BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 273

BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 274

A Philosopher’s View: The Toulmin Model

All my ideas hold together, but I cannot elaborate them all at once. — JEAN-JACQUES ROUSSEAU

Clarity has been said to be not enough. But perhaps it will be time to go into that when we are within measurable distance of achieving clarity on some matter.


[Philosophy is] a peculiarly stubborn effort to think clearly. — WILLIAM JAMES

Philosophy is like trying to open a safe with a combination lock: Each little adjustment of the dials seems to achieve nothing, only when everything is in place does the door open.


In Chapter 3, we explained the contrast between deductive and inductive arguments to focus on the two main ways in which we reason, either

• Making explicit something concealed in what we already accept (deduction) or

• Using what we have observed as a basis for asserting or pro- posing something new (induction).

Both types of reasoning share some structural features, as we also noticed. Thus, all reasoning is aimed at establishing some thesis (or



BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 275

conclusion) and does so by means of some reasons. These are two basic characteristics that any argument contains.

After a little scrutiny we can in fact point to several features shared by all arguments, deductive and inductive, good and bad alike. We use the vocabulary popularized by Stephen Toulmin, Richard Rieke, and Allan Janik in their book An Introduction to Reasoning (1979; second edition 1984) to explore the various ele- ments of argument.


Every argument has a purpose, goal, or aim — namely, to establish a claim (conclusion or thesis). Suppose you were arguing in favor of equal rights for women. You might state your thesis or claim as follows:

Men and women ought to have equal rights.

A more precise formulation of the claim might be

Men and women ought to have equal legal rights.

A still more precise formulation might be

Equal legal rights for men and women ought to be protected by our Constitution.

The third version of this claim states what the controversy in the 1970s over the Equal Rights Amendment was all about.

Consequently, in reading or analyzing someone else’s argu- ment, your first question should naturally be: What is the argu- ment intended to prove or establish? What claim is it making? Has this claim been clearly and precisely formulated, so that it unam- biguously asserts what its advocate wants to assert?


Once we have the argument’s purpose or point clearly in mind and thus know what the arguer is claiming to establish, then we can ask for the evidence, reasons, support—in short, for the grounds—on which that claim is based. In a deductive argument these grounds are the premises from which the claim is deduced; in


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 276

an inductive argument the grounds are the evidence—a sample, an observation, or an experiment—that makes the claim plausible or probable.

Not every kind of claim can be supported by every kind of ground, and conversely, not every kind of ground gives equally good support for every kind of claim. Suppose I claim that half the students in the classroom are women. I can ground this claim in any of several ways.

1. I can count all the women and all the men. Suppose the total equals fifty. If the number of women is twenty-five and the number of men is twenty-five, I have vindicated my claim.

2. I can count a sample of, say, ten students and find that in the sample five of the students are women. I thus have induc- tive—plausible but not conclusive—grounds for my claim.

3. I can point out that the students in the college divide equally into men and women and claim that this class is a represen- tative sample of the whole college.

Obviously, ground 1 is stronger than ground 2, and 2 is far stronger than ground 3.

So far we have merely restated points about premises and con- clusions covered in Chapter 3. But now we want to consider four additional features of arguments.


Once we have the claim or the point of an argument fixed in mind and the evidence or reasons offered in its support, the next question to ask is why these reasons support this conclusion. What is the warrant, or guarantee, that the reasons proffered do support the claim or lead to the conclusion? In simple deductive arguments, the warrant takes different forms, as we shall see. In the simplest cases, we can point to the way in which the meanings of the key terms are really equivalent. Thus, if John is taller than Bill, then Bill must be shorter than John because of the meaning in English of “is shorter than” and “is taller than.” In this case, the warrant is something we can state quite literally and explicitly.

