I’ve been doing some reading on the relationship between historical philosophy and contemporary philosophy (much of which also concerns methodological considerations for either or both of those pursuits). While many of the papers I’ve read from the volume acknowledge that knowing contemporary philosophy can enrich one’s historical work, and people I know who work on contemporary philosophy occasionally acknowledge that understanding the views of historical figures can enrich one’s contemporary philosophizing, I wanted to give a relatively straightforward example of a place where it seems extremely likely to me that strong benefits are to be had from both approaching historical/interpretive issues through the lens of certain contemporary debates, and from approaching contemporary issues through the lens of certain historical debates. In this instance, it is the case of meta-ethical expressivism and Locke’s theory of language (and more specifically: problems arising from sentences that are “logically complex” or are “attitude” ascriptions).
Generally speaking, Expressivism is a sort of mentalistic approach to giving the meaning or significance of language. An expressivist semantics for a language L assigns, to each meaningful sentence of L, a mental state, which that sentence is said to express. Roughly, a sentence S expresses a mental state M, when ordinary, sincere, assertive utterances of S are properly related to the speaker’s being in state M (this way of stating it sets aside complications arising from context-sensitivity, and I will continue to ignore those throughout this post).
On the assumption that belief is a propositional attitude, it is relatively easy to translate any traditional truth-conditional semantic theory into a corresponding expressivist theory. For any proposition P and any sentence S, if the truth-conditional theory assigns P to S, there is a correlate expressivist theory which assigns the belief that P to S.
Though this fact about the ability to translate such theories is itself interesting, it primarily helps us demarcate an especially uninteresting sub-class of expressivist semantic theories (i.e. those that are the belief-y correlates of traditional truth-conditional theories). More often, the reason contemporary philosophers adopt an expressivist view of language would be out of a belief in the inadequacy both of traditional truth-conditional semantic theories and of the uninteresting expressivist correlates of such theories for a given portion of language. Consequently, when I describe a view as “expressivism about X”, I mean to be picking out views on which the move to an expressivist framework is intended to capture the way in which X-type sentences do not simply express belief in the-proposition-traditional-truth-conditional-semantics-would-have-assigned, but, instead, express some other attitude.
So, non-cognitivists in metaethics who adopt an expressivist approach to theorizing about language do so because they think (many) sentences with moral/normative terms express states that are importantly distinct from the beliefs expressed by descriptive sentences. Typically, they regard moral thought as involving states that feature some desire-like phenomena (broadly construed), and thus, think that moral/normative sentences express some non-cognitive mental states, rather than ordinary belief (or perhaps, in addition to expressing some ordinary belief). Not all expressivist positions have to be strictly non-cognitivist, however. One version of expressivism about epistemic modals is to treat “Must(P)” as differing from “P” not in terms of which proposition is believed when one utters the sentences sincerely, but, instead, as differing in the amount of confidence in P required for sincere utterance, and thus, treat “Must(P)” as expressing highly confident belief in P, while “P” merely expresses belief in P (more likely, such an account would be done in terms of subjective credences, rather than all-out belief and different levels of confidence, but the point stands).
In defense of the claim that a little anachronism can provide benefits for history, I am going to cherry-pick the example of Locke, who was helpfully explicit about the relationship of thought and language. While it is clearly somewhat anachronistic, it is relatively easy to read expressivist positions back into someone like Locke, due to his view that the role of language is principally to publicly manifest one’s mental life:
Locke provides a nice example of this (selection from Locke’s Essay, 4.VII, “Of Particles”):
The Mind, in communicating its thought to others, does not only need signs of the Ideas it has then before it, but others also, to shew or intimate some particular action of its own, at that time, relating to those Ideas. This it does several ways; as, Is, and Is not, are the general marks of the Mind, affirming or denying. But besides affirmation, or negation, without which, there is in Words no Truth or Falshood, the Mind does, in declaring its Sentiments to others, connect, not only the parts of Propositions, but whole Sentences one to another, with their several Relations and Dependencies, to make a coherent Discourse.
Neither is it enough, for the explaining of these Words, to render them, as is usually in Dictionaries, by Words of another Tongue which came nearest to their signification: For what is meant by them, is commonly as hard to be understood in one, as another Language. They are all marks of some Action, or Intimation of the Mind; and therefore to understand them rightly, the several views, postures, stands, turns, limitations, and exceptions, and several other Thoughts of the Mind, for which we have either none, or very deficient Names, are diligently to be studied. Of these, there are a great variety, much exceeding the number of Particles, that most Languages have, to express them by: and therefore it is not to be wondred, that most of these Particles have divers, and sometimes almost opposite significations.
Locke goes on to give a quasi-treament of the various actions of the mind that can be signified by the term “but”, which I won’t get into here.
