Grazer Philosophische Studien, 45, 125-134 (1993)
My title's question is borrowed from a section of the first chapter of Thomson's recent book, Needs (1987). His is one of at least three recent, philosophically sophisticated works that have challenged what I once took, no doubt through reliance on too exiguous a data base, to be a movement towards consensus on a positive answer to the question (1980, p. 18)
My main aim in this brief note is to examine the reasons offered by these authors for rejecting the claim that 'A needs X' is always elliptical. But before we can begin, it is necessary to say a little about the terms of the debate, the notion of ellipsis here appealed to.
As I have shown at length elsewhere (1986), there are several distinguishable notions tied to the term ellipsis in traditional grammar, rhetoric, and the philosophizing that shared a common background with those activities. The root notion is simply of the marked absence of something from a sentence that is nevertheless used correctly. It is particularly clear when the resulting sentence could not be used correctly in isolation. I say 'Jim would like tea'; you ask 'With milk and sugar?' It would be possible, albeit pedantic, for you to utter the full 'Would Jim like tea with milk and sugar?' Once given this possibility of characterising sentences as reduced variants of something fuller, various reasons can be found for claiming that something has been left out — some good, some bad. Notoriously, grammatical theories could too easily be saved from refutation by appeal to elided elements.
In this tradition, the philosophical use of 'ellipsis' has typically been unreflective, a convenient tool for some purposes but not one to pay much attention to. It is easy to think that it is just a matter of any old lack of specificity — a criticism Hamblin (1970) offers of his reconstruction of Boethius' use of the notion. This is a trivializing move that attracts some who often make use of the notion for their own purposes, as for instance W.E. Johnson, who said à propos the claim 'all voters are males' that, in its context (1914; England; election to Parliament), it is perfectly intelligible since "the phrase, though elliptical — like all phrases in discussion or conversation — does not require the explicit introduction of every well-understood qualification" (1921, p. 161). But if one reviews the typical examples philosophers have used to illustrate ellipsis, it is clear that the notion is not the totally useless one of something or other that has been omitted. Rather, one offers an ellipsis as an answer to the question: "what else, over and above the workings of proper names and token-reflexives, do we need to know to begin to ascertain the truth value of the sentence in question?" (Brandon, 1986, p. 99).
What Johnson's example suggests is that we need to distinguish parameters (a term borrowed from Barwise, 1986) that are usually filled contextually for all sentences of a particular type — such as those associated with tenses and token-reflexives, or differently those that restrict quantifiers to a changing "universe of discourse" — from parameters that are attached to particular phrases. If we imagine someone in Johnson's context issuing an instruction that all voters over 40 should go to the left-hand booth, the former parameters would not give us enough to be able to pick out which booth is meant; we would have to add the perspective that comes with the phrase 'left'. 'A is to the left of B' is then elliptical, or one could say that what looks like a binary relation 'to the left of' is really elliptical for what is (at least)1 a ternary relation, whereas the predicate 'voter' is not itself elliptical, though the qualifications Johnson alludes to have to be filled in from the context of its various uses. I have no doubt that this distinction is sometimes a matter either for judgement or for decision on grounds of overall theoretical simplicity, as philosophers with only a sketch of a theory like to say; but equally, it does seem a serviceable distinction in a fair range of cases.
Acknowledging, then, the rough and ready nature of the tools we are employing, let us turn to the question Thomson posed. In the course of his discussion, Thomson has already claimed that 'A needs X' implies that 'X is a necessary condition of A's Þ-ing in circumstances C', so that if A needs X, "then there must be something that X is a necessary condition of" (p. 15). But, he says, this does not mean that our original statement is always elliptical since it may be that "the meaning of the term 'need' fixes logically what X is necessary for". At this point he invokes an earlier claim that there are two concepts associated with the verb, need2 'instrumental need' (which behaves elliptically) and 'fundamental need'. When we are using the second concept, we know that without X A must suffer serious harm, so it is part of the meaning of need here that it is the avoidance of serious harm that is in question, so there is no parameter left to be filled by contextual clues. Unfortunately we do not have any clear linguistic signal to pick out which concept is being tied to the verb, need, so Thomson has to admit that 'A needs X' may be ambiguous.
With respect to 'instrumental need', when 'A needs X' is admittedly elliptical, Thomson says that one should not draw attention to the ellipsis by suggesting that the question 'What does A need X for?' requires an answer (he is here responding to a remark by Flew). The question and its answer imply, according to Thomson, that A has a reason for needing X. Whether or not this is so (and I do not see such an implication in what appears to me a non-deviant question, 'What does sunlight need eight minutes for?', to use one of Thomson's examples), what is at stake here is, as he says, that X is instrumental or a means to some end.