In other cases, we may need to be more resourceful. A reliable tactic is to think up a simple parallel argument—that is, an argument exactly parallel in form and structure to the argument we are trying to defend. We then point out that if we are ready to accept the


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 277

simpler argument, then we must accept the more complex argu- ment because both arguments have exactly the same structure. For example, in her much-discussed 1972 essay on the abortion contro- versy, “A Defense of Abortion,” philosopher Judith Thomson argues that a pregnant woman has the right to an abortion to save her life, even if it involves the death of her unborn child. She anticipates that some readers may balk at her reasoning, and so she offers this parallel argument: Suppose you are locked in a tiny room with another human being, which through no fault of its own is growing uncontrollably, with the result that it is slowly crushing you to death. Of course, it would be morally permissible to kill the other person to save your own life. With the reader’s presumed agreement on that conclusion, the parallel argument concerning the abortion situation—so Thomson hopes—is obvious and convincing.

In simple inductive arguments, we are likely to point to the way in which observations or sets of data constitute a representative sample of a whole (unexamined) population. Here, the warrant is the representativeness of the sample. For instance, in projecting a line on a graph through a set of points, we defend one projection over alternatives on the grounds that it makes the smoothest fit through most of the points. In this case, the warrant is simplicity and inclusiveness. Or in defending one explanation against compet- ing explanations of a phenomenon, we appeal to the way in which the preferred explanation can be seen as a special case of generally accepted physical laws. Examples of such warrants for inductive reasoning will be offered in following pages (see Chapter 9, A Logician’s View: Deduction, Induction, Fallacies, pp. 289–339).

Establishing the warrants for our reasoning—that is, explain- ing why our grounds really support our claims—can quickly become a highly technical and exacting procedure that goes far beyond what we can hope to explain in this book. Only a solid course or two in formal deductive logic and statistical methods can do justice to our current state of knowledge about these warrants. Developing a “feel” for why reasons or grounds are or are not rele- vant to what they are alleged to support is the most we can hope to do here without recourse to more rigorous techniques.

Even without formal training, however, one can sense that some- thing is wrong with many bad arguments. Here is an example. British professor C. E. M. Joad found himself standing on a station platform, annoyed because he had just missed his train, when another train, making an unscheduled stop, pulled up to the platform in front of him. He decided to jump aboard, only to hear the porter say “I’m


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 278

afraid you’ll have to get off, sir. This train doesn’t stop here.” “In that case,” replied Joad, “don’t worry. I’m not on it.”


The kinds of reasons appropriate to support an amendment to the Constitution are completely different from the kinds appropriate to settle the question of what caused the defeat of Napoleon’s inva- sion of Russia. Arguments for the amendment might be rooted in an appeal to fairness, whereas arguments about the military defeat might be rooted in letters and other documents in the French and Russian archives. The canons of good argument in each case derive from the ways in which the scholarly communities in law and his- tory, respectively, have developed over the years to support, defend, challenge, and undermine a given kind of argument. Thus, the support or backing appropriate for one kind of argument might be quite inappropriate for another kind of argument.

Another way of stating this point is to recognize that once you have given reasons for a claim, you are then likely to be challenged to explain why these reasons are good reasons—why, that is, one should believe these reasons rather than regard them skeptically. Why (a simple example) should we accept the testimony of Dr. X when Dr. Y, equally renowned, supports the opposite side? Or why is it safe to rest a prediction on a small though admittedly carefully selected sample? Or why is it legitimate to argue that (1) if I dream I am the King of France, then I must exist, whereas it is illegitimate to argue that (2) if I dream I am the King of France, then the King of France must exist? To answer these kinds of challenges is to back up one’s reasoning, and no argument is any better than its backing.


As we have seen, all arguments are made up of assertions or propo- sitions, which can be sorted into four categories:

• The claim (conclusion, thesis to be established),

• The grounds (explicit reasons advanced),

• The warrant (the principle that connects the ground to the claim), and

• The backing (implicit assumptions).


