ISSA Proceedings 2010 – Points And Purposes Of Argumentative Texts

1. Preliminaries
What does it take to understand an argument? We can’t hope to provide the answer to this question in full here. We will instead focus on an obvious point about which there is universal agreement: understanding an argument requires that one be able to identify the argument’s conclusion. This apparent truism, however, might not be quite as simple as at first it appears. Arguments do not spring forth from the universe by themselves; they have authors. And their authors have purposes in making their arguments that are not necessarily identical to their conclusions. Indeed, it is another common idea that it in order to fully understand an argument, it is a good idea to identify the author’s purpose in making that argument. We want to suggest that these two claims are in fact closely related, and that, in fact, comprehension of the conclusion of the argument and the argument as a whole is often heightened by seeing how the author’s activity of attempting to establish that conclusion can be re-described as activity of another (but related) sort.

We approach this matter as developers of a high-level reading and reasoning test—the LSAT (Law School Admission Test). The LSAT is a high-stakes test used for admission into law school in the US and Canada (also currently by one law school in Australia). We will be focusing here on reasoning as found in longish argumentative texts, which appear in the reading comprehension section of the test (our test has four scored multiple choice sections, one of which is reading comprehension).

As is usual for tests of reading comprehension, one of the standard questions on the LSAT asks the test-taker to identify the main point, or main idea, of the passage. The motivation for this is again straightforward. Scholars and instructors of reading agree that understanding a text requires that the reader be able to identify the text’s main idea. What “main idea” designates across text types (e.g. expository, narrative, and argumentative texts) is a matter of some debate (Cunningham & Moore 1986), but if we restrict our attention to argumentative texts, as we largely do on the LSAT, then the main idea of such a text is the conclusion of the argument.[i] We call this the “main point” of the text.

Just as one standard reading comprehension question asks for the main point of the passage, another standard reading comprehension question on the LSAT asks about the author’s “primary purpose” in the text. This may appear puzzling. You might think that they are probably just two ways of asking the same thing because the primary purpose of the text is just to establish the main point (i.e. the conclusion of the argument). If so, there would be no point in making the distinction between main point and primary purpose. On the other hand, if the author of the passage attempts to argue for the conclusion in order to serve some further purpose, then the primary purpose – if it can be identified in the text – might seem itself to be the best candidate for the “main point” of the text, at least at first glance. In other words, if something in the text tells us what the author hoped to accomplish in writing his or her text, why isn’t that the main point? And so, again, the distinction would seem to be doubtful.

We are going to argue that there is a real and significant difference between main points and primary purposes of argumentative texts. We will make our case with the help of some concepts from the philosophy of action. We will conclude with some remarks about why it might be useful from the point of view of understanding argumentative texts to make the distinction between the primary purpose and the main point of the text.

Let’s begin with the above mentioned case, in which the author argues for conclusion p, but in order to, or as part of the effort to establish some further proposition, q. To be explicit (and to somewhat artificially restrict our attention to a special case), let’s imagine that p is a premise in an argument for q. Arguing in support of p will typically only be part of the process of attempting to establish q. We want to reject the notion, first, that the “main point” of such an argumentative text must be q.

Imagine a hypothetical text in which the author argues as her main point that the death penalty, as it is administered in the United States, leads occasionally to the execution of the wrong person. On the one hand, one might think that the author’s purpose is simply to show that the claim in question is true. On the other hand, depending on context and other factors that we will discuss later, it might be more accurate to say that the author’s purpose is to persuade the reader that the death penalty should be abolished. We maintain that the main point of such a text is not that the death penalty should be abolished because (so we stipulate for this hypothetical text) it does not completely make the case for this further conclusion. We claim that the same will hold for any text with a similar argumentative structure. If establishing p is only one small part of the argument on behalf of q, and it can be gleaned from the text and surrounding clues that the author aims to establish q, we maintain that establishing q is (of the cases we will discuss) most evidently and appropriately described as the author’s primary purpose in writing the text, and distinct from the effort to establish the local conclusion of argument in the text.