Or perhaps not quite that. It is a curious fact that many writers move unhesitatingly from talking of F being a necessary condition of G to thinking of F and G being causally or means-end related.3 But as they know full well in other contexts, there can be logically necessary conditions; and depending on how broadly one takes causation, there may be other sorts of non-causal necessary conditions (to be accurate, in many cases both causal and non-causal, what we have are at least inus [insufficient but necessary parts of unnecessary but sufficient] conditions, in Mackie's (1965) terminology, but I shall leave this sophistication aside). Parts, for instance, may be necessary for the wholes they constitute.
Given that necessary condition is not restricted to causal contexts, I do not see any reason for reading in such a restriction to the meaning of those bits of language we analyse by reference to such conditions. We thereby eliminate, for instance, David Miller's (1976) reason for rejecting the elliptical analysis for all cases of 'A needs X'. (Miller's target, Brian Barry (1965, p. 48), left himself vulnerable by saying it always makes sense to ask for what purpose or end something is needed.) Against a general means-end, X to achieve Y, analysis, Miller offers 'functional' need-statements such as 'surgeons need manual dexterity' where the Y-slot could only be filled 'vacuously' by something like 'to be surgeons' or 'to do their job' and 'intrinsic' need-statements, Thomson's 'fundamental needs', ( 'Men need food', 'He needs someone to understand him') where he says "what is needed in these cases is not a means to an end but a part of the end itself" (p. 128). Once we read 'A needs X' as elliptical for 'X is a condition necessary in the circumstances for A to be Y', Miller's counter-examples to Barry fit quite happily. We may find it useful to distinguish causal conditions from preconditions, or to make any other distinctions amongst the things we talk about, but we do not have to make these shifting distinctions part of the meaning of the language we are using. As we saw Thomson conceding, there is no linguistic way to signal what he supposes to be a distinct sense of the verb, need. Better to suppose there is no such sense, only uses in one context, and uses in others.
Thomson's position rests on the prior claim that there are both normative and non-normative need-statements. He contrasts claims such as 'I need your love' with 'To die one's brain needs to stop working'. The former is said to carry a practical force lacking in the latter; if 'need' is replaced with 'want' in the former, its seriousness diminishes. Undoubtedly, typical discourses involving such sentences would differ in these ways. What is doubtful is whether we should locate these differences in different senses of the verb, need. Gricean considerations of the point of language use and the "rules" of polite conversation might well enable us to construe all uses of 'A needs X' uniformly, as regards basic meaning, while understanding why some uses are more likely than others to carry practical force. When A both wants and needs X, the former is just a psychological fact about A, the latter portrays the situation as one in which X is essential for some unspecified state of A; by not specifying we encourage the thought that it is essential for A's survival (or something not too far removed therefrom); and so its greater urgency is hardly surprising. Nor, as Thomson himself admits, is the fact that this urgency can be wiped out by actually filling in the ellipsis with some unimpressive Y. Why does not-specifying point us in the direction of 'fundamental needs'? Simply because not specifying lets us assume that the audience can supply the Y off their own bat, without knowing anything special about A; A's adventitious purposes would not be thus available, so all we can fall back on are matters related to persistence as A.
The same Gricean account would also allow us to say (apparently against Thomson — "the concept of an instrumental need is non-normative" (p. 7)) that some need-statements of the kind Thomson would regard as 'instrumental' would also, in typical conversational uses, carry practical force: your car's engine chugs to a halt and I say 'It needs petrol'. Am I not recommending you to get some petrol put in the car? And to follow Thomson's story (p. 6), the petrol is practically necessary because it is a necessary condition of the car's Y-ing. But is its Y-ing a matter of importance, as Thomson says? Maybe it is, maybe it isn't. If you've taken it to the wrecker's yard and are just seeing which bits still work, then probably it isn't; but in this conversational context, it would still carry something like a recommendation to put some petrol in if you want it to Y. This is not because of any normative meaning of need but simply as a result of the fact that I've chosen to say something of the form 'X is necessary for A to Y'. Why (implicitly) refer to Y? Because it might be something you want. If you do, then I've told you something useful which is meant to guide your future action. I am suggesting, then, that there is no more normative content or practical force in the verb need than there is in the verb one uses in saying 'That tickles!' in the hope of obtaining more or less of the same from one's playmate.
Thomson notes (p. 8) another common move which might encourage him in thinking that there is a separate sense of 'fundamental' needs — granted that A needs X for Y, A doesn't need Y, so A doesn't (really) need X. More generally, one might invoke the fact that we do reject some need-statements since if anything is allowed in the Y-slot we could probably find some way of making 'A needs X' true, whatever X may be. But once again it would appear that Grice can come to the rescue. To resuscitate an example used elsewhere (1987, p. 107 ftn.), I do not need a bottle of Lafite each day, because I could not hope to sustain conversational interest in the Ys for which a bottle of Lafite each day is necessary. There is no point thinking about them. Perhaps we could regard many conversations about needs as constrained by a challenge: give us a Y we can take seriously. To return to Thomson's example, A needs money to buy a bracelet, but for what Z does A need a bracelet? None that we can take seriously, so with respect to all serious Z A doesn't need a bracelet and so doesn't need the money either. Harm, as Thomson and Miller recognize, is clearly something we are disposed to take seriously, though surely it is not the only factor we normally use in these contexts.