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 279

All these kinds of propositions have an explicit or tacit modality in which they are asserted, indicating the scope and character with which they are believed to hold true. Is the claim, for instance, believed to be necessary —or only probable? Is the claim believed to be plausible —or only possible? Of two reasons for a claim, both may be good, but one may be better than the other. Indicating the modal- ity with which an assertion is advanced is crucial to any argument for or against it.

Empirical generalizations are typically contingent on various fac- tors, and it is important to indicate such contingencies to protect the generalization against obvious counterexamples. Thus, consider this empirical generalization:

Students do best on final examinations if they study hard for them.

Are we really to believe that students who study regularly through- out the whole course and so do not need to cram for the final will do less well than students who neglect regular work in favor of sev- eral all-nighters at the last minute? Probably not; what is really meant is that all other things being equal (in Latin, ceteris paribus), concentrated study just before an exam will yield good results. Alluding to the contingencies in this way shows that the writer is aware of possible exceptions and that they are conceded right from the start.

Assertions also have varying scope, and indicating their scope is equally crucial to the role that an assertion plays in argument. Thus, suppose you are arguing against smoking, and the ground for your claim is this:

Heavy smokers cut short their life span.

Such an assertion will be clearer, as well as more likely to be true, if it is explicitly quantified. Here, there are three obvious alterna- tive quantifications to choose among: all smokers cut short their life span, most do, or only some do. Until the assertion is quantified in one of these ways, we really do not know what is being asserted — and so we do not know what degree and kind of evi- dence and counterevidence is relevant. Other quantifiers include few, rarely, many, often, sometimes, perhaps, usually, more or less, regu- larly, occasionally.

In sum, sensitivity to the quantifiers and qualifiers appropriate for each of our assertions, whatever their role in an argument, will


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 280

help prevent you from asserting exaggerations and other misguided generalizations.


Very few arguments of any interest are beyond dispute, conclu- sively knockdown affairs in which the claim of the argument is so rigidly tied to its grounds, warrants, and backing and its quantifiers and qualifiers so precisely orchestrated that it really proves its con- clusion beyond any possibility of doubt. On the contrary, most arguments have many counterarguments, and sometimes one of these counterarguments is the most convincing.

Suppose one has taken a sample that appears to be random: An interviewer on your campus accosts the first ten students she encounters, and seven of them happen to be fraternity or sorority members. She is now ready to argue that seven-tenths of enrolled students belong to Greek organizations.

You believe, however, that the Greeks are in the minority and point out that she happens to have conducted her interview around the corner from the Panhellenic Society’s office just off Sorority Row. Her random sample is anything but. The ball is now back in her court as you await her response to your rebuttal.

As this example illustrates, it is safe to say that we do not under- stand our own arguments very well until we have tried to get a grip on the places in which they are vulnerable to criticism, counterat- tack, or refutation. Edmund Burke (quoted in Chapter 3 but worth repeating) said, “He that wrestles with us strengthens our nerves, and sharpens our skill. Our antagonist is our helper.” Therefore, cul- tivating alertness to such weak spots, girding one’s loins to defend at these places, always helps strengthen one’s position.


To see how the Toulmin method can be used, let’s apply it to an argument in this book, Susan Jacoby’s “A First Amendment Junkie” (p. 43).

The Claim Jacoby’s central thesis or claim is this: Any form of censorship — including feminist censorship of pornography in particular — is wrong.


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 281

Grounds Jacoby offers six main reasons or grounds for her claim, roughly in this sequence (but arguably not in this order of importance).

First, feminists exaggerate the harm caused by pornography because they confuse expression of offensive ideas with harmful conduct.

Second, letting the government censor the expression of ideas and attitudes is the wrong response to the failure of parents to con- trol the printed materials that get into the hands of their children.

Third, there is no unanimity even among feminists over what is pornography and what isn’t.

Fourth, permitting censorship of pornography to please feminists could well lead to censorship on many issues of concern to feminists (“rape, abortion, menstruation, contraception, lesbianism”).

