Carnap's Empiricism, Semantics, and Ontology: Internal vs External Questions
Rudolf Carnap's 1950 distinction between internal questions (within a framework) and external questions (about frameworks), dissolving traditional metaphysical disputes.
PhilosophyAll chapters Reading tools Notes & comments
Questions within and about a framework › What Would Settle the Question?Full chapter · Sections in this chapter1 / 9 ↓
What Would Settle the Question?
Is there a prime number greater than one hundred? The number 101 supplies an answer. Its only possible prime divisors up to its square root are 2, 3, 5 and 7, and none divides it. Now ask whether there are numbers at all. The same calculation may sound beside the point. Someone who grants the proof can still ask whether it commits us to an independent realm of mathematical objects. What exactly are they asking the proof to do?
In his 1950 essay Empiricism, Semantics, and Ontology, §§1–2, Rudolf Carnap hears two questions in this conversation. One uses the expressions of arithmetic to ask whether a number meets a condition. The other seems to ask about the reality of the system of numbers as a whole. The first has a technique for giving an answer. Carnap wants anyone who asks the second to say what a theoretical answer would establish, how it would do so, and how that differs from recommending or declining the language of arithmetic.
His concern is practical even in the setting of the essay. A semanticist who uses propositions or properties wants to write statements that can be understood and assessed. Must that person first defend a metaphysical theory of abstract objects? Carnap argues that they can use this language and still think like an empiricist. The force of the argument depends on keeping the kinds of permission he grants separate.
Learning a Framework
A linguistic framework, in Carnap's sense, gives speakers forms of expression and rules for their use. Adding a few names to an existing vocabulary may leave the larger question untouched. To introduce a framework for numbers, we need to know what counts as a numerical expression, how expressions for properties and relations apply to numbers, and how to make general claims about them. A speaker should be able to say that every number meeting a certain condition meets another condition, or that some number has both of two properties.
The rules have different jobs. Some tell us how terms can go together. Others tell us what follows from a statement, or what observations support it. In arithmetic, deducing a consequence from the accepted rules can establish an answer. In ordinary talk about material things, a question about the contents of a room calls for observation. The framework tells us why looking into the room bears on the question and why calculating the factors of 101 does not. Carnap develops both cases in §2, “The world of things” and “The system of numbers”.
An internal question uses the terms and rules of a framework. Depending on the question, the grounds for an answer may be a deduction, observations, or a combination of available reasoning and evidence. Using a framework leaves room for false statements and for questions we cannot yet settle.
Consider whether a particular library owns a copy of the essay. A catalogue entry is evidence, but it could have survived after the copy was withdrawn. A conversation with the librarian or a search of the shelf might change our answer. Nothing about this uncertainty undermines our grasp of what the question asks. Carnap's account of ordinary object talk in §2 makes room for that familiar distinction between having a method of inquiry and already knowing how the inquiry will turn out.
Two Ways to Hear “There Are Numbers”
Once the numerical framework is in place, the sentence “There are numbers” can express an internal claim. Start with the rule that allows a numeral such as five to be used as a number. Grant the corresponding instance and you can infer that at least one number exists. On Carnap's account, that particular internal existential statement is analytic. He contrasts it with questions brought up in the same framework whose answers require actual work, such as finding a prime above a specified bound. Both varieties belong to the practice of arithmetic. They differ in the work needed to justify an answer. Empiricism, Semantics, and Ontology, §2, “The system of numbers”
Suppose someone agrees that the number-language permits the inference but repeats their question. They may mean to ask whether they should use that language in a theory. That is intelligible. We can ask whether a numerical formulation lets us make precise comparisons, produces useful predictions in an application, or combines well with other commitments of the theory. We can then decide whether to use it. The discussion may appeal to facts about rival formulations. Its practical conclusion recommends a way to proceed.
