Song Page - Lyrify.me

Lyrify.me

Word and Object Ch 2 - Translation and Meaning §7 - §14 by Willard V. O. Quine Lyrics

Genre: misc | Year: 1960

Chapter Two: Translation and Meaning

§7. First steps of radical translation

We have been reflecting in a general way on how surface irritations generate, though language, one's knowledge of the world. One is taught so to associate words with words and other stimulations that there emerges something recognizable as talk of things, and not to be distinguished from truth about the world. The voluminous and intricately structured talk that comes out bears little evident correspondencе to the past and present barragе of non-verbal stimulation; yet it is to such stimulation that we must look for whatever empirical content there may be. In this chapter we shall consider how much of language can be made sense of in terms of its stimulus conditions, and what scope this leaves for empirically unconditioned variation in one's conceptual scheme.

A first uncritical way of picturing this scope for empirically un-conditioned variation is as follows: two men could be just alike in all their dispositions to verbal behavior under all possible sensory stimulations, and yet the meanings or ideas expressed in their identically triggered and identically sounded utterances could diverge radically, for the two men, in a wide range of cases. To put the matter thus invites, however, the charge of meaninglessness: one may protest that a distinction of meaning unreflected in the totality of dispositions to verbal behavior is a distinction without a difference.

Sense can be made of the point by recasting it as follows: 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.

The same point can be put less abstractly and more realistically by switching to translation. The thesis is then this: manuals for translating one language into another can be set up in divergent ways, all compatible with the totality of speech dispositions, yet incompatible with one another. In countless places they will diverge in giving, as their respective translations of a sentence of the one language, sentences of the other language which stand to each other in no plausible sort of equivalence however loose. The firmer the direct links of a sentence with non-verbal stimulation, of course, the less drastically its translations can diverge from one another from manual to manual. It is in this last form, as a principle of indeterminacy of translation, that I shall try to make the point plausible in the course of this chapter. But the chapter will run longer than it would if various of the concepts and considerations ancillary to this theme did not seem worthy of treatment also on their own account.

We are concerned here with language as the complex of present dispositions to verbal behavior, in which speakers of the same language have perforce come to resemble one another; not with the processes of acquisition, whose variations from individual to individual it is to the interests Of communication to efface (cf. § 2). The sentence 'That man shoots well’, said while pointing to an unarmed man, has as present stimulation the glimpse of the
marksman's familiar face. The contributory past stimulation includes past observations of the man's shooting, as well as remote episodes that trained the speaker in the use of the words. The past stimulation is thus commonly reckoned in part to the acquisition of language and in part to the acquisition of collateral information; however, this subsidiary dichotomy can await some indication of what it is good for and what general clues there are for it in observable verbal behavior. (Cf. §§ 9, 12, 14.)

Meanwhile what is before us is the going concern of verbal behavior and its currently observable correlations with stimulation. Reckon a man's current language by his current dispositions to respond verbally to current stimulation, and you automatically refer all past stimulation to the learning phase. Not but that even this way of drawing a boundary between language in acquisition and language in use has its fluctuations, inasmuch as we can consult our convenience in what bound we set to the length of stimulations counted as current. This bound, a working standard of what to count as specious present, I call the modulus of stimulation.

The recovery of a man's current language from his currently observed responses is the task of the linguist who, unaided by an interpreter, is out to penetrate and translate a language hitherto unknown. All the objective data he has to go on are the forces that he sees impinging on the native's surfaces and the observable behavior, vocal and otherwise, of the native. Such data evince native “meanings” only of the most objectively empirical or stimulus-linked variety. And yet the linguist apparently ends up with native “meanings” in some quite unrestricted sense; purported translations, anyway, of all possible native sentences.

Translation between kindred languages, e.g., Tristan and English, is aided by resemblance of cognate word forms. Translation between unrelated languages, e.g., Hungarian and English, may be aided by traditional equations that have evolved in step with a shared culture. What is relevant rather to our purposes is radical translation, i.e., translation of the language of a hitherto untouched people. The task is one that is not in practice undertaken in its extreme form, since a chain of interpreters of a sort can be recruited of marginal persons across the darkest archipelago. But the problem is the more nearly approximated the poorer the hints available from interpreters; thus attention to techniques of utterly radical translation has not been wanting.° I shall imagine that all help of interpreters is excluded. Incidentally I shall here ignore phonematic analysis (§ 18), early though it would come in our field linguist's enterprise; for it does not affect the philosophical point I want to make.
° See Pike.

The utterances first and most surely translated in such a case are ones keyed to present events that are conspicuous to the linguist and his informant. A rabbit scurries by, the native says ‘Gavagai’, and the linguist notes down the sentence ‘Rabbit’ (or ‘Lo, a rabbit’) as tentative translation, subject to testing in further cases. The linguist will at first refrain from putting words into his informant's mouth, if only for lack of words to put. When he can, though, the linguist has to supply native sentences for his informant's approval, despite the risk of slanting the data by suggestion. Otherwise he can do little with native terms that have references in common. For, suppose the native language includes sentences S1, S2, and S3, really translatable respectively as ‘Animal’, ‘White’, and ‘Rabbit’. Stimulus situations always differ, whether relevantly or not; and, just because volunteered responses come singly, the classes of situations under which the native happens to have volunteered S1, S2, and S3, are of course mutually exclusive, despite the hidden actual meanings of the words.

How then is the linguist to perceive that the native would have been willing to assent to S1 in all the situations where he happened to volunteer S3, and in some but perhaps not all of the situations where he happened to volunteer S2? Only by taking the initiative and querying combinations of native sentences and stimulus situations so as to narrow down his guesses to his eventual satisfaction.

So we have the linguist asking ‘Gavagai?’ in each of various stimulatory situations, and noting each time whether the native assents, dissents, or neither. But how is he to recognize native assent and dissent when he sees or hears them? Gestures are not to be taken at face value; the Turks' are nearly the reverse of our own. What he must do is guess from observation and then see how well his guesses work. Thus suppose that in asking ‘Gavagai?’ and the like, in the conspicuous presence of rabbits and the like, he has chatted the responses ‘Evet’ and ‘Yok’ often enough to surmise that they may correspond to ‘Yes’ and ‘No’, but has no notion which is which. Then he tries the experiment of echoing the native's own volunteered pronouncements. If thereby he pretty regularly elicits “Evet’ rather than ‘Yok’, he is encouraged to take ‘Evet’ as ‘Yes’. Also he tries responding with ‘Evet’ and Yok’ to the native's remarks; the one that is the more serene in its effect is the better candidate for Yes’. However inconclusive these methods, they generate a working hypothesis. If extraordinary difficulties attend all his subsequent steps, the linguist may decide to discard that hypothesis and guess again.°

° See Firth, Elements of Social Organization, p. 23, on the analogous matter of identifying a gesture of greeting.

