Meanings and Understandings in the History of Ideas

This paper presents a framework of four types of meaning and understanding in the history of political thought and intellectual history. Previous frameworks have overlooked a whole type of meaning – the type often prioritised by political theorists and philosophers. I call this “extended meaning.” Correcting a wrong turn in philosophy of language in the 1950s, I show how extended meaning has robust intellectual founda-tions, and I illustrate its value for textual interpreters. Even historians often need extended meaning, for example to help resolve ambiguous passages. So, the main types of meaning are not alternatives: scholars interested in one kind of meaning still need others. This paper thus celebrates both diversity and unity.


Introduction
Fifty years ago, Quentin Skinner published his seminal essay "Meaning and Understanding in the History of Ideas."1 Skinner's essay implicitly uses one idea of meaning: intended meaning. In a 1972 essay, Skinner covers three ideas: Blau Journal of the Philosophy of History 14 (2020) 232-256 that Mill is wrong"). Second, this and the other kinds of meaning each have their own, different, and important kind of understanding. Third, some scholars already include extended meaning in their categorisations, albeit imperfectly. I build on these efforts. In short, extended meaning is a legitimate kind of meaning with its own kind of understanding, and fits plausibly into typologies of meaning.
Worrying that extended meaning is not a kind of meaning may already reflect the hegemony of intended meaning. Ultimately, though, one can deny that "extended meaning" is a kind of meaning and still accept that we understand texts better or differently by grasping their implications. What we call this matters less than establishing it as a valid and important focus in history of political thought and intellectual history -and not just for philosophers.
Sections 2 and 3 thus offer a fuller typology of meaning and understanding, correcting Paul Grice's wrong turn in the philosophy of language in the 1950s. Section 4 explains extended meaning further. Section 5 justifies its importance in textual interpretation, including for historians. Section 6 shows that the main frameworks of meaning largely overlook extended meaning. Section 7 connects meaning and understanding.
My framework is not intended to be complete.5 And as I explain elsewhere,6 I largely sidestep the work of hermeneuticists. A single paper can only do so much.

The Old Framework
Philosophers of language have been deeply influenced by Paul Grice's account of meaning, in articles from 1957 onwards.7 The details no longer convince but his general stance remains prominent.8 Yet a critical gap in his framework is missed by all commentators, including Skinner, the leading methodologist of the history of ideas.9 Grice distinguishes two families of meaning: natural and non-natural. These names mislead, as we will see, so I use Wayne Davis's terms: evidential and semantic meaning.10 Evidential meaning involves physical signs, for example "clouds mean rain," "spots mean measles," "smoke means fire," and "boulders mean glacial activity."11 Semantic meaning has two main parts, which I call literal meaning (some philosophers of language prefer "word meaning" or, misleadingly, "timeless meaning") and intended meaning (which philosophers of language usually call "speaker meaning" or "utterer meaning" -terms not very apt for written texts).
Much work on the history of political thought involves the distinction between literal and intended meaning. Consider irony, the "textbook case" of "a divergence between what a person says and what she means."12 Taken literally, Defoe's The Shortest-Way with the Dissenters recommends executing dissenters. But Defoe wrote ironically, implicitly advocating toleration, not persecution.13 To conclude that Defoe meant to defend persecution would be to confuse literal and intended meanings, to confuse what the words say and what Defoe meant by them.
Most writers agree that we must not take words literally where their authors had something else in mind. This is abundantly clear in the methodological work of Skinner and Leo Strauss,14 and in their substantive interpretations.15 Recovering intended meaning is also one aim of political philosophers such 10  It is not surprising that such diverse writers agree about recovering intended meaning. We cannot understand the simplest text without such inferences. For example, when Mill mentions "the tyranny of the majority," he clearly means "the tyranny of the majority [of people or citizens in a country]," not "the tyranny of the majority [of anchovies]." But filling such gaps can be hard. When Rousseau writes that a citizen can "obey only himself and remain as free as before,"18 he could mean "and [also]," "and [that is to say]," or "and [thereby]" remain as free as before.19 Each statement implies something different.
Intended meanings need not be precise. If you insult me and I exasperatedly reply "F*** off!," my swearing does not translate to anything specific; it is not identical to "shut up" or "go away," for instance. Much that Rousseau said may have been intentionally or unintentionally ambiguous, contemporaries like Voltaire suspected. Yet no one can always have clear, unambiguous intended meanings: ambiguity is "rampant" in natural languages,20 and perhaps no statement is ever unambiguous.21 In one of the most astonishing passages in the history of political thought, the normally concrete James Madison suggests that God himself could not explain to humans the difference between the legislative, executive and judiciary, because of the indistinctness of the ideas and of language.22 Intended meanings can be unconscious.23 No one is conscious of how they use every word. Yet analysts may still be able to recover intended meanings. For example, anglophone environmental scientists understand "pollution" as "the effects of pollutants," but most of us equate "pollution" and "pollut- ants." Environmental scientists say "smoke causes pollution," we say "smoke is pollution." We may not be conscious of this, but analysts could recover our intended meanings by analysing how the terms are used. The same applies to many terms in historical texts. Obviously, though, one never knows if one has recovered intended meanings correctly. But the task is not hopeless. Even sceptics about truth will defend themselves against outrageous misrepresentations, for example that they defended Nazism ("I never said that!") or that Nazism is what they intended ("I didn't mean that!"). For most statements in most texts in the history of political thought, authors have some intended meaning, more or less ambiguous, more or less conscious -but never perfectly unambiguous or conscious.
Definition of literal meaning: the conventional sense of a term, phrase or proposition. Examples: "ice" means "frozen water." Definition of intended meaning: the sense of the term, phrase or proposition that one intends to communicate. Examples: you say "it's icy," (a) literally, meaning that it is icy (NB literal and intended meaning overlap here); (b) exaggeratingly, merely meaning "it's cold compared to indoors"; or (c) sarcastically, meaning "it's hot."