Thomson's argument here is acknowledged to derive from Wiggins, to whose own discussion we may now turn. Wiggins (1985) notes that arguments about the distinctness of senses are somewhat hazardous, but claims that our understanding of both 'A needs the money for a suit' and 'A doesn't need the money' as together true, or to put it differently, our reluctance to allow the detachment of 'A needs the money' from 'A needs the money to buy a suit' is enough to establish two distinct senses of need. Casting it in my terms, what Wiggins is doing is putting the 'default' values of the Y slot into the meaning of one of his items.4 To the extent that Wiggins' account of a discussion about A's need for money to buy a suit involves, at some stage, a shift from one sense of need to another sense, distinct in its logical properties, I suggest that it erects too rigid a barrier; repeated questions, 'do you really need that?' may continually shift the Y-slot closer to sheer survival, but we do not need to suppose, if Wiggins does, that at some determinate point we have flipped from elliptical instrumental needs to non-elliptical categorical ones.
The integrity of Wiggins' categorical sense of need is seriously compromised by his own subtle reflections on the various relativities involved in his categorical need-statements. The first of these, he says, is a relation to the idea of well-being or flourishing against which we judge harm.5 The second is that "what constitutes suffering or wretchedness or harm is an essentially contestable matter, and is to some extent relative to a culture, even to some extent relative to people's conceptions of suffering, wretchedness and harm" (1985, p. 155). The third is "relativity to the particular circumstances of the time or times associated with the need and the background of (no doubt normative) assumptions associated with those circumstances" (p. 156). Part of this third relativity relates to the fact that needs are typically inus conditions of the Y in question. Wiggins notes that his man's need for money to acquire a suit is not impugned by the fact that he might be able to steal it. But my present point is not to explore these issues themselves; it is simply to note that if his objection to the elliptical analysis is that for 'categorical' needs the Y-slot is given by the meaning of the word 'need', once we recognize his relativities we can surely wonder what exactly we are being given on any particular occasion of use. To the extent that it is slippery, we are going to require just that further specification that the elliptical analysis insists upon.
The same conclusion could be based on Braybrooke's much lengthier discussions of the ins and outs of his "course-of-life" needs (1987). For my own very limited present purpose, Braybrooke does not require much further discussion since his disdain for the elliptical analysis is largely based on a misconception. He is prepared to concede that the relational formula, as he calls it, can be fitted to every need-statement, although "it gives disappointingly unilluminating results when it is applied to certain course-of-life needs, to wit, the most obvious and basic ones" (p. 30). He notes that in such cases where Y is a matter of life or health, we do not have to justify aiming at it;6 but advocates of the elliptical analysis such as White never said that we did. Braybrooke characterizes White's view as one that "makes every need relative to an end calling for justification" (p. 308), but White concludes his discussion by saying "the confusion between the existence and the desirability of an end-state has arisen from a confusion between specifying the end-state in virtue of which the need arises and justifying it. One cannot decide whether A needs X unless one knows what he is alleged to need it for; but one does not have to pronounce on the merits of the latter" (1975, p. 121).
To sum up, then, I have suggested
That is to argue for one way of dealing with the English lexicon; it is also perhaps to hazard the guess that our political (and educational, psychological, and other) thinking will make greater progress by attending to the differences amongst the implicit fillers of the Y-slots than by sinking them in a putative sense of fundamental need. The actual practice of our authors in ferreting out such differences seems to me to belie their attachment to a categorical sense. Nothing I have said is meant to underwrite Thomson's belief that on the elliptical view "all talk of needs relating to human welfare would be inappropriate, and the contrast between essential needs and luxuries would be bogus" (p. 2). Where we may really differ is in my seeing what Thomson offers as one basis for a variety of claims as a possibly equal variety of distinguishable bases. No doubt, as all are agreed, rhetorically the appearance of unity is more persuasive; but for accuracy's sake, the diversity should be acknowledged. Thomson notes the dangers of talk of needs but says that virtually any other terminology is also open to abuse. True, but when he says that anyone using the concept of a need "should treat the term cautiously and be alert for ambiguities" (p. 102), or when Braybrooke, noting relativities analogous to those mentioned by Wiggins, says that "before one proceeds with questions about individual items, one needs to ask where are we, questioner and respondent, with respect to the schema?" (pp. 120-1), we are simply being given the same warning the elliptical analysis is designed to emphasize. We are being told to get straight about the Y-slot, except that these authors wish to believe there isn't really a Y-slot there in these cases. Sharing Thomson's wish for a unified account of the semantics of 'need', it is surely better to tell the same story when the same story has to be told somehow.