Let us then suppose the linguist has settled on what to treat as native signs of assent and dissent. He is thereupon in a position to accumulate inductive evidence for translating ‘Gavagai’ as the sentence ‘Rabbit'. The general law for which he is assembling instances is roughly that the native will assent to ‘Gavagai?’ under just those stimulations under which we, if asked, would assent to ‘Rabbit?’; and correspondingly for dissent.

But we can do somewhat more justice to what the linguist is after in such a case if, instead of speaking merely of stimulations under which the native will assent or dissent to the queried sentence, we speak in a more causal vein of stimulations that will prompt the native to assent or dissent to the queried sentence. For suppose the queried sentence were one rather to the effect that someone is away tracking a giraffe. All day long the native will assent to it whenever asked, under all manner of irrelevant attendant stimulations; and on another day he will dissent from it under the same irrelevant stimulations. It is important to know that in the case of ‘Gavagai?’ the rabbit-presenting stimulations actually prompt the assent, and that the others actually prompt the dissent. In practice the linguist will usually settle these questions of causality, however tentatively, by intuitive judgment based on details of the native's behavior: his scanning movements, his sudden look of recognition, and the like. Also there are more formal considerations which, under favorable circumstances, can assure him of the prompting relation. If, just after the native has been asked S and has assented or dissented, the linguist springs stimulation s on him, asks S again, and gets the opposite verdict, then he may conclude that s did the prompting.

Note that to prompt, in our sense, is not to elicit. What elicits the native's ‘Evet’ or ‘Yok’ is a combination: the prompting stimulation plus the ensuing query ‘Gavagai?’.

§ 8. Stimulation and Stimulus Meaning

It is important to think of what prompts the native's assent to ‘Gavagai?’ as stimulations and not rabbits. Stimulation can remain the same though the rabbit be supplanted by a counterfeit. Conversely, stimulation can vary in its power to prompt assent to Gavagai’ because of variations in angle, lighting, and color contrast, though the rabbit remains the same. In experimentally equating the uses of ‘Gavagai’ and ‘Rabbit’ it is stimulations that must be made to match, not animals.

A visual stimulation is perhaps best identified, for present purposes, with the pattern of chromatic irradiation of the eye. To look deep into the subject's head would be inappropriate even if feasible, for we want to keep clear of his idiosyncratic neural routings or private history of habit formation. We are after his socially inculcated linguistic usage, hence his responses to conditions normally subject to social assessment. (Cf. § 2.) Ocular irradiation is intersubjectively checked to some degree by society and linguist alike, by making allowances for the speaker's orientation and the relative disposition of objects.

In taking the visual stimulations as irradiation patterns we invest them with a fineness of detail beyond anything that our linguist can be called upon to check for. But this is all right. He can reasonably conjecture that the native would be prompted to assent to ‘Gavagai’ by the microscopically same irradiations that would prompt him, the linguist, to assent to ‘Rabbit’, even though this conjecture rests wholly on samples where the irradiations concerned can at best be hazarded merely to be pretty much alike.

It is not, however, adequate to think of the visual stimulations as momentary static irradiation patterns. To do so would obstruct examples which, unlike ‘Rabbit’, affirm movement. And it would make trouble even with examples like ‘Rabbit’, on another account: too much depends on what immediately precedes and follows a momentary irradiation. A momentary lepiform image flashed by some artifice in the midst of an otherwise rabbitless sequence might not prompt assent to ‘Rabbit’ even though the same image would have done so if ensconced in a more favorable sequence. The difficulty would thus arise that far from hoping to match the irradiation patterns favorable to ‘Gavagai’ with those favorable to Rabbit’, we could not even say unequivocally of an irradiation pattern, of itself and without regard to those just before and after, that it is favorable to ‘Rabbit’ or that it is not.1 Better, therefore, to take as the relevant stimulations not momentary irradiation patterns, but evolving irradiation patterns of all durations up to some convenient limit or modulus. Furthermore we may think of the ideal experimental situation as one in which the desired ocular exposure concerned is preceded and followed by a blindfold.
1 This difficulty was raised by Davidson.

In general the ocular irradiation patterns are best conceived in their spatial entirety. For there are examples such as ‘Fine weather’ which, unlike ‘Rabbit’, are not keyed to any readily segregated fragments of the scene. Also there are all those rabbit-free patterns that are wanted as prompting dissent from ‘Rabbit’. And as for the patterns wanted as prompting assent to ‘Rabbit’, whole scenes will still serve better than selected portions might; for the difference between center and periphery, which is such an important determinant of visual attention, is then automatically allowed for. Total ocular irradiation patterns that differ in centering differ also in limits, and so are simply different patterns. One that shows the rabbit too peripherally simply will not be one that prompts assent to ‘Gavagai’ or ‘Rabbit’.

Certain sentences of the type of ‘Gavagai’ are the sentences with which our jungle linguist must begin, and for these we now have before us the makings of a crude concept of empirical meaning. For meaning, supposedly, is what a sentence shares with its translation; and translation at the present stage turns solely on correlations with non-verbal stimulation.

Let us make this concept of meaning more explicit and give it a neutrally technical name. We may begin by defining the affirmative stimulus meaning of a sentence such as ‘Gavagai’, for a given speaker, as the class of all the stimulations (hence evolving ocular irradiation patterns between properly timed blindfoldings) that would prompt his assent. More explicitly, in view of the end of §7, a stimulation s belongs to the affirmative stimulus meaning of a sentence S for a given speaker if and only if there is a stimulation s’ such that if the speaker were given s‘, then were asked S, then were given s, and then were asked S again, he would dissent the first time and assent the second. We may define the negative stimulus meaning similarly with ’assent’ and ‘dissent’ interchanged, and then define the stimulus meaning as the ordered pair of the two. We could refine the notion of stimulus meaning by distinguishing degrees of doubtfulness of assent and dissent, say by reaction time; but for the sake of fluent exposition let us forbear. The imagined equating of ‘Gavagai’ and 'Rabbit’ can now be stated thus: they have the same stimulus meaning.