Fifth, censorship under law shows a lack of confidence in the democratic process.

Finally, censorship of words and pictures is suppression of self- expression, and that violates the First Amendment.

Warrants Each of these six grounds needs its own warrant, and the warrants vary considerably in their complexity. Jacoby (like most writers) is not so didactic as to make these warrants explicit. Taking them in order, this is what they look like.

First, since the First Amendment protects speech in the broad- est sense, the censorship that the feminist attack on pornography advocates is inconsistent with the First Amendment.

Second, if feminists want to be consistent, then they must advocate censorship of all offensive self-expression, but such a radi- cal interference with free speech (amounting virtually to repeal of the First Amendment) is indefensible.

Third, if feminists can’t agree over what is pornographic, the cen- sorship of pornography they propose is bound to be arbitrary.

Fourth, feminists ought to see that they risk losing more than they can hope to gain if they succeed in censoring pornography.

Fifth, the democratic process can be trusted to weed out harm- ful utterances.

Sixth, if feminists have a legal right to censor pornography, antifeminists will claim the same right on other issues.

Backing Why should the reader agree with Jacoby’s grounds? She does not appeal to expert authority, the results of experimental tests or other statistical data, or the support of popular opinion.


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 282

Instead, she relies principally on two things—but without saying so explicitly.

First, she assumes that the reader accepts the propositions that freedom of self-expression is valuable and that censoring self-expression requires the strongest of reasons. If there is no fundamental agreement on these propositions, several of her reasons cease to support her claim.

Second, she relies on the reader’s open-mindedness and will- ingness to evaluate common sense (untechnical, ordinary, familiar) considerations at each step of the way. She relies also on the reader having had some personal experience with erotica, pornography, and art. Without that open-mindedness and experience, a reader is not likely to be persuaded by her rejection of the feminist demand for censorship.

Modal Qualifiers Jacoby defends what she calls an “absolute interpretation” of the First Amendment—that is, the view that all censorship of words, pictures, and ideas is not only inconsistent with the First Amendment but is also politically unwise and morally objectionable. She allows that some pornography is highly offensive (it offends her, she insists); she allows that some pornogra- phy (“kiddie porn”) may even be harmful to some viewers. But she also insists that more harm than good would result from the censor- ship of pornography. She points out that some paintings of nude women are art, not pornography; she implies that it is impossible to draw a sharp line between permissible erotic pornography and impermissible offensive pornography. She clearly believes that all Americans ought to understand and defend the First Amendment under the “absolute interpretation” she favors.

Rebuttals Jacoby mentions several objections to her views, and perhaps the most effective aspect of her entire argument is her skill in identifying possible objections and meeting them effectively. (Notice the diversity of the objections and the various ways in which she replies.)

Objection: Some of her women friends tell her she is wrong.

Rebuttal: She admits she’s a “First Amendment junkie,” and she doesn’t apologize for it.

Objection: “Kiddie porn” is harmful and deserves censorship.

Rebuttal: Such material is not protected by the First Amendment because it is an “abuse of power” of adults over children.


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 283

Objection: Pornography is a form of violence against women, and therefore it is especially harmful.

Rebuttal: (1) No, it really isn’t harmful, but it is disgusting and offensive. (2) In any case, it’s surely not as harmful as allowing American neo-Nazis to parade in Jewish neighbor- hoods. (Jacoby is referring to the march in Skokie, Illinois, in 1977, upheld by the courts as permissible political expression under the First Amendment despite its offensiveness to sur- vivors of the Nazi concentration camps.)

Objection: Censoring pornography advances public respect for women.

Rebuttal: Censoring Ms. magazine, which antifeminists have already done, undermines women’s freedom and self- expression.

Objection: Reasonable people can tell pornography when they see it, so censoring it poses no problems.