A third reading is the one Carnap challenges. Here a person claims to ask a theoretical question about the reality or subsistence of the numerical system independently of the framework. Their question sounds like the internal existential one, but an internal proof is set aside as irrelevant. Their question could be reinterpreted as a decision to adopt or reject the language, but the speaker insists they want to discover a fact about what there is. Carnap asks for a formulation that supplies this third question with cognitive content. He argues that the familiar debate has yet to furnish such a formulation. Empiricism, Semantics, and Ontology, §§2–3
The distinction matters because there can be a legitimate dispute about adopting the language without there being a second theoretical fact that settles that choice. It also avoids the implication that every question asked by a philosopher is external. If someone can specify what a claim means and what bears on its answer, Carnap's challenge has done its job even if the question remains difficult.
What Reality Means in the Thing Language
Carnap begins his examples with ordinary things. Imagine asking whether a white piece of paper is on a desk. We can look. If the scene is a reflection, the actual paper might be elsewhere. If a colleague claims the paper was there yesterday, we may compare observations made at different times. Carnap takes a judgment about whether an item or event is real in the scientific sense to involve placing it within a system of things and events, where it fits with what else we have reason to accept.
That account leaves ordinary realism about the paper intact. The disputed question is whether the system of things itself has the reality attributed to an item successfully placed within the system. In his treatment of the traditional dispute between realism and idealism, Carnap argues that this stretches the internal test beyond its domain. Locating an item among others cannot locate the entire system among its own members. The argument depends on his account of what “real” does in scientific thing talk. It would take a different account of the predicate to make the attempted external application intelligible.
Carnap observes that we ordinarily take up the language of things early in life, before deciding whether to keep using it. Carnap treats continuing to use thing language as acceptance in a practical sense. The fact that it communicates our observations effectively may give us a reason to continue. It does not supply a new observational finding concerning the reality of the thing system as a whole. Empiricism, Semantics, and Ontology, §2, “The world of things”
What Accepting a Framework Commits Us To
The essay makes the distinction consequential for semantics. Suppose we add variables into whose places declarative sentences can be substituted, a term for propositions, and rules for reasoning about them. We gain a systematic way to speak about what sentences express. Carnap's example uses a formal rule that makes the assertion that a given declarative content is a proposition analytic within the proposed framework. Consequently the claim that there are propositions has a straightforward internal reading. This does not require picturing a proposition as a peculiar physical object encountered alongside the paper on our desk.
The framework choice and the assertions it licenses still have to be kept apart. Introducing rules for proposition talk allows us to state claims and draw consequences. It does not make all claims formulated with the resources of that language true. Nor does it require any further metaphysical assertion proclaiming the reality of all propositions from outside that practice. Speakers can be explicit about what expressions they are using and why they think those expressions will help. That is the defense Carnap offers against the charge that semantic reference to propositions entails Platonic metaphysics. Empiricism, Semantics, and Ontology, §2, “The system of propositions,” and §4
Carnap also distinguishes a framework's rules from images we might use to teach it. Telling a beginner how to picture an entity might help them learn to use a term. The picture is a poor guide if we infer that every feature we imagined belongs to the theory. In the case of propositions, using the rules may do more to explain how the expressions work than giving them a metaphorical address. That distinction lets a reader ask whether a philosophical dispute concerns the actual use of a theory or only the story someone has told about its vocabulary. Empiricism, Semantics, and Ontology, §2, “The system of propositions”
One Question, Three Different Tasks
Imagine a team measuring travel across a city. It can represent each journey by a start and an end event with spatial and temporal coordinates. We can distinguish three questions it might ask.
First, did two journeys end at the same place and time, within a tolerance specified in the measurement? That is a factual question. The team uses its representation and looks at the observations. Whether the representation is convenient tells it little about the particular journeys.
Second, should the team introduce a coordinate system at all? It may have a less precise method of reporting places and times that is cheaper to use. If the team needs to compare departure and arrival sequences closely, the extra precision may be worth the cost. Evidence about uncertainty, time spent recording measurements, and the performance of past analyses can affect the choice. Different uses of the data may call for different decisions.
