Word and Object. W. V. O. Quine, first edition 1960. 2013 MIT Press edition, ISBN 978-0-262-51831-4.

I am taking these notes mainly from an interest in philosophy of language. I am focused most clearly on questions of meaning and, of course, the analytic–synthetic divide. I will freely omit sections of chapters from these notes which are not relevant to these interests — for example, asides on epistemology that don’t seem to affect the main points.

1. Language and truth
2. Translation and meaning

Language and truth

(last edited 14 May 2017)

Executive summary

Linguistic meaning (of sentences) consists of two main sorts of things:¹

1. Behaviorist meaning: Sentences which are directly connected to causal stimuli, and analogies of such sentences: “The floor is wet,” “I see my father.” We learn the meanings of these sentences by observation of causal association, and make them productively apply to new situations by learning (or already knowing?) how to substitute in new words for old ones.
2. Conceptual meaning: Sentences which get their meaning solely from their relation with other sentences (including necessarily some relation to sentences with behaviorist meaning). The meaning of each sentence bottoms out at some point in stimulus-response associations.

Quine describes these two components with a metaphor of a stone arch. Conceptual meaning corresponds to the stones in the middle of the arch. They get their support from their nearby stones, and transitively from the stones on the sides. These side stones are the behaviorist meaning, which are grounded in actual stimulus-response observations.

In Quine’s words (p. 11):²

What comes of the association of sentences with sentences is a vast verbal structure, which, primarily as a whole, is multifariously linked to non-verbal stimulation. These links attach to separate sentences (for each person), but the same sentences are so bound up in turn with one another and with further sentences that the non-verbal attachments themselves may stretch or give way under strain. In an obvious way this structure of interconnected sentences is a single connected fabric including all sciences, and indeed everything we ever say about the world; for the logical truths at least, and no doubt many more commonplace sentences too, are germane to all topics and thus provide connections. However, some middle-sized scrap of theory usually will embody all the connections that are likely to affect our adjudication of a given sentence.

In a rough sentence, this seems like a straightforward (if radical) combination of a theory of causal meaning with some flavor of what I know as inferential role semantics.³

Behaviorist meaning

Quine presents a very strong theory of word learning which smacks of something like behaviorism.

The first examples he gives are of brute causal links between sensory experiences (assigned the tongue-in-cheek label “surface irritations”) and utterances. There are two ways that these “surface irritations” are linked to language:

1. Direct ostension: point to X and say, “that’s an X.”
2. Collocation: “the blue thing” is spoken while some object on the scene is blue

Quine presents the act of speech here as basicaly a stimulus-response process: a speaker sees something red (a “cue”; p. 5) and utters something with the word “red” in response.

As language learners, we first observe one-word sentences of this sort, and eventually have enough utterance->referent data to start building up actual multi-word sentences. Our sentences are productive because we can make analogical substitutions. If I hear someone say “The boat is red,” I can use this same sentence structure to describe a red car by substituting “car” for “boat.”

Beyond stimulus-response: conceptual meaning

Of course, much of daily language does not conform to this stimulus-response setup. How do we explain the fluent production and comprehension of sentences like “That punishment was unjust?”

These conceptual sentences get their meaning from two sources:

1. from their logical and causal relations with other conceptual sentences, and
2. directly or indirectly, from their connection to instances of behaviorist meaning (observations of stimulus-response associations).

Quine suggests that we acquire the meaning of entirely conceptual words such as molecule entirely through analogy with behaviorist meaning.

p. 13:

What makes insensible things intelligibly describable is analogy … consider molecules, which are described as smaller than anything seen. This term “smaller” is initially meaningful to us through some manner of association with such observable contrasts as that of a bee to a bird, a gnat to a bee, or a mote of dust to a gnat. The extrapolation that leads to talk of wholly invisible particles, microbes for example, can be represented as an analogy of relation: microbes are supposed to compare in size to the motes of dust as these do to the bees. …

Translation and meaning

(last edited 25 September 2018)

Here the story of the previous chapter falls apart. Stated formally in the beginning of the chapter:

the infinite totality of sentences of any given speaker’ s language can be so permuted, or mapped onto itself, that (a) the totality of the speaker’s dispositions to verbal behavior remains invariant, and yet (b) the mapping is no mere correlation of sentences with equivalent sentences, in any plausible sense of equivalence however loose. Sentences without number can diverge drastically from their respective correlates, yet the divergences can systematically so offset one another that the overall pattern of associations of sentences with one another and with non-verbal stimulation is preserved. The firmer the direct links of a sentence with non-verbal stimulation, of course, the less that sentence can diverge from its correlate under any such mapping (23–24).

[I won’t recount the radical translation story here.]

The stimulus meaning of a sentence $S$ is a pair $(\Sigma, \sigma)$ joining

• affirmative stimulus meanings $\Sigma$ — conditions (e.g. “ocular irradiation patterns”) under which a listener would be disposed to assent to $S$ holding, and
• negative stimulus meanings $\sigma$ — conditions under which a listener would be disposed dissent to $S$ holding.

$\Sigma$ and $\sigma$ must be exclusive, but are by no means exhaustive.

No problem here with imaginary concepts with no extension, given a sufficiently loose interpretation of “conditions”:

Stimulus meaning has [this] virtue, since there are stimulation patterns that would prompt assent to ‘Unicorn?’ and not to ‘Goblin?’.

A typology of sentences

• Occasion sentences have truth conditions determined by the immediate stimulus conditions: “Gavagai,” “Red,” “His face is dirty.”
• Standing sentences have truth conditions determined by distal stimulus conditions: “The crocuses are out.”

The meanings of occasion sentences are (maybe??) fully identified by their stimulus meanings. But we have to grant that even the most stereotyped occasion sentence may also have truth conditions for a listener which depend on distal stimuli:

Even for such favored occasion sentences as ‘Gavagai’ and ‘Rabbit’, actu- ally, sameness of stimulus meaning has its shortcomings as a synonymy relation. The difficulty is that an informant’ s assent to or dissent from ‘Gavagai?’ can depend excessively on prior collateral information as a supplement to the present prompting stimulus. He may assent on the occasion of nothing better than an ill-glimpsed movement in the grass, because of his earlier observation, unknown to the linguist, of rabbits near the spot (33).