (TVJ). The results supported our hypothesis: demonstrative reference is not determined by referential intentions, at least not directly, because intentions can be overridden by more objective salience factors, including demonstration and description. At the same time, our studies showed that speakers’ intentions play a role in language users’ referential interpretations and can be more important than an object’s perceptual salience alone. The studies thus generally supported the adequacy of the pluralist or hybrid theories of reference developed in the project. These results were presented in “Against intentionalism: an experimental study on demonstrative reference”. We investigated anaphoric relations and the role of different salience factors in shaping them using self-paced reading and coherence tests within the theoretical framework of Centering. We conducted a series of experiments on the importance of grammatical structure, word order, an object’s discourse status (discourse-old versus discourse-new), and additional description through a subordinate clause. Overall, our results indicated that each of these factors affects the salience ranking of discourse entities and consequently shapes anaphoric relations, with grammatical structure (subject versus object) having the strongest influence. A paper presenting these results has been submitted for publication.

The final stage of the project is to develop a salience-based semantic theory. According to the initial intuition, the referent of a term used in context is, roughly, the object that (i) satisfies the conventional descriptive condition contained in the expression and (ii) occupies the highest position in the salience ranking in that context of use. Developing this intuition, the principal investigator formulated a theory that uses the analysis of anaphoric reference as a universal model of reference, covering demonstrative reference in particular, which was the main focus of the project. On the proposed theory, a demonstrative is understood as semantically dependent on an element (an object or property) that is “already present” in the linguistic or extralinguistic context, analogously to the relation between an anaphor and its antecedent. Semantically interpreting a demonstrative thus proceeds like resolving an anaphor, that is, linking it to an appropriate antecedent. In a typically referential or deictic use, however, the “antecedent” is an element of the extralinguistic context. The choice of antecedent depends on the salience of objects in the context or discourse, involving the factors identified in the empirical studies described above.

Since linguistic theories offer specific tools for analysing anaphora, the proposed approach allows the mechanism of demonstrative reference to be described as precisely as possible. We developed a theory based on Discourse Representation Theory (DRT) and, in part, the binding theory of anaphora. We proposed a version of DRT that represents linguistic and extralinguistic context uniformly and applied this analysis to complex demonstratives. A key development was to incorporate into the anaphora-resolution algorithm a salience ranking that includes factors such as demonstration and extraconversational salience. An integral part of our analysis of complex demonstratives was treating descriptive content (for example, “black dog” in “that black dog”) as carrying a presupposition and analysing these expressions’ different functions in terms of different types of anaphora. The proposed analysis can readily be applied to pronouns and definite descriptions. A paper presenting this stage of the project is in preparation.

Publications arising from the project:
Rostworowski, W., Kuś, K. & Maćkiewicz, B. Against intentionalism: an experimental study on demonstrative reference. Linguist and Philos (2021). https://doi.org/10.1007/s10988-021-09340-z