The argumentative litotes in The Analects

1. Introduction

Scholarship on figures of speech in the past decades has contributed greatly to exploring the functions beyond the aesthetic ([13,27,29]); however, most research only investigates a few popular tropes (metaphor and metonymy in particular), leaving many important figures still unattended. Litotes, a powerful argumentative figure, is unduly and unfortunately neglected, in rhetorical scholarship as well as in argumentation studies, and, despite its correlation with morpholexical signatures of negation, it is virtually unknown in computational argument studies. This essay begins a systematic investigation of the argumentative functions of litotes, first by briefly tracing its scholarly history and then by closely examining the Chinese classic of Confucius’ Analects, a particularly rich text for the exploration of litotes, in part because of its argumentative style and in part because of the special resources of the Chinese language. This inquiry combines qualitative research methods with a quantitative approach, progressing logically with description illustrated by preliminary statistics and sample analyses. The theoretical support comes mainly from Aristotle’s Categories [2], Rhetoric [1], and Interpretation [2], and from Perelman & Olbrecht-Tyteca’s The New Rhetoric: A Treatise on Argumentation [37] with valuable reliance on Wang’s Ancient Chinese Language [42] and Gao’s On Chinese Grammar [17].

This research is through manual inspection. The complications of litotes types make automatic detection challenging at this stage; certainly, computational detection could not catch them all currently (unless tuned so broadly as also to generate many false positives), so I proceeded by manual inspection of The Analects: multiple careful readings, line-by-line, to ensure completeness and precision. The results not only justify this process, but they identify the features which can help automate such processes in the future. The inquiry, while confirming litotes familiar roles, discovers several new argumentative functions.

2. A universal figure with negligible research

Litotes – most generally known as an affirmative expressed by the negation of its opposite (as in “litotes is not an insignificant figure”) – is ubiquitous. Negation, of course, is universal. Horn in his renowned A Natural History of Negation claims, “All human systems of communication contain a representation of negation.” ([21]:xiii), and later in The Expression of Negation he adds it “is a sine qua non of every human language, but is absent from otherwise complex systems of animal communication” ([23]:1). Burke, indeed, defines man as “Inventor of the negative” ([6]:498), and Royce calls “[t]he not-relation… one of the simplest and most fundamental relations known to the human mind” ([39]:265). But it is important to note that the function of this relation is not always direct negation, and litotes is the most pervasive way that negation is deployed to a different purpose; in fact, to its inverse purpose, affirmation.

Litotes is universal in ordinary language and in texts of all domains and genres – literary and historical; philosophical and scientific; trivial and profound. Litotes by negation frequently appear in Homer’s Iliad and Odyssey, Thucydides’ History of the Peloponnesian War, Cicero’s Brutus and De Oratore, Kafkas’ Das Schloss, More’s Utopia and King Richard III, Confucius’ Analects, etc. (cf. [12,19,30,31,38]) However, in sharp contrast to its prevalence in language, its prevalence in research is meager – only a few dozen articles and two monographs.1

The most systematic of this existing research is perhaps Hoffmann’s dissertation-based monograph, Negatio Contrarii: A Study of Latin Litotes [19]. She defines litotes as an evaluative expression expressed through the negation of its opposite ([19]:42) – we will return to this definition – and plots out the syntactic features, semantic properties and pragmatic functions associated with this account. The functions she finds are mainly argumentative: periphrastic, evaluative, focusing. The first two can be regarded as ethotic, and the last as effecting salience. This is a good start, but there are other argumentative functions; for example, logical functions. This study aims to prove that litotes can help establish the enthymeme, “the substance of rhetorical persuasion” ([1]:20), with higher probability and more resistance to refutation, and that it can serve as a sophisticated major premise. Further, while Hoffmann and other researchers (such as [14,24,25,43]) have offered their respective definitions and touched upon litotic types, they have not provided adequate justifications, leading to considerable confusion. We seek, in this paper, theoretical support for understanding the key concepts of litotes, chiefly negation and opposition (shown in the next section), so to establish a solid definition and classification for the figure as a foundation for the argumentative functions and corresponding forms.