Evidential Meaning
So, the standard framework covers evidential, literal and intended meaning. Evidential meaning is usually then ignored. But to see what philosophers of language have missed, we must take evidential meaning more seriously. On semantic (literal and intended) meaning, there are tens of thousands of pages; on evidential meaning, probably not even a hundred. Philosophers of language have thus missed its importance for language. Grice did not define evidential meaning, but does relate it to signs.24 For example, "those clouds mean rain" is shorthand for "those clouds are a sign that rain will fall." Davis's account is clearer: to say that P means Q, in an evidential sense, is to say that P indicates or provides evidence of Q.25 So, "boulders mean Blau Journal of the Philosophy of History 14 (2020) 232-256 glacial activity" is shorthand for "boulders are evidence of glacial activity." This notion goes back to Augustine, Ockham and Hobbes.26 Obviously, clouds do not always mean rain. But the meaning of "mean" in that proposition is the same whether or not rain actually falls. "Those clouds mean rain" equals "those clouds indicate that rain will fall." If this turns out to be false, we could reply "no, they didn't mean rain," i.e. "those clouds did not actually indicate rain." "Mean" means "indicate" in both propositions, and that is all my position involves -not that clouds always actually mean rain, that smoke always actually means fire. Consider, likewise, the propositions "2+2=4" and "2+2=5." "Equals" means the same in both cases, but one claim is false. So, when we say "those clouds mean rain," we make an empirical inference which, as with all empirical inferences, is falsifiable/defeasible.
The standard account requires two modifications. The first, more obvious, highlights the importance of evidential meaning for communication. The second, more subtle, opens the door to this key contribution of this paper: extended meaning.
The first modification is to broaden the scope of evidential meaning, beyond natural phenomena alone. Grice is slightly ambiguous here.27 From the way he initially discusses the natural and non-natural "sense" of meaning, it is possible that he did not actually relate it to "nature."28 Indeed, one of his first examples of natural meaning is: "The recent budget means that we shall have a hard year."29 Nonetheless, his other examples all involve nature, and many scholars understandably read him as relating natural meaning to natural phenomena, including Skinner,30 and Davis who rightly rejects the view that signs must be natural.31 "Clouds mean rain" is similar to "being caught means prison," which is social, not natural. Evidential meaning is thus the same as Hirsch's "symptomatic meaning": stumbling and looking shifty means the speaker in Hirsch's example is lying.32 Denaturalising evidential meaning is a significant step. It explains whyunsurprisingly, and importantly -the language of evidential meaning is used by textual interpreters, as in Viroli's comment that "[t]he fact that they wrote 26  in this manner meant that the Signori were well aware that Machiavelli was familiar with rhetoric."33 The second modification is to move beyond signs (Grice) and evidence (Davis). "Being caught means prison" is not really about signs or evidence. True, being imprisoned is both sign and evidence of being caught; but that does not seem to capture the sense of "being caught means prison." I think we are saying something about consequences -specifically, empirical consequences.34 Prison is a consequence of being caught; boulders are consequences of glacial activity; looking shifty is a consequence of lying. (Not a necessary consequence, obviously; but again, signs and evidence can mislead.) Definition of evidential meaning: to say that P means Q is to say that if P happens, Q happens. Examples: clouds mean rain, boulders mean glacial activity, looking shifty means I'm lying, being caught means prison.