While there are obviously some lurking interpretive questions, it is at least natural to come away from passages like this with the thought that Locke thinks a) the basic account of the meaning of simple declarative/descriptive sentences is to be given in terms of the judgments they signify, and that b) a complete account of linguistic significance will invoke actions other affirmation and denial (as attested by the case of particles other than the copula, such as “but”).
Note that this commitment places Locke’s view fairly squarely within a framework fitting the definition of “expressivism” given above, and further, that the view of particles like “but” would clearly render it what I was thinking of as an interesting version of expressivism.
In fact, given i) his claim that we signify the connections among “not only the parts of Propositions, but whole Sentences one to another”, and ii) his use of a sentential connective like “but” as a central example of this phenomenon, one would expect him to be an expressivist about other connectives, such as “if…then…” (and thus for the conditional sentences containing them). Unfortunately, the chapter on particles is only about 2 pages long, and doesn’t contain (or even suggest, really) all that much in the way of guidance on how we could systematically construct the rest of his theory of language.
At the outset of this post, I said I was going to illustrate how projects in historical philosophy can benefit from attention to contemporary philosophy as well as how projects in contemporary philosophy can benefit from attention to the history of philosophy. So far, I have, at best, shown that we can draw some connections between historical and contemporary issues. Now I’d like to explain how such connections can produce the benefits I’ve indicated:
At the New England Colloquium on Early Modern Philosophy, Jennifer Smalligan Marusic presented a paper investigating issues arising for Locke and Arnauld’s theories of judgment, on which propositions are formed in the mind by way of acts of judgment (acts of affirmation and denial). One concern surrounding such judgment-based theories of proposition-formation is that they appear to rule out merely entertaining a proposition (or, at least, merely entertaining a proposition without having previously judged it). Another natural concern is that, if “is” signifies affirmation, then we get surprising predictions about the judgments possessed by someone competently uttering disjunctions (“Either Susan is smartest or Nancy is smartest”), externally negations (“it is not the case that Philip is loud”), conditionals (“if Carol is happy, then Lisa is happy”), or attitude ascriptions (“Beth thinks Lupe is smartest”, “Tom hopes that Alejandro is home”).
If “is” signifies that the speaker is doing some affirming of the ideas signified by the terms/phrases flanking the “is”, then the sentences just mentioned should all signify affirmations which they do not, in fact, seem to signify. For instance, sincerely uttering “Beth thinks Lupe is smartest” doesn’t seem to involve the speaker judging Lupe to be smartest, and “Tom hopes that Alejandro is home” doesn’t signify that anyone judges that Alejandro is home).
Some have (in the face of such worries) opted for proto-Fregean readings of Lockean judgment, in which “affirmation” is forming-but-not-necessarily-endorsing an affirmative proposition. In her talk (which was excellent), Marusic made a compelling case against such proto-Fregean interpretations of Locke and Arnauld and in favor of the reading which generates these worries, while also exploring ways that Locke and Arnauld, on her preferred interpretation, might be able to resolve the worries.
My point in discussing this may already be apparent to those familiar with non-cognitivist accounts of moral terms, who no doubt recognize in this challenge, a belief-based cousin to the Frege-Geach problem for emotivism about, say, “good” (I am here referring to the version of the problem presented by Geach and later by Searle, as described on pp. 705-7 of Mark Schroeder’s paper).
If, as those views maintain, the moral sense of “good” indicates/expresses commendation, then something (though it is unclear what) should be commended in an utterance of the sentence “if giving money to charity is good, then giving money to Oxfam is good”. But, prima facie, one could endorse the conditional without commending charitable donation (generally) or donations to Oxfam (specifically).
I won’t take the time here to carefully detail why I think the parallel is more than just superficial similarity, but I will observe that both problems arise because a claim about what a term is doing appears to work well in certain basic/ground-level sentential contexts, but does not seem to generalize well across all possible occurrences of the term, and seems to fail, in particular, with respect to occurrences of the term either in “logically complex” sentences or inside the “scope” of an “attitude ascription”. As this post is getting quite long, I will attempt to briefly wrap up, and indicate the potential benefits:
If the problem faced by Locke’s view of judgment and language is as similar to the problem faced by contemporary meta-ethical expressivism as I have suggested, we can learn quite a bit about the individual problems by observing the similarities and differences they exhibit when it comes to evaluating proposed solutions. If a proposal looks promising in the one case, but not the other, that would be useful information for the people working to solve either one individually. If some proposals look equally promising for both problems, that is also useful information. If the approach to language embodied in Locke’s view can avoid the problems (without simply invoking Fregean or proto-Fregean resources), we might find that working within the range of currently dominant frameworks (which were heavily influenced by Frege and Russell’s approaches to philosophy of language) unduly constrains various expressivist projects. Solving the Frege-Geach problem is one thing, but it is an additional constraint to do so while preserving a particular story about why there is no Frege-Geach problem for the descriptive case.