Litotes, a Greek word, like many rhetorical terms, means ‘plainness,’ ‘simplicity,’ or ‘smallness,’ with alternate spellings such as liptote, lyptote and leptotes ([9]:258–9). It first shows up in the rhetorical tradition with the first century BCE rhetorician Dionysius of Halicarnassus, who uses it in a general way to characterize the muted, understated style of Isocrates [38], and begins to associate with a specific device in the fourth century CE. The Grammarian Servius, for instance, calls it a “kind of figure” which “indicates the contrary” ([38]:353n4). But – like many other rhetorical terms, most notoriously the figures – litotes accrues various overlapping and sometimes conflicting definitions over the millennia. The common features can, however, be extracted as a first pass at a stable and fruitful definition. In the established dictionaries we find accounts such as these:

The Analects is a collection of 20 passages recording mainly the dialogues between Confucius and his disciples on all major aspects of life. Since its assembly after the Master’s death (479 BCE), the text has exerted tremendous influence both inside and outside of China. Of the various Chinese and English versions, we have chosen the authoritative Liu Baonan edition (Qing Dynasty [11]) and D.C. Lau’s highly recognized English translation [10] to extract the typical litotes data for analysis.

Taking into account the three above-mentioned subtypes, it is no exaggeration to say that this classic is brimming with litotes. By concentrating on contradictory litotes – most identifiable for its highly schematic double negation (DN), and therefore also the most computationally amenable – we have detected throughout the entire collection (11750 Chinese characters), 97 DN pairs.5

The high frequency of DN litotes (near 100 pairs) may result from various factors, among which we attribute it first to the unusual richness of Chinese negators. According to Grammarian Wang ([42]:260), there exist in ancient Chinese nine frequently used negators: “” (bù), “” (fú), “” (mú), “” (wù), “” (wèi), “” (fǒu), “” (fēi), “” (wú), “” (mò). In The Analects, all except “” (fú) and “” (fǒu) have appeared and we actually find one more adverb negator: “” (mò) collocated with “” (bù) for three times. Among all the negator collocations (about 20) in this analyzed text, the highest frequency goes to “” (bù)… “” (bù) (48 pairs), next “” (wú)… “” (bù) and “” (bù)… “” (wú) (11 for each). As stated by Lv [28] and Gao [17], “” (bù) (like not) is a dominating Chinese negator. Almost all the Chinese DN litotes are formed via two negators. As to the translated versions, the first negation is often formed by the negator of not, no, or never, in the order of frequency, the second negation either by a negator or a semantically richer negative word, such as unless or without. Comparatively speaking, DNs seem easier to formulate in Chinese (esp. ancient Chinese) because of its richer inventory of negators, and DN appears to achieve a stronger force with the prominently aligned negators than with one negator plus a negative lexeme (conjunction or preposition, etc.) or a negative-affix word.

As we will argue, however, litotes confers a distinctive style of argumentation. It condenses a subtle and resilient epistemic argument structure that both characterizes the ethos of The Analects and reflects the culture that has taken up and propagated its litotic formulations. On whether the preponderance of negators in ancient Chinese molded the argumentation of Confucius or whether Confucius harnessed that preponderance to his style, we have no position. It is a chicken or egg question. But The Analects presents us with a laboratory of litotes in which we can explore the argumentation of affirming by denying.

Typical examples follow, illustrating the two structures of litotic double negations. (Note: as you read through these examples you will see other instances of figuration – such as the epanaphora of 6 and the epistrophe of 8 – which exist in both the original text and in the English translation. We have not undertaken a study of these figures at this time. Litotes is a good day’s work on its own. But these examples should be enough to indicate how rich Confucius is, not to say how rich Chinese is, for further and broader research.)