Extended Meaning
Rethinking evidential meaning has small but surprisingly far-reaching effects.
Including non-natural phenomena explains why textual interpreters can and do talk about meaning in an evidential sense. But addressing consequences, not signs and evidence, is far more important: this points to a largely overlooked idea of meaning, which I call "extended meaning." Whereas evidential meaning involves empirical consequences, extended meaning involves logical consequences. Imagine that I say "Paris is the capital of Germany," and that I use the conventional understanding of these terms. But Paris is not the capital of Germany. This means that I am wrong. Or, if someone is an unmarried man, this means that he is a bachelor.
I call this idea "extended meaning," partly to contrast it with "intended meaning," and partly because it involves extending things to see what follows from them. Extended meaning is equivalent to a standard deductive inference: in a syllogism which starts "all men are mortals, Socrates is a man," we could conclude "therefore Socrates is mortal," or "it follows that Socrates is mortal," or "this means that Socrates is mortal." Definition of extended meaning: to say that P means Q is to say that P logically implies Q. Examples: if I say that Paris is the capital of Germany, but Paris is not actually the capital of Germany, this means that I am wrong.
Claims about extended meaning may be wrong, as with claims about intended and evidential meaning (e.g. "this error means Hegel's argument is contradictory"; "when Rousseau says 'moral liberty' he means 'autonomy'"; "this means Nietzsche has gone mad"). But the ideas of extended, intended and evidential meaning still apply: we understand each proposition as involving particular ideas of meaning, just as we understand the objectively false statements "2+2=5" or "Caesar murdered Foucault." My paper contrasts different ideas of meaning, but I do not pretend that all claims about meaning are right.
Ideally, textual interpreters should talk of evidential and extended meaning only where our claims are right. Of course, we never know if claims about evidential meaning are right: there is no certain knowledge about empirical consequences. Textual interpreters should ideally indicate degrees of uncertainty,35 e.g. "this probably means that Dewey influenced Rawls." We can know if some claims about extended meaning are right, provided our terms are clearly defined and our deductive inferences are correct. But I have no problem with people talking of extended meanings more loosely, e.g. "Bentham's errors mean we should ignore him." Such claims are only claims and are always contestable. So, philosophers of language may prefer us to use the language of evidential meaning (and, presumably, extended meaning) only when it is objectively correct.36 But I see no harm in using these ideas more subjectively.
Some philosophers dislike talking of "logical consequences," but the Stanford Encyclopedia of Philosophy uses this term too.37 I thus propose that evidential and extended meaning are subsets of what I will call "consequential" meaning, involving empirical and logical consequences respectively. I now seek to show two things: academics already use the language of extended meaning, and the idea thus merits theorising. This helps deflect the view that extended meaning is not really a kind of meaning -a view which reveals the hegemony of other kinds of meaning, especially intended meaning. I challenge this hegemony, without for a moment implying that intended meaning is not extremely important.38 Extended meaning is very common in everyday English. Unsurprisingly, academics often talk this way too. For example, Machiavelli's defence of political expediency is often depicted as highly original, but as Strauss notes, some politicians had previously grasped it. "This means that his achievement … cannot be understood in terms of politics itself … but only in terms of political thought."39 In other words, it follows logically that Machiavelli's contribution was only original in political theory, not in the practice of politics.
Again, the language of extended meaning is widespread in our field.40 In short, textual interpreters already use the term "mean" in different respects, including intended, evidential and extended meaning. These are different ideas of meaning. Hence my paper.
The Machiavelli example helps me address two further issues. First, is this a meaning of Machiavelli's text? I avoid this phrasing in general, partly due to doubts about whether long and complex texts like Leviathan or The Social Contact have "a meaning" (as opposed to arguments therein do). The term is 38

