Position Paper on Common-sense and Formal Semantics 
Geoffrey Nunberg 
Xerox PARC and CSLI, Stanford 
1. A philological excursus 
I'm not sure what I'm doing on this panel, but I thought it would be 
helpful if we could start at the beginning. It's interesting to note that both 
the dictionary and common sense were eighteenth-century inventions. This 
is no coincidence; in fact, it's entirely appropriate that the most celebrated 
blow that was ever struck on behalf of common sense was delivered by Dr. 
Johnson himself, when he kicked a stone in refutation of Berkeley's 
idealism. If you'll indulge me, I'd like briefly to explore this connection, with 
the promise that I'll wind up by drawing a moral about n_atural-language 
understanding. 
First for "common sense." To be sure, the phrase had been in use 
since the fourteenth century, but it was not until the eighteenth century that 
it acquired three important philosophical uses, which we confuse at our 
peril. First, there was the understanding of common sense as the faculty of 
judgement possessed by the average person as part of his birthright. This is 
the sense of the phrase that Priestly had in mind when he wrote (1775): 
"Common sense...in common acceptation...has long been appropriated...to 
that capacity for judging of common things that persons of middling 
capacities are capable of." In this sense, common sense is variously a 
general or universal faculty, "the same in every time and clime," as Hume 
(somewhere) put it; to avoid confusion here, I'll sometimes refer to this as 
the faculty of common sense. Now for many philosophers of the period, it 
was an article of doctrine that natural and moral law were accessible to 
common sense, and hence that the body of common man could achieve 
both the degree of understanding of the world and the refinement of moral 
judgment necessary to self-government. This was the view of Hume, of 
Johnson, and of course of Thomas Paine, who entitled his famous treatise 
on natural rights Common Sense. 
The second use of the phase was more narrowly associated with the 
Scottish School of philosophy(also called the "Philosophers of Common 
Sense"); on this understanding it is a universal endowment, not just of 
critical faculties, but of beliefs. As the philosopher Hamilton put it, common 
sense is in this interpretation is "the complement of those cognitions or 
convictions which we receive from nature; which all men possess in 
common .... " This group made common sense the test of all philosophical 
doctrines, and in particular referred to the common-sense belief in the 
reality of the physical world in the course of their Berkeley.bashing (a 
129 
position to which Hume, unlike Johnson, was not sympathetic). More recently, this 
is the sense of the phrase that a lot of AI researchers appear to have in mind when 
they talk about "common-sense knowledge" as a body of generally accepted beliefs 
about the world. 
The third eighteenth-century use of "common sense" is somewhat archaic 
nowadays, and usually appeared with the definite article, as "the common sense." 
Here, the common sense is the body of beliefs and values possessed by a community 
as a whole--what we might now call "received wisdom," or "custom" or perhaps 
what Durkheim referred to as the "collective representation." For now, I'll call this 
the "collective sense," to avoid confusion. Roughly, you could think of this as a kind 
of systematic representation that exists only collectively, and which is realized only in 
interaction. 
Now for the dictionary. Before the eighteenth century, such dictionaries as 
existed were mostly lists of "hard words," and no one seemed to mind the absence 
of a dictionary of the modern sort, which would set out the rules for using all the 
words in general currency. It's a complicated matter to explain why this changed in 
Johnson's time. In part, he was animated by a widespread contemporary Sentiment 
that knowledge was expanding and becoming specialized beyond the ability of any 
one person to master it (recall that this was also the time at which the "division of 
labor" entered the general discussion). But he was also sensitive to the doctrines 
about common sense and politics that I mentioned a moment ago. What he hoped to 
do, finding the language "copious without order," was nothing less than to 
systematize the collective sense of the English-speaking community, or at least such 
part of it as was embodied in the uses of words, so as to make it generally accessible 
to the ordinary reader. This step was crucial to the democratic program, which 
relied on the ability of the common man to coordinate an informed discourse about 
the world as it was and as it ought to be; to use words, if not in exactly the same 
ways, then at least in line with general practice. 
Note that the possibility of succeeding in this enterprise rested on two 
assumptions about the role that the individual faculty of common sense would have 
to play in the process. First, as I noted earlier, Johnson and his contemporaries 
assumed that the ordinary individual was capable of grasping that part of the 
collective sense embodied in the definition (or as he called it, the "explanation") of a 
word; as he put it in an essay in the Idler, the most obscure doctrines of philosophy 
and science would be "found to contain nothing more than very plain truths," which 
could be "delivered in p!ain language." The second assumption is somewhat subtler. 