5.2. Conditional double negations (CDNs)

Zǐ yuē: “Bú fèn bù qǐ, bù fěi bù fā.”

The Master said, “I never enlighten anyone who has been driven to distraction by trying to understand a difficulty or who has got into a frenzy trying to put his ideas into words.” – Passage VII, Section 8

Zǐ yuē: “Fēi lǐ wù shì, fēi lǐ wù tı¯ng, fēi lǐ wù yán, fēi lǐ wù dòng.”

The Master said, “Do not look it is in accordance with the rites; do not listen it is in accordance with the rites; do not speak it is in accordance with the rites; do not move it is in accordance with the rites.” – Passage XII, Section 1

Zǐ yuē: “Míng bú zhèng, zé yán bú shùn; yán bú shùn, zé shì bù

chéng; shì bù chéng, zé lǐ yuè bù xı¯ng; lǐ yuè bù xı¯ng, zé xíng fá

bù zhōng; xíng fá bù zhōng, zé mín wú suǒ cuò shǒu zú.”

The Master said, “When names are correct, what is said will not sound reasonable; when what is said does sound reasonable, affairs will not culminate in success; when affairs do culminate in success, rites and music will not flourish; when rites and music do flourish, punishments will not be exactly right; when punishments are exactly right, the common people will not know where to put hand and foot.” – Passage XIII, Section 3

Zǐ yuē: “Jǐ suǒ bú yù, wù shı¯ yú rén.”

The Master said, “Do not impose on others what you yourself do desire.” – Passage XV, Section 24

Kǒng Zǐ yuē: “Bù zhı¯ mìng, wú yǐ wéi jūn zi yě; bù zhı¯ lǐ, wú yǐ lì

yě; bù zhı¯ yán, wú yǐ zhı¯ rén yě.”

Confucius said, “A man has no way of becoming a gentleman he understands Destiny; he has no way of taking his stand he understands the rites; he has no way of judging men he understands words.” – Passage XX, Section 3, ending sentence of The Analects

There are 15 double negations in the five sentences. All of them are condition-result DNs. All the first Chinese negators marked out with waves perform as conditions, though in their English versions they may turn out in the attributive clause, as with Example 4, or in the clause of time, as with Example 6, and even in the object clause, as with Example 7, in addition to the typical negative conditional clause of as evidenced in Examples 5 and 8. Of each pair, the second Chinese negator, line-marked, indicates the corresponding result or effect. The condition-result or cause-effect DNs stand out most saliently in The Analects, not only due to the overwhelming quantity of 66 pairs, but also because of the enduring significant lessons they have successfully conveyed. However, previous research, both into litotes (which has not obviously noted this subtype7

⁷

Quite a few books on figures of speech have touched upon this type (not… + without/unless), but not yet used the name of conditional double negation, e.g. Xu ([43]:283): “We can’t learn English well without watching out for idiomatic ways of saying things.”; and “You will not succeed unless you work harder.” In Farnsworth ([14]:210), when discussing litotes’ function of definition, he actually cites a CDN created by Melville in Moby-Dick: “Had these Leviathans been but a flock of simple sheep, pursued over the pasture by three fierce wolves, they could not possibly have evinced such excessive dismay” ([32]:362) (The italics in the last example are made by the author to be consistent with the other two.) So we have a reason to treat CDN as a subtype of contradictory litotes though a CDN is not entirely the same as its corresponding affirmative.

) and into The Analects (which has not well commented on this pattern), leaves much to explain. Why should the ancient sages (represented by Confucius) so frequently state their views in condition-result double negations? What privileged function do CDNs serve and why is this function so consonant with the argumentative style of The Analects?