In other argumentative texts, however, the case for a difference between main point and primary purpose is harder to make. It is to these latter cases that we now wish to turn, and to which the bulk of our argument is dedicated. Our main aim will be to suggest that there can be a primary purpose of an argumentative text that is distinct from the main point of the text, even when the case for p thereby completely (or very nearly so) suffices for establishing q. Even in these cases, the main point need not be identified as q. Also, more generally, even when there is no further conclusion to which the case for p contributes, there can be, we suggest, an identifiable primary purpose of the text that is distinct from the main point.

So how are we to understand the relation between main points and primary purposes for these latter types of cases? They are not identical, though they are obviously closely related. How do they bear on each other? We will argue that the main point and primary purpose of an argumentative text are related to one another in these cases as two actions[ii] are related to one another when one action “level-generates” another. This relation, explained in detail by philosopher Alvin Goldman in his A Theory of Human Action, holds between co-temporaneous actions that nevertheless stand in something like a means-end relation. We turn now to a brief examination of Goldman’s framework and then show how it can be applied usefully to the distinction between main points and primary purposes.

2. Level-Generation
One action “level-generates” a second action when the two actions are performed at the same time and the agent performs the second action by performing the first.[iii] A third action can, in turn, be performed by performing the second act, and so on. As the name suggests, therefore, level-generated actions stand in a hierarchy, with more basic actions standing at lower levels. As levels are ascended, more of the purposive content of the agent’s behavior comes into view. For example, Smith signals for a cab by raising his hand in the air. Jones checkmates her opponent by moving her queen to king-knight seven. And so on. Goldman describes four kinds of level-generation: first, causal generation; second, conventional generation; third, simple generation; and finally fourth, augmentation generation. Because its utility to the textual case is limited, we will ignore the case of causal level-generation (the first in Goldman’s taxonomy) and focus instead on the remaining three varieties.

The two examples mentioned above (hand-raising/signaling and checkmating) are cases of conventional generation, which, according to Goldman, “is characterized by the existence of rules, conventions, or social practices in virtue of which an act A’ can be ascribed to an agent S, given [the agent’s] performance of another act, A.” (Goldman 1970, p. 25) There is a conventional rule, for example, that raising one’s hand in a particular way counts as hailing a cab. In addition to a rule, conventional generation often requires certain circumstances to be in place. Raising one’s hand in a classroom, for instance, counts as a very different action as raising one’s hand at the side of a road.

In simple generation (the second type in Goldman’s taxonomy), circumstances but no rules come into play in the generation. Goldman’s examples include: S out-jumps George by jumping 6 feet 3 inches and S fishes by dangling a line in the water (Goldman 1970, p. 27). Here circumstances alone dictate that performing one type of action counts as the performance of another type of action. No conventions or rules need come into play.

he final variety of level-generation Goldman discusses is augmentation generation. The key idea here is that one can perform an act and also perform that act in a specific manner. Goldman relates the act described in terms of the manner in which it is performed as the generated (higher-level) act. So S‘s extending his arm level-generates the action of extending his arm out the car window; S’s saying “hello” level-generates his saying “hello” loudly; and S’s running level-generates his running at 8 m.p.h. Goldman emphasizes that the performance of the generated act (e.g. running at 8 m.p.h.) entails the performance of the generating act (running), but not vice versa. He also notes that this form of level-generation is “not as intuitively attractive” as the other types of level-generation, in part because the actions can’t be easily described as standing in the “by” relation. It would be somewhat odd to say that S runs 8 m.p.h. by running, in the same way we say that Smith signals for a cab by raising his hand (Goldman 1970, pp. 28-29).

Part of the reason that Goldman finds augmentation generation intuitively unappealing may be because he has reversed the direction of generating and generated actions.[iv] In conventional and simple generation, it is the performance of the generating action that “entails” the performance of the generated action (together with some circumstances and/or rules). If that pattern held for augmentation generation, then it would be S’s running 8 m.p.h. that level-generates his running, not the other way around, as Goldman claims. Also, consider that in raising his hand at the curb, S signaled, but not vice versa (for he could signal a variety of ways). Applied to the augmentation case, we would say that in running 8 m.p.h., S runs, but not vice versa. So while we may not get the “by” description of the relation of the two actions, in other ways the augmented actions are, contrary to Goldman, best conceived as standing at the lower, generating level. It is this conception of augmentation generation with which we will proceed. We turn now to an application of Goldman’s concept of level-generation to the distinction between main points and primary purposes of argumentative texts.