Blau
Journal of the Philosophy of History 14 (2020) 232-256 particularly unsuitable in this example. The extended meaning Strauss drawsthat Machiavelli is less original than some say -is not really "a meaning of the text." It is a consequence of the overlap between Machiavelli's arguments, previous politicians' actions, and academics' comments. It is a syllogism with more premises than "Socrates is a man, all men are mortal," but the logical deductions are equivalent: "this means that Socrates is mortal," "this means that Machiavelli is less original than some say." This helps address a second issue: is extended meaning about logical or conceptual consequences? Different philosophers I discuss this with see it differently; as far as I can tell, there is no consensus about such matters. Indeed, in practice it amounts to the same thing. But syllogisms can conclude with "therefore," or "we can thus deduce," or "this means"; it seems more natural to depict these as logical consequences. And the Machiavelli example is not exactly about conceptual consequences: the conclusion reflects Machiavelli's ideas, previous politicians' actions, and academics' comments on Machiavelli's originality.
So, we do seem to use the language of what I call extended meaning, which I have theorised in terms of logical consequences, similar to evidential meaning's empirical consequences. Unfortunately, modern philosophers of language have largely overlooked extended meaning. I suspect this partly reflects Grice's wrong turn, relating "natural" meaning almost wholly to natural signs, and relating communication only to literal and intended meaning. Philosophers of language can reply, rightly, that the extended meaning of a proposition is not part of its linguistic content. But nor is its sibling evidential meaning -which is regularly depicted as a kind of meaning. So, we often talk of extended meaning, in everyday speech and in academic contexts, and the kind of understanding it generates matters hugely for textual interpreters.
In summary, we can distinguish two families of meaning: semantic and consequential. Semantic meaning (Grice's "non-natural" meaning) has two types: literal meaning (or "word" meaning) and intended meaning (or "speaker"/ "utterer" meaning). In consequential meaning (my term), P means Q if Q obtains when P obtains. There are two kinds of consequential meaning: extended and evidential. Extended meaning involves logical consequences: P means Q if P logically implies Q. Evidential meaning (Grice's "natural" meaning) involves empirical consequences: P means Q if Q happens when P happens. (If extended meaning involves conceptual not logical consequences, we might be inclined to classify it within semantic meaning. But that clashes with the strong parallel between extended and evidential meaning, involving different kinds of consequences.)

Extended Meaning in the History of Political Thought
I now explain why extended meaning matters so much for textual interpreters. I start with a substantive example before illustrating nine different types of extended meaning. Consider Hobbes's comment that "the Thoughts, are to the Desires, as Scouts, and Spies, to range abroad, and find the way to the things desired."41 I have found about 20 political theorists, philosophers, political scientists and others who read this anachronistically -as a Humean idea of passion enslaving reason, or suchlike -but only one historian who commits this error.42 Let us assume that linguistic and political contexts do not settle this. We could look at similar comments, such as "From Desire, ariseth the Thought of some means we have seen produce the like of that which we ayme at."43 Given the past tense -a means "we have seen" -it seems that "memory" makes more sense than "reason" here. True, memory provides material for reasoning, but in Hobbes's terms they are fundamentally different faculties.44 Even "ability to compute," another Hobbesian synonym of "reason," does not really work.
Moreover, if the "Scouts and Spies" passage is about reason, why would Hobbes have just written "But of Reason and Science, I have already spoken in the fifth and sixth Chapters"?45 This sounds like he will say no more of reason in this chapter; and indeed the next paragraphs do not involve reason. Returning to reason in the "Scouts and Spies" passage would be odd: Hobbes does not usually jump about so disjointedly.
To cut a longer story short, Hobbes seems not to be discussing reason here. This example thus illustrates how we assess ambiguous passages. My process was philosophical, but not very philosophical -no twenty-first century logical tools, for instance. I used a complicated example, but "plugging in" different meanings and testing their consistency is a fundamental and almost universal Let us now systematise this. Extended meaning involves three main things: consequences, consistency, and correctness: (a) logical consequences, e.g. "an author says P, and P implies Q"; (b) consistency and inconsistency, e.g. "an author believes F and G, but these are inconsistent"; (c) correctness and error, e.g. "an author argues H, but this is wrong." Strictly speaking, (b) and (c) are subsets of (a). (For example, the third example might be spelled out as "the author says H exists, but H does not exist; this means that the author is wrong.") Nonetheless, it helps to think of these three things as different questions when analysing a text. What are its implications? How consistent is it, within itself and with other texts? Which parts of it work well and which less so? I will illustrate these categories with three