Rebuttal: Yes, there are clear cases of gross pornography; but there are lots of borderline cases, as women themselves prove when they disagree over whether a photo in Penthouse is offen- sively erotic or “lovely” and “sensuous.”


✓ A CHECKLIST FOR USING THE TOULMIN METHOD Have I asked the following questions? � What claim does the argument make? � What grounds are offered for the claim? � What warrants the inferences from the grounds to the claim? � What backing supports the claim? � With what modalities are the claim and grounds asserted? � To what rebuttals are the claim, grounds, and backing vulnerable?

See the companion Web site for links related to the Toulmin model.

BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 284


Let’s look at an argument—it happens to be a proposal concerning illegal immigration—and see how the Toulmin method can be applied.

Michael S. Dukakis and Daniel J. B. Mitchell

Michael S. Dukakis, a professor of political science at Northeastern University, served as the governor of Massachusetts from 1975 to 1979 and from 1983 to 1991. Daniel J. B. Mitchell is a professor of management and public policy at the University of California at Los Angeles. The essay that follows originally appeared in the New York Times (July 25, 2006).

Raise Wages, Not Walls

There are two approaches to illegal immigration currently being debated in Congress. One, supported by the House, empha- sizes border control and law enforcement, including a wall along the Mexican border and increased border patrols. The other, which is supported by the Bush administration and has been passed by the Senate, relies on employers to police the workplace. Both pro- posals have serious flaws.

As opponents of the House plan have rightly pointed out, walls rarely work; illegal immigrants will get around them one way or another. Unless we erect something akin to the Berlin Wall, which would cost billions to build and police, a barrier on the border would be monitored by largely symbolic patrols and easily evaded.

The Senate approach is more realistic but it, too, has problems. It creates a temporary worker program but requires employers first to attempt to recruit American workers to fill job openings. It allows for more border fencing, but makes no effort to disguise the basic futility of the enterprise. Instead, it calls on employers to enforce immigration laws in the workplace, a plan that can only succeed through the creation and distribution of a costly national identification card.


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 285

A national ID card raises serious questions about civil liberties, but they are not the sole concern. The cost estimates for producing and distributing a counterfeit-proof card for the roughly 150 mil- lion people currently in the labor force—and the millions more who will seek work in the near future—extend into the billions of dollars. Employers would have to verify the identity of every American worker, otherwise the program would be as unreliable as the one in place now. Anyone erroneously denied a card in this bureaucratic labyrinth would be unemployable.

There is a simpler alternative. If we are really serious about turning back the tide of illegal immigration, we should start by rais- ing the minimum wage from $5.15 per hour to something closer to $8. The Massachusetts legislature recently voted to raise the state minimum to $8 and California may soon set its minimum even higher. Once the minimum wage has been significantly increased, we can begin vigorously enforcing the wage law and other basic labor standards.

Millions of illegal immigrants work for minimum and even sub-minimum wages in workplaces that don’t come close to meet- ing health and safety standards. It is nonsense to say, as President Bush did recently, that these jobs are filled by illegal immigrants because Americans won’t do them. Before we had mass illegal immigration in this country, hotel beds were made, office floors were cleaned, restaurant dishes were washed and crops were picked—by Americans.

Americans will work at jobs that are risky, dirty, or unpleasant so long as they provide decent wages and working conditions, especially if employers also provide health insurance. Plenty of Americans now work in such jobs, from mining coal to picking up garbage. The difference is they are paid a decent wage and provided benefits for their labor.

However, Americans won’t work for peanuts, and these days the national minimum wage is less than peanuts. For full-time work, it doesn’t even come close to the poverty line for an individual, let alone provide a family with a living wage. It hasn’t been raised since 1997 and isn’t enforced even at its currently ridiculous level.

Yet enforcing the minimum wage doesn’t require walling off a porous border or trying to distinguish yesterday’s illegal immigrant from tomorrow’s “guest worker.” All it takes is a willingness by the federal government to inspect workplaces to determine which employers obey the law.