A stimulus meaning is the stimulus meaning of a sentence for a speaker at a date; for we must allow our speaker to change his ways. Also it varies with the modulus, or maximum duration recognized for stimulations. For, by increasing the modulus we supplement the stimulus meaning with some stimulations that were too long to count before. Fully ticketed, therefore, a stimulus meaning is the stimulus meaning modulo n seconds of sentence S for speaker a at time t.

The stimulations to be gathered into the stimulus meaning of a sentence have for vividness been thought of thus far as visual, unlike the queries that follow them. Actually, of course, we should bring the other senses in on a par with vision, identifying stimulations not with just ocular irradiation patterns but with these and the various barrages of other senses, separately and in all synchronous combinations. Perhaps we can pass over the detail of this.

The affirmative and negative stimulus meanings of a sentence (for a given speaker at a given time) are mutually exclusive. Granted, our subject might be prompted once by a given stimulation s to assent to S, and later, by a recurrence of s, to dissent from S; but then we would simply conclude that his meaning for S had changed. We would then reckon s to his affirmative stimulus meaning of S as of the one date and to his negative stimulus meaning of S as of the other date. Yet the affirmative and negative stimulus meanings do not determine each other; for many stimulations may be expected to belong to neither. In general, therefore, comparison of whole stimulus meanings can be a better basis for translations than comparison merely of affirmative stimulus meanings.

What now of that strong conditional, the ‘would’ in our definition of stimulus meaning? Its use here is no worse than its use when we explain ‘x is soluble in water’ as meaning that z would dissolve if it were in water. What the strong conditional defines is a disposition, in this case a disposition to assent to or dissent from S when variously stimulated. The disposition may be presumed to be some subtle structural condition, like an allergy and like solubility; like an allergy, more particularly, in not being understood. The ontological status of dispositions, or the philosophical status of talk of dispositions, is a matter which I defer to § 46; but meanwhile we are familiar enough in a general way with how one sets about guessing, from judicious tests and samples and observed uniformities, whether there is a disposition of a specified sort.

The stimulus meaning of a sentence for a subject sums up his disposition to assent to or dissent from the sentence in response to present stimulation. The stimulation is what activates the disposition, as opposed to what instills it (even though the stimulation chance to contribute somehow to the instilling of some further disposition).

Yet a stimulation must be conceived for these purposes not as a dated particular event but as a universal, a repeatable event form. We are to say not that two like stimulations have occurred, but that the same stimulation has recurred. Such an attitude is implied the moment we speak of sameness of stimulus meaning for two speakers. We could indeed overrule this consideration, if we liked, by readjusting our terminology. But there would be no point, for there remains elsewhere a compelling reason for taking the stimulations as universals; viz., the strong conditional in the definition of stimulus meaning. For, consider again the affirmative stimulus meaning of a sentence S: the class X of all those stimulations that would prompt assent to S. If the stimulations were taken as events rather than event forms, then X would have to be a class of events which largely did not and will not happen, but which would prompt assent to S if they were to happen. Whenever X contained one realized or unrealized particular stimulatory event s, it would have to contain all other unrealized duplicates of s; and how many are there of these? Certainly it is hopeless nonsense to talk thus of unrealized particulars and try to assemble them into classes. Unrealized entities have to be construed as universals.

We were impressed in §3 with the interdependence of sentences. We may well have begun then to wonder whether meanings even of whole sentences (let alone shorter expressions) could reasonably be talked of at all, except relative to the other sentences of an inclusive theory. Such relativity would be awkward, since, conversely, the individual component sentences offer the only way into the theory. Now the notion of stimulus meaning partially resolves the predicament. It isolates a sort of net empirical import of each of various single sentences without regard to the containing theory, even though without loss of what the sentence owes to that containing theory. It is a device, as far as it goes, for exploring the fabric of interlocking sentences, a sentence at a time.

Between the notion of stimulus meaning and Carnap's remarks on empirical semantics there are connections and differences worth noting. He suggests exploring the meaning of a term by asking the subject whether he would apply it under various imaginary circumstances, to be described to him. That approach has the virtue of preserving contrasts between such terms as ‘goblin’ and ‘unicorn’ despite the non-existence of contrasting instances in the world. Stimulus meaning has the same virtue, since there are stimulation patterns that would prompt assent to ‘Unicorns’ and not to ‘Goblins’. Carnap's approach presupposes some decision as to what descriptions of imaginary circumstances are admissible; e.g., ‘unicorn’ would be not wanted in descriptions used in probing the meaning of ‘unicorn’. He hints of appropriate restrictions for the purpose, mentioning “size, shape, color”; and my notion of stimulus meaning itself amounts to a firmer definition in that same direction. There remains a significant contrast in the uses the two of us make of subjunctive conditionals: I limit them to my investigator's considered judgment of what the informant would do if stimulated; Carnap has his investigator putting such conditionals to the judgment of the informant. Certainly my investigator would in practice ask the same questions as Carnap's investigator, as a quick way of estimating stimulus meanings, if language for such questions happened to be available. But stimulus meaning can be explored also at the first stages of radical translation, where Carnap's type of questionnaire is unavailable. On this score it is important, as we shall see in §12, that my theory has to do primarily with sentences of a sort and not, like Carnap‘s, with terms.

9. Occasion sentences. Intrusive information
Occasion sentences, as against standing sentences, are sentences such as ‘Gavagai’, ‘bed’, ‘It hurts’, ‘His face is dirty’, which command assent or dissent only if queried after an appropriate prompting stimulation. Verdicts to standing sentences can be prompted too: stimulation implemented by an interferometer once prompted Michelson and Morley to dissent from the standing sentence ‘There is ether drift’, and a speaker's assent can be prompted yearly to ‘The crocuses are out’, daily to ‘The Times has come’. But these standing sentences contrast with occasion sentences in that the subject may repeat his old assent or dissent unprompted by current stimulation when we ask him again on later occasions, whereas an occasion sentence commands assent or dissent only as prompted all over again by current stimulation. Standing sentences grade off toward occasion sentences as the interval between possible repromptings diminishes; and the occasion sentence is the extreme case where that interval is less than the modulus. Like the stimulus meanings themselves, the distinction between standing sentences and occasion sentences is relative to the modulus; an occasion sentence modulo n seconds can be a standing sentence modulo n - 1.