These questions have compelling cultural implications as well. Many of The Analects’ most efficient expressions have entered Chinese culture with proverbial status, often as chengyu [8], and this includes its CDNs. Burke long ago suggested that proverbs, adages, and other condensed opinions or attitudes provide us with “equipment for living,” highly convenient and portable ways “of arming us to confront perplexities and risks” in everyday life ([5]:61; see also 293–304). They provide a strategy for acting and they frequently imply the reasons for doing so. They are, in a very real sense, the taglines of arguments for culturally sanctioned conduct and belief (as illustrated in 5.3). We also know that proverbs, adages, and so on, are heavily figured in all languages: they frequently exhibit rhyme, assonance, isocolon, various strategies of repetition, and the like. Figuration is what makes them cognitively ‘sticky’; that is, salient upon first hearing, subsequently memorable, prone to cultural propagation. Fahnestock, picking up from a strain in Aristotle, tells us that figures

epitomize lines of argument that have great applicability and durability, and though these lines can be paraphrased in roundabout ways, they gain their greatest force in the stylistic concision of a recognizable figure. [13]:xii

Combining Burke’s and Fahnestock’s (and Aristotle’s) insights, then, we see that the CDNs of The Analects not only deploy important argumentative functions for Confucius and his milieu, but they have done so culturally in Chinese for millennia, and continue to do so.

In answering the questions of what function CDN litotes serve and why that function is consonant with the argumentative style of The Analects, then, we are also answering questions that are suggestive about Chinese discourse and Chinese styles of argumentation as well. There are potentialities here for corpus research of a new sort: Are litotes more common in Chinese than in English? Is the presence of litotes proportionally significant among Chinese adages? In what contexts are they most prevalent? And the same questions can be asked for the entire panoply of rhetorical figures, with tremendous possibilities for charting universal, intra-cultural, inter-cultural, and cross-linguistic patterns of reasoning.

5.3. Logical function of CDN litotes

Logos in Aristotle’s Rhetoric comes in two basic flavours – by way of example (induction) or by way of enthymeme, which Aristotle glosses as rhetorical syllogism (deduction) – and Aristotle demonstrates his preference for the latter: “[s]peeches that rely on examples are as persuasive as the other kind, but those which rely on enthymemes excite the louder applause” ([1]:26). The two modes are equal in persuasiveness, but deduction always feels somehow more powerful, certainly in Aristotle’s treatment. The Philosopher would have approved, then, of The Analects. All the above conditional double negation pairs function as, or contribute to, enthymemes, often characterized by two traits ([1]:28): incompleteness (omission of one or more propositions) and probability (rather than certainty) of one or more premises. The CDN litotes of The Analects fulfill both of these requirements.

5.3.1. Achieving high probability

The conditional double negation sayings of The Analects are almost all of a probable nature, but they impress us as highly acceptable or extremely convincing. However – and this is the crucial point for the argumentative deployment of litotes – their affirmative counterparts often sound vulnerable. Take Example 4, (“I never enlighten anyone who has been driven to distraction…”), which is an adage frequently cited to justify kairotic education. Its affirmative sounds far from bright (“I always enlighten anyone who has been driven to distraction…”), and perhaps arrogant; we assume that even if the pupil has been driven to distraction, the Master may still not enlighten him if he does not show due respect, or if they are not at the right place or if some other required conditions are missing. Or, take 7, (“Do not impose on others what you yourself do desire.”) is irrefutable; but its direct affirmative, “Impose on others what you yourself desire” is highly problematic. The same with Example 8, (“A man has no way of becoming a gentleman he understands Destiny”) is unquestionably much more acceptable than its affirmative, (“A man becomes a gentleman if he understands Destiny”), for “understands Destiny” is just one necessary condition to be a gentleman who actually needs many other conditions to satisfy the qualifications.

To sum up, in each condition-result double negation pair, the condition is very often a necessary but not sufficient one. A non-sufficient condition cannot lead to a solid result. However, in the double negation, the Master wisely avoids the vulnerability of the corresponding affirmative proposition, by only stressing the indispensability of the necessary condition.8

⁸

Even if the condition is sufficient, it is still often seen in DN, for example in Laotse’s adage:

It is because he does contend that he is without reproach. [26]

Here, “ (contend)” is a sufficient condition of “ (reproach)”.