Johnson had to assume as well that readers would come generally to acknowledge 
the authority of his dictionary, or in more humility, the authority of the authors he 
cited in explication of the uses of words. So far as I know, Johnson himself never 
addressed this problem head-on, but the importance of common sense in this 
process was acknowledged by his contemporary George Campbell, a rhetorician 
associated with the common-sense school, in his Philosophy ot Rhetoric (1776): 
"The source...of that preference which distinguisheth good use from bad in 
language, is a natural propensity of the human mind to believe that those are the best 
judges of the proper signs, and of the proper applications of them, who understand 
best the things which they represent." It is fair to say, then, that eighteenth-century 
130 
beliefs about the faculty of common sense both made the Dictionary necessary, and 
made it possible. 
2. Common sense and the social 
Now let me wrench this out of context. What makes the eighteenth century 
interesting from a linguistic point of view is not the peculiarities of its situation--the 
fact, for example, that an actual written dictionary was first produced then, or that the 
enterprise was connected to explicit political notions. The latter is important only in 
that it forced people to become self-consciously aware of certain problems of 
language, and set up a scheme for taking them on. (Analogously, it was in the 
eighteenth century that philosophers first became aware of the market as a social 
type, thouch certainly markets themselves had been around long before.) In 
particular, they made an explicit effort to talk about problems of linguistic 
coordination. And stripped of local details, their formulation of the situation is 
applicable to all discourse in all languages. You could say that all linguistic 
communities are democracies, in the sense that deference to linguistic authority is 
always consensual. 
If these questions haven't loomed very large in recent research in linguistics, 
AI and related fields, it's only because those fields have been wed to a doctrine of 
what Putnam calls "methodological solipsism." I'll come back to this point presently, 
but right now I want to try to show how these considerations bear on problems of 
natural-language interpretation. 
Suppose we put the problem in a schematic way. On the one hand, you have 
this extensive body of knowledge and assumptions--the collective sense-- which 
underlies the use of natural-language expressions. A part of this knowledge is 
actually possessed by all discourse participants when they interpret utterances--this 
is what constitutes their "common-sense beliefs" in the accepted use of the term. 
But the rest of this material is not represented by most users when they use most 
expressions, though parts of it are available to some speakers in some contexts. And 
yet this information is at least tacitly accepted as licencing all uses of all expressions. 
Most talk, that is, is carried out on credit, not cash on the barrel head, but the 
success of particular exhanges, as well as the overall coordination of the larger 
discourse, depends crucially on a general belief that the credit is sound; that the 
linguistic chits can be cashed in if circumstances required. And as with any sound 
currency--and linguistic currency is more stable even that the Swiss Franc--the 
interesting and somewhat paradoxical consequence is that people almost never do 
call in the notes, because it's taken for granted that they're good. 
Some examples may help. Now the only area in which the social 
determination of meaning has been considered at all is in the use of natural-kind 
terms (though no one seems to know what to make of these, either). But I want to 
stay away from these cases, because in fact the situation here is rather complicated, 
and is obscured in Putnam's account. What is remarkable is that no one has given 
much thought to the fairly obvious point that the linguistic division of labor is 
pervasive in any direction you look. Take words from any domain: grade-A prime, 
felony, squeeze play, sentence, jazz, braise, or hacker; you'll find that most people 
131 
use these words in imperfect knowledge of the things they are held to denote, and in 
implicit deference to the collective sense that regulates their use. 
This in turn raises several questions for theories of natural-language 
understanding. First, how do people get away with talking abOdt the world in the 
absence of what economists would call "perfect information" about the meanings of 
the words they use? The answer is that the task of regulating words and world is 
taken care of socially. I don't know how to make a Peking Duck (or do we call it a 
Beijing Duck nowadays?), not even declaratively. But man and boy, I can't count the 
number of times I've succeeded in ordering Peking Ducks, or recommending Peking 
Ducks, or warning people off Peking Ducks. And I can get away with this because 
there's someone in the back room whom I trust to keep the phrase "Peking Duck" in 
line with its denotation. 