Footnotes
1. See my 1986, p. 96, for some discussion of Swinburne's claim that there is a fourth relatum (a frame of reference) here too. Return to text.
2. I am restricting this discussion to the verb, and ignoring Thomson's remarks about the noun, need. He seems to gloss over the distinction White (1975) made between cases in which a need exists and the more restricted set of cases in which A has a need; but it may be a case of differing intuitions of grammatical acceptability. Return to text.
3. For instance, in his magisterial contribution, White (1975) continually speaks of the "end-state" for which something is needed, although it is clear from his discussion that this should be understood simply as a label for the Y-slot in 'A needs X for/to Y' since he notes that in many cases A needs X to be an A, and that the X may be related to Y logically, legally, according to the regulations, etc. (p. 105). Return to text.
4. As he did in an earlier paper he refers to (1971) with respect to the word 'good' in 'She has good legs'. There he claimed that the only permissible interpretations of that as a complete sentence were from an aesthetic, functional, or medical perspective, although any number of other perspectives could be introduced explicitly (to play the part of a dwarf; to eat;...). Speaking for myself, the de Bonoesque perspectives could all be supplied by the appropriate context; those Wiggins plumps for might be the only ones available to a virtually decontextualized audience and so be the 'default' values of a slot that is always there. Return to text.
5. While writing of "some however minimal level of flourishing" (p. 156), it seems to me that Wiggins tends to operate with a binary opposition between harm and flourishing when our ordinary thinking makes a three way contrast: harm; average existence; flourishing. In an obscure passage, he claims he is reluctant to say that human beings need to flourish but happy to say (equivalently?) that they need to avoid harm (p. 166). But these do seem rather different; and flourishing, as against living unharmed, does not seem to be the sort of non-negotiable good Wiggins posits. Thomson allows that it is worse to lack what one needs than to fail to acquire an unneeded benefit, and maintains that the latter is distinct from a luxury which is superfluous to our well-being (p. 96); but once again he appears to regard anything above the needs-threshold as some sort of flourishing, and his talk of rich and poor would allow the 'middle sort' to be too easily overlooked. Return to text.
6. Thomson contrives to tell a story in which someone can sensibly be said to need to survive for a little while longer (pp. 19-20); but the general point he and Braybrooke make that we do not normally regard living as a matter of need (or perhaps more persuasively that the question 'Does A need to live?' is usually inappropriate and pointless) can easily be accommodated by the Gricean picture I have adumbrated: since, as Thomson notes, the value of our lives is not tied to specific projects, we should not offer, so cannot presume audiences to supply, appropriate Ys to make informative truth out of 'A needs to live'. Return to text.
REFERENCES
Barry, B. (1965). Political Argument. London: Routledge & Kegan Paul.
Barwise, J. (1986). Conditionals and conditional information. In E.C. Traugott, A. ter Meulen, J.S. Reilly, and C.A. Ferguson (Eds.), On Conditionals, Cambridge: Cambridge University Press.
Brandon, E.P. (1980). O reason not the need. Education for Development, 6, 18-25.
Brandon, E.P. (1986). Ellipsis: history and prospects. Informal Logic, 8, 93-103.
Brandon, E.P. (1987). Ellipsis and ideology. In D.N. Perkins, J. Lochhead, and J.C. Bishop (Eds.), Thinking: the Second International Conference, Hillsdale, NJ: Lawrence Erlbaum.
Braybrooke, D. (1987). Meeting Needs. Princeton: Princeton University Press.
Hamblin, C.L. (1970). Fallacies. London: Methuen.
Johnson, W.E. (1921). Logic, Part I. Cambridge: Cambridge University Press.
Mackie, J.L. (1965). Causes and conditions. American Philosophical Quarterly, 2, 245-264.
Miller, D. (1976). Social Justice. Oxford: Oxford University Press.
Thomson, G. (1987). Needs. London: Routledge & Kegan Paul.
White, A.R. (1975). Modal Thinking. Oxford: Basil Blackwell.
Wiggins, D. (1971). Sentence sense, word sense and difference of word sense. In D. Steinberg and D. Jacobovits (Eds.), Semantics: An Interdisciplinary Reader, Cambridge: Cambridge University Press.
Wiggins, D. (1985). The claims of need. In T. Honderich (Ed.), Morality and Objectivity, London: Routledge & Kegan Paul.
HTML prepared December 17th, 1999. Return to publications; Return Home.
URL: http://www.uwichill.edu.bb/bnccde/epb/graz.html