The stimulations belonging to neither the affirmative nor the negative stimulus meaning of an occasion sentence are just those that would inhibit a verdict on the queried sentence, whether through indecisiveness (as in the case of a poor glimpse) or through shocking the subject out of his wits. On the other hand the stimulations belonging to neither the affirmative nor the negative stimulus meaning of a standing sentence are of two sorts: besides the inhibitory ones there are the irrelevant ones, which neither prompt nor inhibit. Querying the sentence on the heels of such a stimulation would elicit a verdict, but always the one that the query would have elicited without the attendant stimulation; never a change of verdict.

The stimulus meaning is a full cross-section of the subject's evolving dispositions to assent to or dissent from a sentence, if the sentence is an occasion sentence; less so if it is a standing sentence. Standing sentences can differ among themselves in ”meaning,” by any intuitive account°, as freely as occasion sentences; but, the less susceptible they are to prompted assent and dissent, the fewer clues are present in stimulus meaning. The notion of stimulus meaning is thus most important for occasion sentences, and we shall limit our attention for a while to them.

°Twice I have been startled to find my use of ‘intuitive’ misconstrued as alluding to some special and mysterious avenue of knowledge. By an intuitive account I mean one in which terms are used in habitual ways, without reflecting on how they might be defined or what presuppositions they might conceal.

Even for such favored occasion sentences as ‘Gavagai’ and ‘Rabbit’, actually, 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. Since the linguist would not on his own information be prompted by that same poor glimpse to assent to ‘Rabbit?’, we have here a discrepancy between the present stimulus meaning of ‘Gavagai’ for the informant and that of ‘Rabbit’ for the linguist.

More persistent discrepancies of the same type can be imagined, affecting not one native but all, and not once but regularly. There may be a local rabbit-fly,° unknown to the linguist, and recognizable some way off by its long wings and erratic movements; and seeing such a fly in the neighborhood of an ill-glimpsed animal could help a naive to recognize the latter as a rabbit. Ocular irradiations combining poor glimpses of rabbits with good ones of rabbit-flies would belong to the stimulus meaning of ‘Gavagai’ for natives, generally, and not to that of ‘Rabbit’ for the linguist.

And, to be less fanciful, there are all those stimulations that incorporate verbal hints from native kibitzers. Thus suppose that the stimulation on the heels of which the informant is asked ‘Gavagai?’ is a composite stimulation presenting a bystander pointing to an ill-glimpsed object and saying ‘Gavagai’. This composite stimulation will probably turn out to belong to the affirmative stimulus meaning of ‘Gavagai’ for the informant, and not to the stimulus meaning of ‘Rabbit’ for most English speakers, on whom the force of the bystander's verbal intervention would be lost. Such cases would not fool our linguist, but they do count against defining synonymy as sameness of stimulus meaning. For we must remember that every sufficiently brief stimulation pattern, though it be one that never gets actualized or that the linguist would never use, still by definition belongs to the stimulus meaning of ‘Gavagai’ for a man at a given time if it is one that would prompt his assent at that time.

° Here I am indebted to Davidson.

Intuitively the ideal would be to accord to the affirmative meaning of ‘Gavagai’ just those stimulations that would prompt assent to ‘Gavagai?’ on the strength purely of an understanding of ‘Gavagai’, unaided by collateral information: unaided by recent observation of rabbits near the spot, unaided by knowledge of the nature and habits of the rabbit-fly, unaided by conversance with the kibitzer's language. On the face of it there is a difficulty in excluding this third aid, considering our continuing dependence on the subject's understanding of ‘Gavagai’. But also the trouble is more wide- spread. It is precisely that we have made no general experimental sense of a distinction between what goes into a native's learning to apply an expression and what goes into his learning supplementary matters about the objects concerned. True, the linguist can press such a distinction part way; he can filter out such idiosyncratic bits of collateral matter as the informant's recent observation of rabbits near the spot, by varying his times and his informants and so isolating a more stable and more social stimulus meaning as common denominator. But any socially shared information, such as that about the rabbit-fly or the ability to understand a bystander's remark, will continue to affect even that common denominator. There is no evident criterion whereby to strip such effects away and leave just the meaning of ‘Gavagai’ properly so-called - whatever meaning properly so-called may be.

Thus, to depict the difficulty in more general terms, suppose it said that a particular class X comprises just those stimulations each of which suffices to prompt assent to a sentence S outright, without benefit of collateral information. Suppose it said that the stimulations comprised in a further class X’, likewise sufficient to prompt assent to S, owe their efficacy rather to certain widely disseminated collateral information, C. Now couldn't we just as well have said, instead, that on acquiring C, men have found it convenient implicitly to change the very ”meaning” of S, so that the members of X’ now suffice outright like members of X? I suggest that we may say either; even historical clairvoyance would reveal no distinction, though it reveal all stages in the acquisition of C, since meaning can evolve pari passu. The distinction is illusory: as mistaken as the notion, scouted in §4, that we can determine separately what to talk about and what to say about it. It is simply a question whether to call the transitivity shortcuts (§3) changes of meaning or condensations of proof; and in fact an unreal question. What we objectively have is just an evolving adjustment to nature, reflected in an evolving set of dispositions to be prompted by stimulations to assent to or dissent from sentences. These dispositions may be conceded to be impure in the sense of including worldly knowledge, but they contain it in a solution which there is no precipitating.

Incidentally, note that stimulus meanings as defined in § 8 can even suffer some discrepancies that are intuitively attributable neither to differences of meaning nor to differences of collateral information. Thus take shocked silence. To begin with, if the speaker is already stunned at time t, all stimulus meanings for him at t will be empty. This outcome of the definition of stimulus meaning is unnatural but harmless, since we can ignore stimulus meanings for stunned persons. But in the case of a speaker alert at t there are stimulations that would stun him at t and so would preclude any assent to or dissent from the ensuing ‘Gavagai?’. These, by definition, belong to neither the affirmative nor the negative stimulus meaning of ‘Gavagai’ for him at t. Now where a discrepancy in stimulus meanings will ensue is where a stimulation is such as would stun one speaker and not another; for it could be- long say to the negative stimulus meaning of ‘Gavagai’ or ‘Rabbit’ for the latter speaker and to neither the affirmative nor the negative stimulus meaning for the former speaker. This again is a discrepancy that would not puzzle the linguist, but that exists under our definition. Also there are interferences of less drastic sorts. The native may dissent from ‘Gavagai’ in plain sight of the rabbit‘s ears, because the rabbit is in no position for shooting;° he has misjudged the linguist's motive for asking ‘Gavagai?’.

°Here I am indebted to Raymond Pirth.