Its probability, therefore, is increased high enough to resist almost all potential attacks. Or we may conclude that a proposition established by the conditional double negation proves much more acceptable with high probability in logic.

5.3.2. Making a sophisticated major premise

A typical rhetorical syllogism/enthymeme not only features probability, but truncation as well, often appearing with one or two propositions, leaving an implied reason or conclusion for the audience to infer, which not only achieves economy but also engages more interaction between the two parties [4]. This is where conditional double negation expressions perform their outstanding role. The two brief dialogues below between Confucius and his son, Boyu, illustrate well the persuasive brevity of the CDNs.

(Have you studied Shi?)

(Not yet.)

(Whoever study Shicannot speak well.)

(Have you studied Li?)

(Not yet.)

(Whoever study Licannot establish himself.)9

⁹

These translations are partially done by the author, as Lau’s version is not desirable enough. Shi and Li are two classics, the former is a collection of poems reflecting life in Zhou Dynasty (11–6BC), the latter, a compilation of rituals observed in preQin periods.

 – From Passage XVI, Session 13

These two dialogues took place in two days when Confucius inquired of his son about his study of the two classics. After receiving his frank replies, the Father each time instructed with just one non-imperative sentence; however, according to the context of that passage the son immediately studied the relevant text. That is, Boyu obeyed both instructions, though no instructions were in fact given; rather, he was given epitomized arguments framed by litotes. These two conversations present us with two forceful enthymemes with the major and minor premises, but leaving the conclusions for the interlocutor to deduce by himself:

OPEN IN VIEWER

Whoever study Shicannot speak well.	(Major premise)
[I have] Not [studied Shi] yet.	(Minor premise)
I cannot speak well unless I study Shi.	(Inferred conclusion)
Whoever study Licannot establish himself.	(Major premise)
[I have]Not [studied Li] yet.	(Minor premise)
I cannot establish myself unless I study Li.	(Inferred conclusion)

(For convenience, instead of displaying the minor premises first as in the dialogues, we list all the propositions in the sequence of a normal syllogism.) Here we clearly see the Father’s “instruction” in two major premises with high probability which impels his argument, and we see it from his son’s perspective, as the inferential structure of enthymemes should be constructed.

Almost all the conditional double negation expressions in The Analects serve as major premises to construct enthymemes either in combination with another proposition, or in certain contexts as one-proposition rhetorical syllogisms. For example, suppose one student, curious about the kitchen of the school canteen, is discussing with his classmate:

A: Let’s have a look at the kitchen.

B: Do not look unless it is in accordance with the rites/allowed or invited.

B’s terse indirect answer is an enthymeme which entails a persuasive syllogism:

OPEN IN VIEWER

Do not look unless allowed or invited.	(Major premise)
We are not allowed or invited.	(Implied minor premise)
We should not have a look [at the kitchen].	(Inferred conclusion)

The above three cases indicate that by using a CDN statement as a major premise, the speaker can adroitly achieve conversational sophistication of economically putting forth his opinion, in a convincing and even euphemistic way. Litotes provides a “highly resilient” strategy of claim-formulation ([29]:129), because it crystalizes the “two-sided vision inherent in every denial of its contrary” ([31]:121).

5.3.3. Elaborating a strict definition

In addition to achieving high probability and serving as nuanced major premises, the conditional double negations in The Analects sometimes are also used to aid in making a strict but elaborate definition, for example:

Yán Yuān wèn rén. Zǐ yuē: “kè jǐ fù lǐ wéi rén. …” Yán Yuān yuē:

“Qǐng wèn qí mù.” Zǐ yuē: “Fēi lǐ wù shì, fēi lǐ wù tı¯ng, fēi lǐ wù

yán, fēi lǐ wù dòng.” Yán Yuān yuē: “Huí suı¯ bù mǐn, qǐng shì sı¯ yǔ

yǐ.”