Third, do the coordinate points really exist, quite independently of the ways to refer to them? If this just asks whether the coordinate framework includes points, we can read the rules. If it asks whether the team ought to quantify over them in making its measurements, we have returned to its choice of representation. To make it a third theoretical question, we need some account of what would count as its answer. The example adapts Carnap's discussion of introducing spacetime coordinates after a thing language is already in use. The third question here is our way of testing a purported metaphysical claim. In Carnap's own list, the third reading asks whether experience shows the scheme will be useful. He treats that as an empirical question about how well the scheme works. Establishing its usefulness would give us a reason to adopt it. That result would leave the purported metaphysical claim unsettled. Empiricism, Semantics, and Ontology, §§2–3
This example also clarifies a limit of the approach. The team can disagree over whether its observations place the two arrivals within the agreed tolerance. Calling such a dispute a matter of framework choice would be a mistake. Conversely, if the dispute really is over which accuracy and coverage to require, examining the same pair of records again might leave the question exactly where it was.
Where a Critic Can Press
An objector can agree that a physical claim and a decision about using a language play different roles, yet doubt the boundary Carnap draws. Suppose choosing a theory requires asking both how well it works and what its claims commit us to. If the best overall theory refers to numbers, the objector may regard that as a reason to believe in numbers. Why should we describe the same commitment as a practical acceptance on one side and a theoretical one on the other?
Quine presses a related point in “Two Dogmas of Empiricism,” §6. He accepts a pragmatic element in scientific revision, but extends it across the whole scheme of belief. Observations can force adjustments, and more than one adjustment can sometimes accommodate the same experience. A change in how we formulate or organize the theory need not fall on a separate side of a fixed division between choices of language and claims about facts. His discussion challenges the sharpness of the division on which Carnap relies. It does not claim that every sentence is individually refuted or confirmed by a particular observation.
A Carnapian reply begins by separating the roles even when the choices are made together. Someone who chooses to express a theory using a certain system of number terms can then be wrong about a calculation in that system. A separate inquiry into the efficiency of the notation can affect the initial choice. That much can be granted even if the boundary is hard to draw in a large scientific theory. Quine's challenge, in turn, is whether the local distinction between selecting a practice and making a move within it justifies Carnap's stronger diagnosis of philosophical ontology.
Another objection concerns the burden of showing that a question has content. A realist might hold that theoretical questions need not always have an experiment that would answer them, and might propose to justify an ontological claim by how it contributes to an explanation. This is a genuine reply only if it says what explanatory work the claim does and how its work differs from the success of the language used to state it. Carnap can then ask whether those benefits give us grounds for accepting a framework or grounds for asserting something additional. The critic need not assume in advance that those options exhaust the possibilities. This is a place to dispute the argument, rather than simply choosing which definition of reality sounds right.
Reading the Essay with Its Distinction in View
When you come across an existence sentence in the essay, try to identify its use before evaluating the answer. Ask which expressions its speaker is using, what rules or observations are relevant, and what would change if they chose a different vocabulary. If those observations could be different while the vocabulary stays fixed, there is room for a substantive factual dispute. If the issue is which vocabulary gives an inquiry the reach or simplicity it needs, compare the available choices. The second exercise can involve more than taste because the alternative frameworks can be assessed against an explicit aim.
A final question remains. If somebody insists there is a theoretical issue left once all of that is settled, ask them to state it in a way that makes its consequences visible. They might succeed, in which case the original classification needs revisiting. If they cannot say what would depend on a positive rather than a negative answer, the stalemate is no mystery. Both speakers may already have agreed on all the internal claims they can formulate and on how the language is used. The emphasis placed on “really” by itself does not make the disagreement clearer.
Carnap's proposal is useful because it makes the next step in a dispute concrete. One person can request a derivation, another can ask for better observations, and a third can demand an argument for adopting a new way of speaking. Keeping those requests separate preserves empirical inquiry and gives abstract language a place in it. The disputed step concerns what would give a further metaphysical question a determinate meaning.
Further Reading
- Rudolf Carnap, Empiricism, Semantics, and Ontology (1950). Start with the successive examples in §2, then see how §§3–4 apply the distinction to accepting entities and to semantics.
- W. V. O. Quine, “Two Dogmas of Empiricism,” §6 (1951, with revisions indicated in the linked text). Read the closing comparison with Carnap after Quine's account of the relation between experience and a whole theory.
Join the discussionAsk a question or share what you learned.
Keep exploring
Reference
Citation details
Cite or share this article.

Comments
Add to the conversation
A question, a correction, or an example from your own work.
No comments yet. Be the first to share your thoughts!