5.1
Implication Johnny's theory states that legitimate authority requires consent. But many married women have not consented to their husbands. This means that on Johnny's theory these men do not have legitimate authority over their wives.
Note the wording: this means that on Johnny's theory these men do not have legitimate authority over their wives -Johnny may or may not mean this. Extended and intended meaning are analytically separate: given a statement's intended meaning, its implications are the same whether or not the author sees them. On Johnny's theory, husbands do not have legitimate authority over their wives, whether or not Johnny (intendedly) means this. Even if Johnny explicitly rejected this conclusion, the extended meaning would be the same.

5.1.1
Factual Implication Carla argues that revolutions only happen in developed societies. If correct, her argument means that revolutions do not happen in undeveloped societies. Again, note the wording: her argument means this (extendedly), whether or not she means this (intendedly).

5.1.2
Conceptual Implication ("Conceptual Redescription") Tom and Jerry define liberty as absence of external constraints. This means that their idea of liberty is a "negative" one. It does not matter if this term was invented later: what Tom and Jerry say is equivalent to this notion.
In effect, the syllogism is: (a) negative liberty has certain criteria, (b) Tom and Jerry's idea of liberty meets those criteria, therefore (c) Tom and Jerry's idea of liberty is a negative one.

5.2
Inconsistency Stewart says we should only exert power over someone to stop him damaging others. But he later says we can exert power over someone who violates good manners, even when this does not damage others. This means that Stewart is being inconsistent.
Yusef believes in God, and argues that God creates all value. He also writes that it would be wrong for God to tell one person to kill another. The second Blau Journal of the Philosophy of History 14 (2020) 232-256 idea contradicts the first, indirectly. In effect, the syllogism is: (a) God creates all value, (b) therefore what God says is good is good; but (c) God would be wrong to tell one person to kill another, therefore (d) there is some external standard of value, therefore (e) God does not create all value. Statements (a) and (e) are inconsistent.
It is unusual for one statement to be contradictory ("it is raining and not raining"), easy for two or more statements to be contradictory (as with Stewart), and very easy for the logical implications of a statement to contradict another (as with Yusef). An argument can even imply the opposite of what the author thinks. We understand texts better when we spot such inconsistencies.

5.2.1
Factual Inconsistency J.J. says all Europeans are corrupt. Later he says the Sardinians, who are Europeans, are not corrupt. This means J.J. is inconsistent. Perhaps he changed his mind; perhaps at least one comment was insincere. Either way, the two descriptions are inconsistent.

5.2.2
Conceptual Inconsistency Stewart writes a lot about harm, but his comments involve several different ideas of harm which are not fully compatible. This means he was inconsistent in how he talked about harm. We would draw the same conclusion whether or not he spotted this.

5.3
Error To understand group fairness, Patti analyses individual fairness, reasoning that fairness must be the same for both. But this is fallacious: fairness might differ in the two situations. Patti's conclusions could be right but they do not follow from her premises. It does not matter that we are comparing Patti's arguments to epistemological standards developed after she died, by her student Harry; Patti's conclusion is logically invalid. This is different to Stewart's and Yusef's inconsistencies, which directly or indirectly imply both P and not-P. Patti's statements, though, are compared not to each other but to external standards of logic. A writer can be completely consistent but logically invalid.

5.3.1
Factual Error Nicki's history of Rome clashes with reality. This means that her descriptions are factually wrong, whether or not she knew this. Just as we can find logical invalidities by comparing logical claims to external standards, so we can find empirical falsities by comparing factual claims to external standards -here, (perceived) reality.
Factual errors can arise out of factual implications. Carla's argument, above, implies that revolutions will not happen in undeveloped societies; if revolution happens there, this means her argument is factually wrong.

5.3.2
Conceptual Error Hal implies that any amount of corruption prevents liberty. Dave rightly responds that the relationship is a matter of degree, and some degree of corruption is compatible with liberty. This means that Hal makes two conceptual errors, or at least, two conceptual over-simplifications. He treats something as all-or-nothing when it is best seen as a matter of degree, and he ignores the threshold below which corruption does not prevent liberty. Perhaps Hal intentionally simplified the issue to make his point, perhaps he was just imprecise. Whether we call this an error or an oversimplification, his conceptualisation is inadequate.