We have now seen that stimulus meaning as defined falls short in various ways of one's intuitive demands on “meaning” as undefined, and that sameness of stimulus meaning is too strict a relation to expect between a native occasion sentence and its translation—even in so benign a case as ‘Gavagai’ and ‘Rabbit’. Yet stimulus meaning, by whatever name, may be properly looked upon still as the objective reality that the linguist has to probe when he undertakes radical translation. For the stimulus meaning of an occasion sentence is by definition the native's total battery of present dispositions to be prompted to assent to or to dissent from the sentence; and these dispositions are just what the linguist has to sample and estimate.

We do best to revise not the notion of stimulus meaning, but only what we represent the linguist as doing with stimulus meanings. The fact is that he translates not by identity of stimulus meanings, but by significant approximation of stimulus meanings. If he translates ‘Gavagai’ as ‘Rabbit’ despite the discrepancies in stimulus meaning imagined above, he does so because the stimulus meanings seem to coincide to an overwhelming degree and the discrepancies, so far as he finds them, seem best explained away or dismissed as effects of unidentified interferences. Some discrepancies he may sift out, as lately suggested, by varying his times and informants. Some, involving poor glimpses or shock or verbal intrusions, he would not even bother to bring to fulfillment by a querying of the sentence. Some, such as those involving the rabbit-fly, he will dismiss as effects of unidentified interferences if he does not encounter them often. In taking this last rather high line, clearly he is much influenced by his natural expectation that any people in rabbit country would have some brief expression that could in the long run be best translated simply as ‘Rabbit’. He conjectures that the now-unexplained discrepancies between ‘Gavagai’ and ‘Rabbit’ are ones that may eventually be reconciled with his translation, after he has somehow got deep enough into the native language to ask sophisticated questions.

In practice, of course, the natural expectation that the natives will have a brief expression for ‘Rabbit’ counts overwhelmingly. The linguist hears ‘Gavagai’ once, in a situation where a rabbit seems to be the object of concern. He will then try ‘Gavagai’ for assent or dissent in a couple of situations designed perhaps to eliminate ‘White’ and ‘Animal’ as alternative translations, and will forthwith settle upon ‘Rabbit’ as translation without further experiment— though always in readiness to discover through some unsought experience that a revision is in order. I made the linguist preternaturally circumspect, and maximized his bad luck in respect of discrepant observations, in order to consider what theoretical bearing a native's collateral information can have upon the linguist's in fact wholly facile opening translation.

§10. Observation sentences

Some stimulus meanings are less susceptible than others to the influences of intrusive information. There is on this score a significant contrast between ‘Red’ and ‘Rabbit’ even when ‘Red’ is taken on a par with ‘Rabbit’ as announcing not a passing sense datum but an enduring objective trait of the physical object. True, there are extreme cases where we may be persuaded, by collateral information about odd lighting and juxtaposition, that something is really red that did not seem so or vice versa; but, despite such cases there is less scope for collateral information in deciding whether a glimpsed thing is red than in deciding whether it is a rabbit. In the case of ‘Red’, therefore, sameness of stimulus meaning comes unusually close to what one intuitively expects of synonymy.

Color words are notoriously ill matched between remote languages, because of differences in customary grouping of shades. But this is no present problem; it means merely that there may well be no native occasion sentence, at least no reasonably simple one, with approximately the stimulus meaning of ‘Red’. Again, even if there is one, there may still be a kind of trouble in equating it to ‘Red’, just because of the vagueness of color boundaries in both languages. But this again is no problem of collateral information; it is a difficulty that would remain even if a distinction between meaning and collateral information were successfully drawn. It can be coped with by a rough matching of statistical scatterings. The penumbra of vagueness of ‘Red’ consists of stimulations in respect of which the stimulus meanings of ‘Red’ tend to vary from speaker to speaker and from occasion to occasion; correspondingly for the penumbra of vagueness of the native sentence; and then ‘Red’ is a good translation to the extent that it resembles the native sentence umbra for umbra and penumbra for penumbra.

In terms of direct behavioral evidence, how do those fluctuations of stimulus meaning that are attributable to a penumbra of vagueness differ from those fluctuations of stimulus meaning (e.g. of ‘Gavagai’) that are laid to variations of collateral information from occasion to occasion? Partly in that the penumbral fluctuations increase rather smoothly as the stimulations grade off, while the fluctuations laid to collateral information are more irregular, suggesting intrusion of extraneous factors. But mainly in that each individual's assent or dissent tends to be marked by doubt and hesitation when the prompting stimulation belongs to the penumbra. If we were to complicate the notion of stimulus meaning to the extent of weighting each stimulation inversely according to reaction time (ct. § 8), then discrepancies in stimulus meaning from speaker to speaker would tend to count for little where due to vagueness, and for more where not.

If ‘Red’ is somewhat less susceptible than ‘Rabbit’ to the influences of intrusive information, there are other sentences that are vastly more so. An example is ‘Bachelor’. An informant's assent to it is prompted genuinely enough by the sight of a face, yet it draws mainly on stored information and none on the prompting stimulation except as needed for recognizing the bachelor friend concerned. As one says in the uncritical jargon of meaning, the trouble with ‘Bachelor’ is that its meaning transcends the looks of the prompting faces and concerns matters that can be known only through other channels. ‘Rabbit’ is a little this way, as witness papier-maché counterfeits; ‘Bachelor’ much more so. The stimulus meaning of ‘Bachelor’ cannot be treated as its “meaning” by any stretch of the imagination, unless perhaps accompanied by a stretch of the modulus.

A mark of the intrusion of collateral information, except when the information is generally shared as in the examples of the kibitzer and the rabbit-fly (§9), was discrepancy in stimulus meaning from speaker to speaker of the same language. In a case like ‘Bachelor’, therefore, we may expect the discrepancies to be overwhelming; and indeed they are. For any two speakers whose social contacts are not virtually identical, the stimulus meanings of ‘Bachelor’ will diverge far more than those of ‘Rabbit’.

The less susceptible the stimulus meaning of an occasion sentence is to the influences of collateral information, the less absurdity there is in thinking of the stimulus meaning of the sentence as the meaning of the sentence. Occasion sentences whose stimulus meanings vary none under the influence of collateral information may naturally be called observation sentences, and their stimulus meanings may without fear of contradiction be said to do full justice to their meanings. These are the occasion sentences that wear their meanings on their sleeves. Or, better, we may speak of degree of observationality; for even the stimulus meaning of ‘Red’ can, we noted, be made to fluctuate a little from occasion to occasion by collateral information on lighting conditions. What we have is a gradation of observationality from one extreme, at ‘Red’ or above, to the other extreme at ‘Bachelor’ or below.