How do you go about assigning an interpretation to my utterances about 
Peking Ducks, then? There are several aspects to the problem. In the first place, 
how do you decide what information I associate with the phrase when I use it, and 
hence what inferences you're entitled to draw about my internal state? Note that this 
is not the same as asking how much I know about Peking Ducks. Nor is it the same 
as asking how much my interlocutor and I can presuppose as common knowledge 
about Peking Ducks. To take another example (since as I said, I don't know a lot 
about Peking Ducks), suppose I'm a gallery owner and I'm trying to sell a particular 
acrylic painting to a client. I may know all sorts of things about acrylics: that they're 
quick-drying, easy to wash off, less noxious and less smelly than oil paints, and 
lacking in a certain tactility. And maybe the client knows that too. But in this context, 
none of that information is relevant to our interests, and you'd be making a big 
mistake if you tried to grind out all of the inferences that that kind of knowledge gives 
rise to. Whereas if I'm i'ecommending a shift from oils to acrylics for a painter who 
has just come down with a rash, you would have to include some of that information 
to get the interpretation right. 
In another way, however, the collective sense does play a role in the 
interpretation of all utterances, even when I am ignorant of it. Whatever my internal 
state vis-a-vis the world, I make certain social commitments about the world when I 
use an expression, and these are determined by the collective sense. Say you go into 
a restaurant and order a bottle of Sauternes, knowing only that it's a white wine; then 
when it comes, you discover that it's very sweet. Now you can ask the waiter to take 
it back on those grounds, I suppose, and he will oblige you if he's a good guy, but 
you'll be a little embarrassed about the exchange, because you realize he is not 
obliged to make the world conform to your idiosyncratic representation of it. He 
could say to you, "Listen, you ordered Sauternes, you got Sauternes. If you don't 
like it, get up, I need the table." 
Now suppose we ask what kinds of linguistic authority we recognize as 
legitimate, and on what grounds? Or to put it another way, what kinds of information 
do we put in the collective sense? Here is where Johnson's assumptions about 
common sense come into play, and where Putnam comes a cropper. Because you 
can't just stick in any information that experts happen to have on hand; in fact, you 
can't even say by fiat who the experts are going to be. Suppose there's some weird 
kind of grape that's a member of the same biological kind as the grape that 
132 
Sauternes wine is made from, and suppose that that grape produces a wine that 
tastes and looks exactly like Gatorade. If the waiter brings me a bottle of that stuff, I 
send it back with full peremtory indignation, because I am certainly not committed to 
defer to biological authority in this matter if it is indifferent to my gastronomic 
interests. I defer to authority only where common sense tells me it is in my interest to 
do so. And 1 require of such authorty that it define categories in ways that are 
consistent with my common-sense appraisal of what those interests are. Until you 
have fleshed out this story about common-sense deference, you can't determine just 
what commitments people are undertaking when they talk about the world. 
3. A Note on Common-sense and formal semantics 
As I noted, nobody has spent much time addressing the sorts of problems I've 
touched on here. The reason for this, I would suggest, is that linguistics and AI have 
been working under a certain methodological assumption that makes these things 
hard to get at. As Chomsky put it in his well known formulation of the goals of 
linguistic theory, linguistics is concerned with the competence of "an ideal 
speaker-listener in a homogeneous speech-community, who knows its language 
perfectly .... " And while virtually every other clause of this formulation has been 
raked over by one or another critic, this one has escaped almost unnoticed, for good 
reason. The fact is that when you look at the sorts of linguistic phenomena most 
formal linguists have been interested in, and especially when you look at them with 
an eye towards describing their, systematic formal properties, it isn't going to be 
particularly interesting to consider the problems raised by variation and imperfect 
knowledge. For one thing, variation in these domains doesn't often coincide with 
any interesting differences in social interests or social knowledge. Then too, from a 
purely structural point of view, an incomplete system is not interestingly different 
from a complete one. Questions about the social determination of meaning only 
arise when you look at domains in which linguistic representations co-vary in an 
interesting way with the social differentiation of knowledge. 
This suggests an interesting way of understanding a part of the distinction 
between the methodologies of formal and common-sense semantics. I suspect that 
formal semantics is capable of making substantial progress in just those domains in 
which the social differentiation of knowledge is not relevant; in those domains, that 
is, in which common-sense beliefs and the collective sense amount to pretty much 
the same thing. But once you leave those domains for the wilds of the lexicon, you 
are going to have to do a lot of common-sense inferencing before you can even say 
what the meanings of expressions are, and how those meanings are relevant in their 
various ways to utterance interpretation. 
133 