5.4
Summary These nine types of extended meaning all involve legitimate uses of the term "meaning" and all potentially advance our understanding of historical texts. Frameworks of meaning should thus include extended meaning.

6
The New Framework and Its Competitors I will now suggest that this framework's inclusion of extended meaning makes it more useful for textual interpreters than the main alternatives, those of Quentin Skinner, Leo Strauss, E.D. Hirsch, A.P. Martinich, and Mark Bevir. I also compare extended meaning to similar ideas implied or discussed by John Plamenatz, Knud Haakonssen, Gad Prudovsky, and Morton White. Skinner presents three core ideas of meaning, which he calls meaning 1 , meaning 2 and meaning 3 . This terminology risks confusion, but Skinner is describing literal, experiential and intended meaning, respectively.49 Literal and intended meaning have been discussed above. Experiential meaning involves what a text means to me.50 For example, a song may mean something special to a couple who shared their first kiss to it. Experiential meaning is rarely discussed in intellectual history.
Elsewhere Skinner touches on extended meaning -but only very briefly:

Blau
Journal of the Philosophy of History 14 (2020) 232-256 It would certainly be amazing if all the meanings, implications, connotations and resonances that an ingenious interpreter might legitimately claim to find in a text could in turn be shown to reflect its author's intentions at every point. And it would be a straightforward mistake to infer that, if we came upon some such obviously unintended element, we should have to exclude it from an account of the meaning of the text.51 This important passage bears scrutiny. Skinner is assuming that these interpretations are legitimate, rather than wilful inventions. In other words, words do have implications and connotations that their authors may not intend. Skinner initially seems to differentiate implications and connotations from meanings. But he then includes all of these in meaning -the meaning of the text. As noted earlier, I phrase things differently. More importantly, intended implications can be treated in the same way, not just unintended implications: it would be odd to see implications of arguments as different kinds of meaning depending on whether authors intended them or not. Alas, Skinner's other methodological writings do not replicate this insightful passage. His classic article "Meaning and Understanding in the History of Ideas"52 only hints at extended meaning when he berates Strauss's condemnation of Machiavelli's alleged immorality: Strauss "does not hesitate to assume that such a tone of denunciation is appropriate to his stated aim of trying to 'understand' Machiavelli's works."53 I am no fan of Strauss,54 but I would defend him here: if Strauss's value-judgement is right, it certainly helps us "understand" Machiavelli's works. Skinner himself passes value-judgements on Machiavelli. 55 In his important critique of John Plamenatz, Skinner runs up against extended meaning but shies away from it. Plamenatz defends a more philosophical than contextual approach to texts. He asserts, misleadingly, that for contextualists, understanding Hobbes simply involves understanding why he wrote what he wrote, when in fact, he replies, "[w]e need understand only the sense in which he is using words."56 Skinner actually recommends both, however.57 But what matters most is that Plamenatz seems to be reaching for what I call extended meaning: we are likelier to get the sense in which Hobbes uses words "by a close study of his argument than by looking at … political controversies in his day…. [We] learn more about [thinkers'] arguments by weighing them over and over again than by extending our knowledge of the circumstances in which they wrote."58 This is rather vague: what does "a close study" and "weighing" of arguments involve? Presumably Plamenatz is thinking of some aspect(s) of extended meaning. But his imprecision invites misreading, and Skinner understandably treats Plamenatz as a pure textualist, advocating reading texts "over and over."59 This cannot be Plamenatz's point, not least because it is not what his book does. 60 Skinner is of course right: reading texts over and over may not tell us what authors mean. But there is more to meaning and understanding than this, and we do understand authors better if we spot extended meanings.
The Skinner-Plamenatz disagreement -a foundational methodological debate in our field -thus missed a huge opportunity, because the idea of extended meaning had not been theorised. Plamenatz mischaracterises contextualism and his preferred approach is unclear. Skinner in turn misreads Plamenatz. Yet Plamenatz is feeling for the idea I am calling extended meaning. If only this idea had been recognised all along.
Elsewhere, Skinner notes, rightly, that "no one can be said fully to understand Machiavelli's text" who does not see that it was an attack on humanist advice-books to princes.61 But equally, no one could be said fully to understand Machiavelli's text who does not see its originality and cunning, its ambiguities and errors, its consistencies and inconsistencies, and so on. (We never fully understand texts, of course.) For example, we understand Machiavelli differently if we think that he talks of virtù with considerable consistency, as Skinner himself argues,62 than if we think the term is used inconsistently, as with Mill on "harm," say.
Indeed, the two ideas of meaning may go hand-in-hand. Chapters 8 and 18 of The Prince seem contradictory as regards whether certain extreme actions have virtù. Different scholars read Machiavelli very differently depending on Blau Journal of the Philosophy of History 14 (2020) 232-256 how they address this. But first one must notice the apparent contradiction, then think it through.
Interestingly, there is a remarkable disjuncture between Skinner's methodological and substantive writings. On the one hand, Skinner's account of meaning and understanding largely overlooks extended meaning, with only two sentences on it in his methodological writings, to my knowledge. Skinner also mischaracterises Plamenatz's admittedly hazy description of the idea.
On the other hand, extended meaning is highly significant in Skinner's substantive interpretations. For example, his penetrating analysis of Hobbes's changing account of liberty shows serious problems in Hobbes's early definitions of liberty, and examines how Hobbes gradually overcame these difficulties. 63 Skinner writes not just as a historian, seeking Hobbes's motivations and intended meanings, but also as a philosopher, showing Hobbes struggling with problems which he then gradually tackled. Again, this feels like a missed opportunity: extended meaning is important to the success of arguably the world's greatest intellectual historian, but largely absent from his and others' theorising.