In the foregoing paragraph we have wallowed most unfastidiously in the conceptual slough of meaning and collateral information. But now it is interesting to note that what we have dredged out, a notion of degree of observationality, is not beyond cleaning up and rendering respectable. For, in behavioral terms, an occasion sentence may be said to be the more observational the more nearly its stimulus meanings for different speakers tend to coincide. Granted, this definition fails to give demerit marks for the effects of generally shared information, such as that about the rabbit-fly. But, as argued in §9, I suspect that no systematic experimental sense is to be made of a distinction between usage due to meaning and usage due to generally shared collateral information.

The notion of observationality is relative to the modulus of stimulation. This is not to be wondered at, since the notion of stimulus meaning was relative to the modulus (ct. §8), and so is the very distinction between habit formation and habit formed (cf. §7). Observationality increases with the modulus, in the following way. A typical case of discrepancy between the stimulus meanings of ‘Gavagai’, for two natives, is the case where one native and not the other has lately seen rabbits near the spot that they are now viewing. An ill-glimpsed movement would now prompt the one native and not the other to assent to ‘Gavagai?’. But if we make the modulus long enough to include as part of the one native's present stimulation his recent observation of rabbits near the spot, then what had been a discrepancy between stimulus meanings is a mere divergence of stimulations: the one stimulation is such as would prompt either native to assent, and the other neither. Increase the modulus sufficiently to take in extended periods of learning about friends and you even increase the observationality of ‘Bachelor’. But let us forget moduli again for a while, thus keeping our variables down.

We have defined observationality for occasion sentences somewhat vaguely, as degree of constancy of stimulus meaning from speaker to speaker. It would not do to use this definition generally among standing sentences, since the stimulus meaning of a standing sentence can show fair constancy from speaker to speaker for the wrong reason: mere sparseness of member stimulations. Among standing sentences that are well over toward the occasion end (cf. §9), however, the notion of observationality works quite as well as among occasion sentences, and is significant in the same way; viz., the higher the observationality, the better we can get on with translation by stimulus meaning. We could hope, e.g., to translate ‘The tide is out’ by a rough matching of stimulus meanings; not so ‘There is a famous novelist on board’.

Viewing the graded notion of observationality as the primary one, we may still speak of sentences simply as observation sentences when they are high in observationality. In a narrow sense, just ‘Red’ would qualify; in a wider sense, also ‘Rabbit’ and ‘The tide is out’. It is for observation sentences in some such sense that the notion of stimulus meaning constitutes a reasonable notion of meaning.

To philosophers ‘observation sentence’ suggests the datum sentences of science. On this score our version is not amiss; for the observation sentences as we have identified them are just the occasion sentences on which there is pretty sure to be firm agreement on the part of well-placed observers. Thus they are just the sentences on which a scientist will tend to fall back when pressed by doubting colleagues. Moreover, the philosophical doctrine of infallibility of observation sentences is sustained under our version. For there is scope for error and dispute only insofar as the connections with experience whereby sentences are appraised are multifarious and indirect, mediated through time by theory in conflicting ways; there is none insofar as verdicts to a sentence are directly keyed to present stimulation. (This immunity to error is, however, like observationality itself, for us a matter of degree.) Our version of observation sentences departs from a philosophical tradition in allowing the sentences to be about ordinary things instead of requiring them to report sense data, but this departure has not lacked proponents.°

° For remarks on this matter and references see von Mises, Positivism, pp. 91-55, 379. To the main theme of this paragraph I sense harmony in Strawson, Individuals, p. 212: “If any facts deserve ... to be called ... atomic facts, it is the facts stated by those propositions which demonstratively indicate the incidence of a general feature.” for the propositions alluded to seem, in the light of adjacent text, to correspond pretty well to what I have called occasion sentences.

In estimating the stimulus meaning of a sentence for a speaker at a given time, the linguist is helped by varying the time and speaker. In choosing a translation, he is helped by comparing native speakers and so eliminating idiosyncrasies of stimulus meaning. Still the notion of stimulus meaning itself, as defined, depends on no multiplicity of speakers. Now the notion of observationality, in contrast, is social. The behavioral definition offered for it above turns on similarities of stimulus meanings over the community.

What makes an occasion sentence low on observationality is, by definition, wide intersubjective variability of stimulus meaning. Language as a socially inculcated set of dispositions is substantially uniform over the community, but it is uniform in different ways for different sentences. If a sentence is one that (like ‘Red’ and ‘Rabbit’) is inculcated mostly by something like direct ostension, the uniformity will lie at the surface and there will be little variation in stimulus meaning; the sentence will be highly observational. If it is one that (like ‘Bachelor’) is inculcated through connections with other sentences, linking up thus indirectly with past stimulations of other sorts than those that serve directly to prompt present assent to the sentence, then its stimulus meaning will vary with the speakers’ pasts, and the sentence will count as very unobservational. The stimulus meaning of a very unobservational occasion sentence for a speaker is a product of two factors, a fairly standard set of sentence-to-sentence connections and a random personal history; hence the largely random character of the stimulus meaning from speaker to speaker.

Now this random character has the effect not only that the stimulus meaning of the sentence for one speaker will differ from the stimulus meaning of that sentence for other speakers. It will differ from the stimulus meaning also of any other discoverable sentence for other speakers, in the same language or any other. Granted, a great complex English sentence can be imagined whose stimulus meaning for one man matches, by sheer exhaustion of cases, another man's stimulus meaning of ‘Bachelor’; but such a sentence would never be spotted, because nobody's stimulus meaning of ’Bachelor’ would ever be suitably inventoried to begin with.

For, consider again how it was with ‘Gavagai’. Here the stimulations belonging to the affirmative stimulus meaning share a distinctive trait that is salient, to us as well as to the native: the containing of rabbit glimpses. The trait is salient enough so that the linguist generalizes on it from samples: he expects the next glimpse of a rabbit to prompt assent to ‘Gavagai’ as past ones have. His generalization is repeatedly borne out, and he concludes with his conjecture that the native's whole stimulus meaning of ‘Gavagai'— never experimentally exhausted, of course—will tend to match ours of ‘Rabbit’. Now a similar effort with a non-observational native occasion sentences of the type of our ‘Bachelor’, would have bogged down in its early stages. Sample stimulations belonging to the affirmative stimulus meaning of such a sentence, for the given native, would show no tempting common traits by which to conjecture further cases, or none but such as fail to hold up on further tries.

