Michael Devitt and Tomasz Zyglewicz
Affiliation: City University of New York - The Graduate Center, The University of Chicago
Category: Philosophy
Keywords: reference, social kind terms, experimental semantics, description theory, causal-historical theory, reference borrowing
Date: Thursday 4th of September
Time: 15:30
Location: Room 232 (232)
View the full session: Content Determination
Our paper reports on experiments testing the reference determination of social kind terms. Until the revolution in the theory of reference started by Saul Kripke (1980), the “description theory” was the received theory for terms in general. According to this theory, the reference of a term is determined by certain descriptions that competent speakers associate with the term: it refers to whatever is described. This view seemed particularly persuasive for social kind terms. Indeed, ‘bachelor’ may have been the most popular example of a term thought to be covered by the description theory: its reference is determined by its association with ‘adult’, ‘unmarried’ and ‘male’. The description theory of proper names has been tested many times (e.g., Machery et al., 2004, 2009; Sytsma & Livengood, 2011) that of “natural” kind terms, quite a lot (e.g., Nichols et al., 2016; Tobia et al., 2020; Devitt & Porter, 2021; Haukioja et al., 2021) that of implement/tool terms, a couple of times recently (Devitt & Porter, forthcoming; Toorman & Haukioja, forthcoming), but that of social kind terms, not at all. Our paper rectifies that.
The revolution was driven by two of Kripke’s brilliant ideas. The negative idea was an “ignorance and error argument” against description theories of proper names and some “natural” kind terms. Competent users of a term are often too ignorant, even too wrong, about the referent of the term to provide the reference-determining descriptions that the description theory requires. Kripke’s positive idea was that ignorant users “borrow the reference” from others in communication situations. There is a “causal chain” of borrowings going back to the first users who “fixed” the reference of the term. That is the central idea of the “causal-historical” theory of reference.
We think that the description theory of some terms has been refuted experimentally: the description theory of proper names (Domaneschi et al., 2017; Devitt & Porot, 2018); that of biological kind terms (Devitt & Porter, 2024), that of implement/tool terms (Devitt & Porter, forthcoming; Toorman & Haukioja, forthcoming). These refutations have led to conclusions that the causal-historical theory is confirmed. Such a conclusion is appropriate if that theory is taken to be just the causal-historical theory of reference borrowing. However, the conclusion is not appropriate in all cases if that theory is taken to include the idea of causal, not descriptive, fixing.
Our online survey experiment (N = 554) was modeled on those on implement terms by Devitt and Porter (2025). We used truth value judgments to test whether a vignette character, Taylor, can successfully refer with a term despite ignorance or error about its referent. The experiment did not test reference fixing. We tested four terms, ‘auditor’, ‘mediator’, ‘podiatrist’, ‘quilter’. For example, here are the three variants of the vignette for ‘mediator’ (“true description” being the control condition):
The local club has money for a new facility but its board is divided over how to spend it. At several heated board meetings, one faction has argued passionately for a swimming pool, while another has argued passionately for a gym. No resolution is in sight. This prompts a board member to suggest that a mediator be called in to help. Taylor, another member of the board, objects to this suggestion saying, “That's a bad idea!” Members of the board scoff, suspecting that Taylor must be confused about mediators. Taylor responds: “Well, it is true that I don’t know much about them. I’ve never dealt with one myself. But I have heard people talking about them. [True description: I know that they attempt to reconcile people in disagreement./ Ignorance: I don’t know what they do./Error: I know that they claim to be in contact with spirits.] But the key thing is that I’ve heard that mediators are pretty useless. So, I am opposed to our inviting a mediator.” But the key thing is that I’ve heard that auditors cost a lot of money. So, I am opposed to our inviting an auditor.” Participants were asked what particular kind of person Taylor was opposed to inviting: a mediator, some other kind of person, or whether it’s unclear what particular kind of person Taylor is opposed to. These answers were coded as “successful reference”, “reference failure”, and “unclear,” respectively.
We predicted analogous results to Devitt and Porter’s (forthcoming) results on implement terms. Their anti-descriptivist results in ignorance conditions counted strongly against the description theory, thus supporting the theory that implement terms can be borrowed. Our results in the ignorance condition for social terms were similarly anti-descriptivist (73%, β = 1.01, p < 0.001). So, our ignorance results supported the view that social terms can be borrowed.
Devitt and Porter’s results in the error conditions were significantly less anti-descriptivist than those in the ignorance condition but were nonetheless still anti-descriptivist. Ours in the error conditions were not (45%, β = -0.20, p = 0.252). Furthermore, participants in the error condition were much less likely to take Taylor to have successfully referred than participants in the ignorance condition (β = -1.23, p < 0.001).
What to make of the error results? We hypothesize that some participants took Taylor to have misunderstood the target term. Where a term can be borrowed, there is always the possibility that a person in a reference-borrowing situation will misunderstand the term and process it to another term in her lexicon that has a different referent (Devitt, 1981, pp. 137–152). Thus, some participants may have supposed that Taylor, when in a position to borrow ‘mediator’, had wrongly processed the term to her lexical entry for ‘medium’. This explains why some participants took Taylor’s resulting confused ‘mediator’-thoughts not to amount to opposition to inviting mediators. A puzzle remains: Why didn’t more of these participants take it to be unclear what kind of person Taylor was opposed to rather than that she was opposed to some kind of person other than mediators (reference failure)?