I now turn to Strauss. He distinguishes interpretation, "the attempt to ascertain what the speaker said and how he actually understood what he said," from explanation, "the attempt to ascertain those implications of his statements of which he was unaware."64 Neither term is very apt: conventionally, "interpretation" has a far wider remit and "explanation" means something different. Moreover, we should not focus only on unconscious implications. The two parts of "interpretation" would now be replaced with literal and intended meaning, respectively, and the latter does not just involve the conscious meanings that Strauss focuses on. But suitably revised, Strauss's basic distinction is sensible.
More problematic is the literary theorist E.D. Hirsch's well-known contrast between "meaning" and "significance."65 Hirsch limits "meaning" to intended meaning only. This is understandable but, I have argued, too narrow. "Significance," though, is less clear. It involves "a relationship between that meaning and a person, or a conception, or a situation, or indeed anything imaginable."66 I confess to struggling with this and similar comments by Hirsch.
"Significance is always 'meaning-to' ," writes Hirsch.67 This sounds like experiential meaning, as do several earlier comments where Hirsch mentions personal significance.68 Yet the next example sounds very different: "If Milton really was of the devil's party without knowing it, that would be … part of [Paradise Lost's] significance." This sounds like extended meaning: it is what the work amounts to, whether Milton knew it or not. But that seems not to be Hirsch's main emphasis, and he later equates significance with "judgment, value, and cultural connection."69 I am not sure what "cultural connection" means. Overall, Hirsch's widely cited contrast seems problematic.
Martinich covers four types of meaning: literal meaning, communicative meaning (which I call "intended" meaning), intention, and significance. 70 Martinich treats significance as close to evidential meaning,71 and is far clearer about it than Hirsch. However, he overlooks extended meaning -despite being a superb analyst of it.72 He mentions logical implications as part of "the total content of what a speaker means," whereby "a speaker means many of the things that are logically implied by what he expresses in virtue of the meaning of the words used."73 But my notion is broader: extended meaning is not itself about what a speaker means but what her comments mean -and a speaker may not intendedly mean many things that her comments logically imply. This is where the history of political thought often gets particularly fun! Bevir's distinctions74 include hermeneutic versus structural meaning. Hermeneutic meaning is similar to intended meaning, with one crucial difference: Bevir addresses "the meaning an utterance has for a particular individual," whether author or reader. 75 This notion of readers constructing meaning is important, although in this paper I keep these ideas separate and restrict intended meaning to authors.
However, Bevir simply defines structural meaning in contrast to hermeneutic meaning, as "any other form of meaning an utterance might bear." I am not sure the term "structural" is very apt, and the definition is too open-ended: it potentially includes, and maybe thus conflates, extended and evidential Blau Journal of the Philosophy of History 14 (2020) 232-256 meaning (perhaps even literal meaning). If Bevir's structural meaning includes extended meaning, I would not describe structural meaning as "irrelevant to historians,"76 as discussed above.
So, extended meaning is barely present in Skinner's frameworks, perhaps present in Hirsch's and Bevir's -but combined with other ideas that are kept separate in my framework -and partly present in Strauss's and Martinich's, although more narrowly than in this paper. Each author clearly has some grasp of the idea, but it is never labelled as a single thing, and is either treated too narrowly or is conflated with other ideas.
Nonetheless, I must keep stressing that despite its problematic status in these frameworks, the idea of extended meaning is widespread. I have already mentioned Skinner and Martinich using it powerfully in their substantive interpretations. Aspects of extended meaning are also explicit in some historical theorising. Prudovsky, defending anachronism, covers one aspect of extended meaning, conceptual implication/redescription: if Galileo's comments are equivalent to inertial mass, we can say that this idea is in Galileo even if he lacked the term. 77 Haakonssen advocates more kinds of extended meaning: by exploring "the logical possibilities in a theory or in a complex of distinctions and problems, the historian can appreciate not only the particular route taken by a past author but also the routes not taken -the logical implications of a theory which were not drawn, the inconsistencies which were not seen, the looseness of distinctions that were taken to be exhaustive."78 Nevertheless, extended meaning still involves more than this. Morton White makes a parallel but different point in arguing that historical analyses "presuppose rather than oppose" logical analyses. White is addressing texts' influences on readers. "If Lenin tells us that reading [Berkeley] led certain Russian idealists to engage in reactionary political behavior, we are entitled to ask what those idealists understood Berkeley to be saying, and this would require asking what they thought the logical consequences of his statement were."79 This is slightly too strong: asking how they understood Berkeley is not the same as thinking through Berkeley's problems, and while the latter probably informs the former, much intellectual history does the former with little emphasis on the latter. My focus is different: understanding Berkeley's texts (not their influence on readers) involves grasping intended and extended meanings. Both projects matter, and overlap in practice -perhaps even more so than for White's examples.
To my knowledge, the idea of extended meaning has not been characterised as a form of meaning since 1923, when Ogden and Richards spent just one paragraph on it in a 25-page account of 15 types of meaning.80 Nearly a century later, this idea of meaning should be emphasised again, and emphasised more strongly -especially given its importance to understanding, as I now argue.