§11. Intrasubjective synonymy of occasion sentences

Stimulus meaning remains defined without regard to observationality. But when applied to non-observational sentences like ‘Bachelor’ it bears little resemblance to what might reasonably be called meaning. Translation of ‘Soltero’ as ‘Bachelor’ manifestly cannot be predicated on identity of stimulus meanings between speakers; nor can synonymy of ‘Bachelor’ and ‘Unmarried man’.

But curiously enough the stimulus meanings of ‘Bachelor’ and ‘Unmarried man’ are, despite all this, identical for any one speaker.° An individual would at any one time be prompted by the same stimulations to assent to ‘Bachelor’ and ‘Unmarried man’; and similarly for dissent. Stimulus synonymy, or sameness of stimulus meaning, is as good a standard of synonymy for non-observational occasion sentences as for observation sentences as long as we stick to one speaker. For each speaker, ‘Bachelor’ and ‘Unmarried man’ are stimulus-synonymous without having the same meaning in any acceptably defined sense of ‘meaning’ (for stimulus meaning is, in the case of ‘Bachelor’, nothing of the kind). Very well; here is a case where we may welcome the synonymy and let the meaning go.

° It can be argued that this much-used example of synonymy has certain imperfections having to do with ages, divorce, and bachelors of arts. Another example much used in philosophy, ‘brother’ and ‘male sibling’, may be held to bog down under certain church usages. An example that is perhaps unassailable is ‘mother's father’ and ‘maternal grandfather’ (poetic connotations not being here in point), or ‘widower’ and ‘man who lost his wife’ (Jakobson). However, with this much by way of caveat against quibbling, perhaps we can keep to our conventional example and overlook its divagations.

The one-speaker restriction presents no obstacle to saying that ‘Bachelor’ and ‘Unmarried man’ are stimulus-synonymous for the whole community, in the sense of being thus for each member. A practical extension even to the two-language case is not far to seek if a bilingual speaker is at hand. ‘Bachelor’ and ‘Soltero’ will be stimulus-synonymous for him. Taking him as a sample, we may treat ‘Bachelor’ and ‘Soltero’ as synonymous for the translation purposes of the two whole linguistic communities that he represents. Whether he is a good enough sample would be checked by observing the fluency of his communication in both communities and by comparing other bilinguals.

Section 10 left the linguist unable to guess the trend of the stimulus meaning of a non-observational occasion sentence from sample cases. We now see a way, though costly, in which he can still accomplish radical translation of such sentences. He can settle down and learn the native language directly as an infant might.° Having thus become bilingual, he can translate the non-observational occasion sentences by introspected stimulus synonymy.

° See Chapter III for reflections on the infant's learning of our own language.

This step has the notable effect of initiating clear recognition of native falsehoods. As long as the linguist does no more than correlate the native's observation sentences with his own by stimulus meaning, he cannot discount any of the native's verdicts as false— unless ad hoc, most restrainedly, to simplify his correlations. But once he becomes bilingual and so transcends the observation sentences, he can bicker with the native as a brother.

Even short of going bilingual there is no difficulty in comparing two non-observational native sentences to see if they are intrasubjectively stimulus-synonymous for the native. The linguist can do this without having intuitively conjectured the trend of stimulus meaning of either sentence. He need merely query the sentences in parallel under random stimulations until he either hits a stimulation that prompts assent or dissent to one sentence and not to the other, or else is satisfied at last that he is not going to. A visiting Martian who never learns under what circumstances to apply ‘Bachelor’, or ‘Unmarried man’ either, can still find out by the above method that ‘Bachelor’ for one English speaker does not have the same stimulus meaning as ‘Bachelor’ for a different English speaker and that it has the same as ‘Unmarried man’ for the same speaker. He can, anyway, apart from one difficulty: there is no evident reason why it should occur to him thus blindly to try comparing ‘Unmarried man’ with ‘Bachelor’. This difficulty mâkes the intrasubjective stimulus synonymy of non-observational occasion sentences less readily accessible to an alien linguist than the stimulus synonymy of observation sentences such as ‘Gavagai’ and ‘Rabbit’. Still the linguist can examine for intrasubjective stimulus synonymy any pair of native occasion sentences that it occurs to him to wonder about; and we shall see in §15 how indirect considerations can even suggest such pairs for examination.

Between the stimulus meaning of any sentence for one man and the stimulus meaning of the same or any other sentence for another man there are almost bound to be countless discrepancies in point of verbally contaminated stimulations, as long as one man understands a language that the other does not. The argument is that of the kibitzer case in §9. The translating linguist had for this reason to discount verbally contaminated discrepancies. But intra-subjective comparisons are free of this trouble. Intrasubjectively we can even compare the occasion sentences ‘Yes’, ‘Uh huh’, and ‘Quite’ for stimulus synonymy, though the stimulations that enter into the stimulus meanings of these sentences are purely verbal in their relevant portions. A further advantage of the intrasubjective situation appears in the ease of stimulations that would at a given time shock one speaker and not another into silence ( cf. § 9); for clearly these will constitute no discrepancies intrasubjectively. Altogether the equating of stimulus meanings works out far better intrasubjectively than between subjects: it goes beyond observation sentences, it absorbs shock, and it better accommodates verbal stimulations.

Verbal stimulations can plague even the intrasubjective comparisons when they are stimulations of “second intention"—i.e., when besides consisting of words they are about words. Second-intention examples are the bane of theoretical linguistics, also apart from synonymy studies. Thus take the linguist engaged in distinguishing between those sequences of sounds or phonemes that can occur in English speech and those that cannot: all his excluded forms can return to confound him in second-intention English, as between quotation marks. Now some second-intention stimulations that could prompt a subject to assent to one of the queries ‘Bachelor?’ and ‘Unmarried man?’ to the exclusion of the other are as follows: a stimulation presenting the spelling of ‘bachelor’; a stimulation presenting the words ’rhymes with harried man’; a stimulation presenting a glimpse of a bachelor friend together with a plea to redefine ‘bachelor’. It is not easy to find a behavioral criterion of second-intention whereby to screen such cases, especially the last.