Curiously, most members of Congress who take a hard line on immigration also strongly oppose increasing the minimum wage,




BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 286

claiming it will hurt businesses and reduce jobs. For some reason, they don’t seem eager to acknowledge that many of the jobs they claim to hold dear are held by the same illegal immigrants they are trying to deport.

But if we want to reduce illegal immigration, it makes sense to reduce the abundance of extremely low-paying jobs that fuels it. If we raise the minimum wage, it’s possible some low-end jobs may be lost; but more Americans would also be willing to work in such jobs, thereby denying them to people who aren’t supposed to be here in the first place. And tough enforcement of wage rules would curtail the growth of an underground economy in which both illegal immigration and employer abuses thrive.

Raising the minimum wage and increasing enforcement would prove far more effective and less costly than either proposal cur- rently under consideration in Congress. If Congress would only remove its blinders about the minimum wage, it may see a plan to deal effectively with illegal immigration, too.


At first blush, what we have in this essay is a twelve-paragraph argument divided into two unequal parts. Paragraphs 1–4 offer some proposals for dealing with illegal immigration and reasons why these proposals won’t work. This preliminary material is fol- lowed by paragraphs 5–12 in which the proposal favored by the authors is introduced, explained, and justified. So much for first impressions.

Let’s now deconstruct this essay by identifying each of the six elements that constitute the Toulmin method.

• First and foremost, what is the claim being made, the main thesis of the essay? Is it in the title? Is it in paragraph 5? Or is it elsewhere? What kind of claim is it—a claim of fact? A claim of value? In any case, write down the claim for further reference.

• Second, what are the grounds, the evidence or reasons advanced in support of the claim? Partly they can be found in what the authors regard as ineffective alternative efforts in paragraphs 1–4. What are these alternatives? Why are they said to be ineffective? Look also at paragraph 5 and later paragraphs. Write down the sentences you have discovered that are playing this role.


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 287

• Third, what are the warrants that Dukakis and Mitchell rely on to carry the burden of their argument? In paragraph 5, for example, the authors rely on examples from Massachusetts and California. The evidence they offer amounts to a minimal inductive argument. Could their argument of this sort be stronger? Paragraphs 6 and 7 rely on general knowledge, as do most of the rest of their argument. Is that the best one can do with this issue? Can you think of ways in which the authors’ argument can be strengthened? Carefully look through the whole essay for whatever evidence you can find of the men- tion of and reliance on this or that warrant.

The essence of the Toulmin method lies in these three ele- ments: the claim(s), the ground(s), and the warrant(s). If you have extracted these from the Dukakis-Mitchell essay, you will have identified most of what will suffice for a good grasp of the argu- ment in question.

Of lesser importance are the three other elements of the Toulmin method: the backing, the modal qualifiers, and the rebuttal.

• Fourth, consider the backing—the reasons for one’s reasons. The authors set out to argue for a claim, which they support on empirical or factual grounds, using warrants appropriate to an argument of that sort. But suppose Dukakis and Mitchell are challenged. How might they back up their rea- sons with further reasons? They are in effect answering the tacit questions “How do you know . . . ?” and “Why do you believe . . . ?” What might the authors offer in support of their views when confronted with such queries?

• Fifth, there are the modal qualifiers—or are there? Can you find any passages in which the authors qualify their assertions (“Perhaps if we tried . . .”) (“Most, although not all, illegal immigration . . .”)? Look at paragraph 11. There’s at least one modal qualifier here—can you spot it?

• Finally, there are the rebuttals, the reasons advanced by someone who rejects the authors’ claim, or who concedes their claim but rejects the grounds offered in its support, and so forth. Paragraphs 1–4 mention alternatives to the propos- als favored by the authors, and the authors reply to these objections. Is their rebuttal convincing? Why, or why not?


BAR_01611_08_ch08_pp273-288.qxd 6/17/10 12:16 PM Page 288