3. Textual Generation
Taking our inspiration from Goldman, we will call level-generation that takes place within texts “textual generation”. If we take arguing for the conclusion (i.e. the main point) to be what an argumentative text, taken as a whole, “does” (Kintsch 1998, p. 66ff), then in some cases we can see that action as level-generating another action – the action in the service of the primary purpose.[v] Broadly speaking, there are two species of textual generation, one corresponding roughly to Goldman’s simple/conventional generation (we leave aside the complex matter of the role of rules) and one corresponding to augmentation generation. We will consider each in turn.

The first type, analogous to simple/conventional generation, takes the main point of the text and re-describes it in some way. It typically yields a description that contains some of the elements of the description of the main point, perhaps even constituting a paraphrase of the description of the text’s main point. In fact, however, there can be new information in such a description – a way of looking at the effort to establish the argument’s conclusion that places the conclusion in a different pattern of significance, e.g. out-jumping George includes the concept of jumping, but shows the significance of jumping 6 feet and 3 inches. And, just as these actions stand in the “by” relation, so too we can say that the author achieves his or her primary purpose by attempting to establish the conclusion of the argument. (More on this later.)

The second kind of textual generation occurs when the description of the primary purpose strips away information from the description of the main point. These are cases where the primary purpose appears to be a description cast in more general terms of the attempt to establish the main point. These are most analogous to cases of augmentation generation, where the main point is the analogue of the augmented action – the action performed in a particular manner – and the level-generated action is the more generic action, e.g. running vs. running at 8 m.p.h.

Note that deriving a generic description of the text’s main point is not necessarily to derive a description that contains less information. Consider the case of augmentation generation again. Suppose that we ask why S says “hello” loudly. Several answers are possible. It could be that S aimed to make a loud sound. Alternatively, it may be that S wished to greet someone; with this answer we learn, in effect, what is not central about S’s intentional action – that saying hello at the volume he did was in some way incidental to his main purpose – saying hello. And that, because of the way level-generated actions are structured, he could have achieved this end perhaps by saying hello at a different volume.

To return to textual generation, then, consider the case of a text that seeks to establish that global warming is real and caused by humans. This is the main point, i.e. the conclusion of the author’s argument. Here too, the purpose of doing so can be described a number of ways, but not all of them would be correct, i.e. supported by the text, signifying the author’s genuine purpose. The main point could be re-described as “discussing a phenomenon caused by humans”. Alternatively, it could be that the author aimed to “defend a position about global warming”. These are very different descriptions of the main point, and, presumably, only one of them will correctly describe the author’s purpose in writing the text. So, while it may appear that the re-description of the main point in these generic terms is a loss of information, being able to derive it correctly requires being able to rule out other possible interpretations. To see these issues in a little more detail, let’s turn to another example.

To take an example from our test: the author of one passage argues that the writing in professional history is terrible:
Part of the joy of reading is in being surprised, but academic historians leave little to the imagination. The perniciousness of the historiographic approach became fully evident to me when I started teaching. Historians require undergraduates to read scholarly monographs that sap the vitality of history; they visit on students what was visited on them in graduate school. (Law School Admission Council 2007, p. 32)

The author goes on to argue that one effort to address this problem focuses on the importance of story, of narrative, in history. This movement encourages historians to tell stories. But, the author complains that even the papers inspired by this movement are dry, dull, and dreary. At professional meetings of historians, he concludes, “we” still do not see historians who tell stories that move readers “to smiles, chills, or tears.”

We might distill the main point as follows: “The writing in professional history is abysmal, and efforts to improve it through attention to narrative are so far not promising.” Here we have the gist of what the author argues, the argumentative thrust of the text. In this case, various indications in the text – for example, his use of the first person “I”, indicating that he is a professional historian himself – suggest that the author has a direct stake in the issue. And the author’s use of certain phrases – for example, the mordant humor in the phrase, “they visit on students what was visited on them in graduate school” – suggests that he would like the situation he discusses to be improved. So we can infer that the author’s purpose in writing the piece is something like: to convince other historians that something should be done about this problem. This purpose is pretty closely related to the main point stated above, but it is logically (and perhaps rhetorically) separable, and, importantly, requires utilizing cues from the text independent from those used to identify the main point. In particular the use of “we” in the last sentence indicates that the author is addressing a community to which he belongs, with all that entails – shared interests, goals, etc. Note that this case is relevantly dissimilar to the death penalty case, in that there was more to do to convince the reader of the further conclusion in the death penalty argument, whereas here making the case for the main point pretty much suffices for the making the case for the larger point.