Leaving that problem unsolved, we have still to note another and more humdrum restriction that needs to be observed in equating sentences by stimulus meanings: we should stick to short sentences. Otherwise subjects’ mere incapacity to digest long questions can, under our definitions, issue in difference of stimulus meanings between long and short sentences which we should prefer to find synonymous. A stimulation may prompt assent to the short sentence and not to the long one just because of the opacity of the long one; yet we should then like to say not that the subject has shown the meaning of the long sentence to be different, but merely that he has failed to encompass it. Still a concept of synonymy initially significant only for short sentences can be extended to long sentences by analogy, e.g. as follows. By a construction, linguistically speaking, let us understand any fixed way of building a composite expression from arbitrary components of appropriate sort, one or more at a time. (What is fixed may include certain additive words, as well as the way of arranging the unfixed components.) Now two sentence-forming constructions may be so related that whenever applied to the same components they yield mutually synonymous results, as long as the results are short enough to be confirmed for synonymy. In this event it is natural, by extension, to count also as mutually synonymous any results of applying those constructions to identical components however long. But to simplify ensuing considerations let us continue to reason without reference to this refinement where we can.

Our success with ‘Bachelor’ and ‘Unmarried man’ has been sufficient, despite the impasse at second intention, to tempt us to over- estimate how well intrasubjective stimulus synonymy withstands collateral information. By way of corrective, consider the Himalayan explorer who has learned to apply ‘Everest’ to a distant mountain seen from Tibet and ‘Gaurisanker’ to one seen from Nepal. As occasion sentences these words have mutually exclusive stimulus meanings for him until his explorations reveal, to the surprise of all concerned, that the peaks are identical. His discovery is painfully empirical, not lexicographic; nevertheless the stimulus meanings of ‘Everest’ and ‘Gaurisanker’ coincide for him thenceforward.°

° I am indebted to Davidson for this point and to Schrödinger, What is Life? for the example.

Or again consider the occasion sentences ‘Indian nickel’ and ‘Buffalo nickel’. These have distinct stimulus meanings for a boy for his first minute or two of passive acquaintance with these coins, and when he gets to turning them over the stimulus meanings tend to fuse.

Do they fully fuse? The question whether ‘Indian nickel’ and ‘Buffalo nickel’ have the same stimulus meaning for a given subject is the question whether any sequence of ocular irradiations or other stimulation (within the modulus), realized or not, would now prompt the subject to assent to or dissent from ‘Indian nickel’ and not ‘Buffalo nickel’ or vice versa. Among such stimulations are those that present, to all appearances, a coin whose obverse is like that of an Indian nickel but whose reverse bears some device other than the buffalo. Such stimulations can with a little felony even be realized. After a modulus-long examination of such a hybrid coin, a novice might conclude with surprise that there are after all two kinds of Indian nickel, while an expert, sure of his numismatics, might conclude that the coin must be fraudulent. For the expert, ‘Indian nickel’ and ‘Buffalo nickel’ are stimulus-synonymous; for the novice not.

The novice does believe and continues to believe, as the expert does, that all Indian nickels are buffalo nickels and vice versa; for the novice has not been and will not be actually subjected to the surprising stimulation described. But the mere fact that there is such a stimulation pattern and that the novice would now thus respond to it (whether we know it or not) is what, by definition, makes the stimulus meanings of ‘Indian nickel’ and ‘Buffalo nickel’ differ for the novice even as of now.

To keep our example pertinent we must abstract from what may be called the conniving mode of speech: the mode in which we knowingly speak of Olivier as Macbeth, of a statue of a horse as a horse, of a false nickel as a nickel. Even the expert would in practice speak of the prepared coin as “that Indian nickel with the whoozis on the back,” adding that it was phony. Here we have a broader usage of ‘nickel’, under which nobody would seriously maintain even that all Indian nickels are in point of fact buffalo nickels and vice versa; whereas our purpose in the example is to examine two supposedly coextensive terms for sameness of stimulus meaning. In the example, therefore, read ‘Indian nickel’ and ‘buffalo nickel’ as ‘real Indian nickel’, ‘real buffalo nickel’.

From the example we see that two terms can in fact be coextensive, or true of the same things, without being intrasubjectively stimulus-synonymous as occasion sentences. They can be believed coextensive without being, even for the believer, stimulus-synonymous as occasion sentences; witness ‘Indian nickel’ and ‘Buffalo nickel’ for the novice. But when as in the expert‘s case the belief is so firm that no pattern of stimulation (within the modulus) would suffice to dislodge it, they are stimulus-synonymous as occasion sentences.

So it is apparent that intrasubjective stimulus synonymy remains open to criticism, from intuitive preconceptions, for relating occasion sentences whose stimulus meanings coincide on account of collateral information. Now there is still a way of cutting out the effects of idiosyncratic information: we can hold out for virtual constancy over the community. In this social sense of stimulus synonymy, ‘Indian nickel’ and ‘Buffalo nickel’ would cease to count as stimulus-synonymous, because of such speakers as our novice; whereas ‘Bachelors’ and ‘Unmarried man’ might still rate as stimulus-synonymous even socially, as being intrasubjectively stimulus-synonymous for nearly everybody. There is still no screen against the effects of collateral information common to the community; but, as urged in §9, I think that at that point the ideal becomes illusory.

§12. Synonymy of Terms

In starting our consideration of meaning with sentences we have hewn the line of §§ 3 and 4, where it was stressed that words are learned only by abstraction from their roles in learned sentences. But there are one-word sentences, such as ‘Red’ and ‘Rabbit’. Insofar as the concept of stimulus meaning may be said to constitute in some strained sense a meaning concept for these, it would seem to constitute a meaning concept for general terms like ‘red’ and ‘rabbit’. This, however, is a mistake. Stimulus synonymy of the occasion sentences ‘Gavagai’ and ‘Rabbit’ does not even guarantee that ‘gavagai’ and ‘rabbit’ are coextensive terms, terms true of the same things. For, consider ‘gavagai’. Who knows but what the objects to which this term applies are not rabbits after all, but mere stages, or brief temporal segments, of rabbits? In either event the stimulus situations that prompt assent to ‘Gavagai’ would be the same as for ‘Rabbit’. Or perhaps the objects to which ‘gavagai’ applies are all and sundry undetached parts of rabbits; again the stimulus meaning would register no difference. When from the sameness of stimulus meanings of ‘Gavagai’ and ‘Rabbit’ the linguist leaps to the conclusion that a gavagai is a whole enduring rabbit, he is just taking for granted that the native is enough like us to have a brief general term for rabbits and no brief general term for rabbit stages or parts.

A further alternative likewise compatible with the same old stimulus meaning is to take