7
Meaning and Understanding Meaning and understanding are correlative: each type of meaning generates its own type of understanding. Literal meaning lets us understand the words being used. Intended meaning lets us understand how authors meant those words. Evidential meaning lets us understand causal consequences. Extended meaning lets us understand logical consequences, including inconsistencies and errors. Consider again "Caesar murdered Foucault," an example from section 4. If I say this sincerely, you can understand my intended meaning, despite my obvious error. And if you see the error, you understand the proposition differently. Each kind of meaning generates its own kind of understanding.
Skinner denies that meaning and understanding are correlative, explicitly in his original essay on meaning and understanding, implicitly in the revised version.81 However, his position reflects a narrow notion of intended meaning. Following J.L. Austin, Skinner differentiates intended meaning from "illocutionary force," i.e. what someone was doing in saying or writing something.82 Consider a police officer who tells skaters "the ice over there is thin."83 The words on their own sound descriptive, but the police officer is presumably also giving a warning. For Skinner and Austin, we cannot understand the police officer only by addressing the meaning of what was said.
However, Austin's speech-act theory is no longer mainstream in philosophy of language. Contemporary philosophers of language include illocutionary another. Some readers of this essay are primarily interested in intended meaning, and for them extended meaning is important insofar as it helps inform intended meaning.91 But they cannot deny that extended meaning is a legitimate kind of understanding in itself, even if it is not the kind of understanding that especially interests them.

Conclusion
Extended meaning is a crucial idea for textual interpreters -and not just political theorists and philosophers. Even the hardest core of hardcore historians will need to think philosophically to some extent.92 Alas, overly narrow typologies of meaning and understanding have obscured this vital principle. Like Skinner, I have rooted my account of meaning and understanding in philosophy of language. This required correcting a serious oversight in philosophy of language dating to Grice, as well as updating J.L. Austin. The philosophy of history, and the history of philosophy, require a close connection between history and philosophy. Theory matters, but it must fit practice. When historical research does not live up to philosophical standards, we might assume that it is the historical research which must change. But this paper has argued that both need to change.