The argument that there is a distinction between main points and primary purposes even in cases dissimilar to the death penalty case is therefore very simple: If the effort to establish the conclusion of the argument textually generates another description of what the author intentionally does in the text, then, ipso facto, there are at least two accurate descriptions of the text’s most global features. The uppermost description deserves the title “primary purpose”, when, like the historian case, it takes some further inference by the reader to derive this description. If this purpose were to be explicitly spelled out as well as its connection to the main conclusion of the argument, then it would most likely be the best candidate for the title “main point.” But in cases where this is not so, an intelligible distinction between main point and primary purpose can be made, and worth making. We will now explore in a little more detail why such a distinction can be worth making.

4. Points and Purposes
We have already alluded to the way that textual generation gives rise to descriptions of the text that put the main point into a new light, in many cases emphasizing the significance of the main point or why it matters that the point be made. The value of this perspective should be self-evident. Even in cases of textual generation that are closer to augmentation generation can still highlight what is significant about the conclusion or put the effort to establish the conclusion in a light that reveals what is at stake. Consider the global warming case again. Identifying the primary purpose as to “defend a position about global warming”, or, even more abstractly, as to “defend a position on a scientific issue”, forces us to see the text as engaged in a debate of social-political significance (in the first case) or of scientific significance (in the second). (Note that both descriptions may not be applicable to the same text.)
As indicated above, the importance of being able to correctly identify the primary purpose of an argumentative text often has much to do with being able to rule out competing possible interpretations of the primary purpose. As in the historian case: identifying the historian’s purpose as advocating for reform requires that the reader rule out other possible uses to which the historian may have put the main point of the text, e.g. to convince writers of history that their efforts to improve are doomed to fail.

Readers of argumentative texts should be able to identify the description of the purpose textually generated from the main point that most accurately captures the author’s actual aim in making the argument. How do readers do this? In the case of physical actions, as we discussed earlier, both circumstances and rules can come into play. Being able to see Smith’s arm-raising as a cab-signaling required knowing the relevant rule (the “counts as” rule regarding cab-signaling) and appreciating the salience of the relevant circumstances (standing by the side of a road). The interpreter brings knowledge of the rule and the salience of certain circumstances to the interpretation – i.e. as background knowledge – but must observe the situation to see which circumstances actually obtain, and which rules actually apply.

Much is the same in the textual case. The reader brings background knowledge to the text that allows her to see that arguing for p counts as an instance of doing q. But background knowledge is only part of the story. The reader must be able to infer from clues in the text itself which “rules” and “circumstances” apply. Unlike the case of physical actions, in some cases the text itself provides information to the reader about how to interpret the main point that was not already part of the reader’s background knowledge. In other cases, the circumstances and rules can be gleaned from indications surrounding the text proper – as in the historian case. These activities require a kind of deep engagement with the text that goes beyond merely being able to reconstruct the conclusion of the author’s argument. Here again we take the value of this kind of engagement with the argumentative text to be self-evident. Its value resides not only in a better understanding of the argument, but, we surmise, can open the door to modes of evaluation of the argument that may not have been available without it.

Finally, we suggest that making an effort to identify the primary purpose of an argumentative text is part of a more general interpretative activity – adopting a “purposive stance” with respect to the argument – the value of which is already well-recognized. We began our paper with the truism that understanding an argument requires being able to identify the argument’s conclusion. Added to this, and perhaps equally as obvious, is the fact that one must be able to see how the elements of the argument fit together in support of the argument’s conclusion. Especially for long arguments, this means being able to decipher the structure of the argument. (Without an understanding of the structure of the argument, one well might not be able to identify the conclusion at all.) One asks, for instance, what role the second paragraph (or section) plays in the author’s argument, or what the function of a paragraph (or even a sentence) has. Another way to describe this is in terms of the purpose of various parts of the argumentative text. One aims to understand how the conclusion is supported by asking why the author does various things in the text.

The purposive stance with regard to the main point of the text can be directed “upward” in addition to “downward”. Textual generation occurs with respect to many elements of the text, and not just the most global aspects of the text. The author argues for p by establishing r, pointing out s, and rejecting the possibility of t (one in each paragraph, say). But, likewise, as we have seen, the author can be said in most cases to seek to achieve the primary purpose of the text by arguing for the main point. So, just as for smaller elements of the text, the main point and primary purpose stand in the why/how relation. The author aims to achieve the primary purpose by making the main point, and makes the main point in order to achieve the primary purpose. In many cases, an understanding of the argument’s structure is incomplete without identifying the primary purpose and how it relates to the main point of the text.[vi] Seeking the primary purpose can sensitize the interpreter to questions of finer-grained purposes that can, in turn, yield valuable insights to the argument’s structure. The chain of “why” questions should not stop once the conclusion of the argument is reached.

So, not only is the distinction between main points and primary purposes (even when they are seeming re-descriptions) real and defensible, being able to distinguish them in an argumentative text is a valuable skill. There is some evidence from reading studies that identifying the main point of a text does not occur automatically; it is an inference task (Kintsch 1998, p. 180). (Even when it is explicitly stated – that it is the main point has to be inferred). Skilled readers have been trained to make this inference. We suspect that even skilled readers often do not go the next step. And so our point is also a pragmatic one, especially as applied to argumentative texts. Once the reader has identified the main point of an argumentative text, he or she should learn to go the next step and identify what we have described as the primary purpose of the text. Granted, rhetoricians have been telling us for a long time that we should identify the author/speaker’s purpose in engaging with some discourse. But what they mean by “purpose” is either something so abstract, “e.g. to persuade, explain, etc.” as to be of little value, or something equivalent to what we have identified as the main point.

The main point of our text, then, is, once again, that there is a real distinction worth making between main points and primary purposes of argumentative texts. Our primary purpose, if it were to be spelled out, might be “to articulate and defend a reading-comprehension distinction.” This primary purpose can be textually generated from the main point on the model of augmentation generation. It appears to merely re-state the main point in a more generic form, i.e. with less information. But being able to correctly identify it plays a crucial role in correctly understanding the text. We are not, for instance, attempting to participate in some debate about the rhetorical structure of arguments – a purpose which is conceivably compatible with our main point.

NOTES
i In many cases, an articulation of the main point will include more information than the isolated conclusion. In fact, the main point of an argumentative text may best be characterized as a tightly compressed “gist” of the argument that is centered on the argument’s conclusion. Note that we are not claiming that that the main point is a summary of the argument, which often includes more information about the argument’s structure than a “gist”.
ii Or two action-descriptions. We won’t take a stand on the question of how actions are individuated, but for expository convenience will adopt Goldman’s way of describing level-generation as a relation between distinct actions.
iii Goldman further distinguishes level-generation from cases in which one performs an act while also performing another, e.g. patting one’s head and rubbing one’s stomach.
iv Another reason may have to do with his insistence on individuating actions so narrowly that each level picks out a distinct action that S performs, rather than the levels describing the same action in different ways.
v The application of level-generation to the textual case is analogical since the existence of a text creates conditions and properties that have no obvious parallel in the behavioral case.
vi Probably less for cases analogous to augmentation generation—which is not to say that even in those cases identifying the primary purpose is without value, as we have already suggested.

REFERENCES
Cunningham, J.W. & Moore, D.W. (1986). The confused world of main idea. In J.F. Baumann (Ed.), Teaching main idea comprehension (pp. 1-17). Newark, DE:
International Reading Association.
Goldman, A. (1970). A Theory of Human Action. Englewood Cliffs, NJ: Prentice-Hall.
Kintsch, W. (1998). Comprehension: A Paradigm For Cognition. Cambridge: Cambridge University Press.
Law School Admission Council (2007). The Official LSAT PrepTest 52. Newtown, PA:  Law School Admission Council, Inc.