Yen Yuan (the Master’s favorite disciple) asked about benevolence. The Master said, “To return to the observance of the rites through overcoming the self constitutes benevolence…” Yen Yuan said, “I should like you to list the items.” The Master said, “Do not look it is in accordance with the rites; do not listen it is in accordance with the rites; do not speak it is in accordance with the rites; do not move it is in accordance with the rites.” Yen Yuan said, “Though I am not quick, I shall direct my efforts towards what you have said.” – Passage XII, Section 1

The term, benevolence (), is the kernel concept in The Analects, which might even be employed to summarize Confucianism as a whole. One brief statement is not adequate to convey its rich content, while loose interpretation may possibly harm its accuracy. A group of conditional double negations, however, can be both elaborate and, with proven high probability, not lacking serious precision. So CDN elaboration is a ready and robust tool for defining.

6. CDN litotes as epistemic

Up to now, we have seen that conditional double negation litotes in The Analects performs a very strong logical function. These highly prized CDN adages, often as major premises, have been passed down, generation after generation, solidly establishing themselves as part of common sense, shared knowledge and guiding principles – the doxa – applied to various arguments in all domains, from daily activity to scholarly research; and they have extended gradually beyond Chinese culture to other cultures.

In this light, the CDN litotes, when employed to state things generally, prove to be knowledge generative or technically speaking, epistemic. Perelman and Olbrechts-Tyteca in the manifesto-like “Conclusion” of The New Rhetoric: A Treatise on Argumentation, note that “[w]e combat uncompromising and irreducible philosophical oppositions presented by all kinds of absolutism: dualisms of reason and imagination, of knowledge and opinion” ([37]:510). This is an ethical statement, of course, enforcing the Belgians’ commitment to an argumentation that fully accommodates, respects, and advances values. But it is an epistemic statement as well, a key part of their credo for an ethics of reason and the warrants of knowledge. Their argumentation justifies that opinions to which the strong “adherence of minds” ([37]:45) is forged should be no other than securely founded knowledge. CDN expressions with high probability as we see above (Examples 4, 7, 8, etc.) are undoubtedly well justified opinions which can be catalogued as one type of knowledge, and litotes, we therefore also see, is a figure that effectively epitomizes and successfully propagates such knowledge.

Masse, in his research on the epistemic dimensions of litotes in early modern scientific discourse, summarizes the conflicting features of oppositional knowledge: “In the one case, oppositional knowledge leads to algorithmic succinctness; on the other, it provides stumbling and building blocks, and thus loquacity” ([29]:121). The first case is epitomized by CDN litotes.

Not all litotes are epistemic, however, from what we have seen; judging the three subtypes identified above – contradictory, contrary, relative – we note that the last two usually offer situational responses which do not generalize or are not serious enough to propagate as knowledge. At least, their epistemic power requires more evidence than we could find, or than The Analects appears to offer. Rather, it is the contradictory type, conditional double negation in particular, that generates views rational enough to be treated as knowledge. As to the features of the litotes-generating knowledge, besides the algorithmic succinctness Masse identifies, we conclude that what CDN knowledge features most, is its high probability which can well justify its objectivity and acceptability.

7. Conclusion

The rich variety of negators in ancient Chinese may partially account for the pervasiveness of DN litotes in The Analects; however, we argue that what is important about them lies in the various argumentative functions they successfully perform. With regard to the function of litotes, previous studies have mainly revealed its highlighting ([38]), evaluative ([19]), understating ([14]) roles, which are primarily vehicles of ethos (especially the last two in combination). While some litotes in The Analects certainly exhibit the highlighting function and the ethotic features of being euphemistic, modest, or cautious, we show that the most prominent function of the most prominent litotes, conditional double negation, is to achieve the strong logical function, the irresistible high probability in particular, and hence correspondingly realize the epistemic power of knowledge producing, accumulating and disseminating.

This epistemic logical function of litotes contributes to the project of shedding a new light on figure studies, of freeing them from the aesthetic cave, to reveal the cognitive dimensions and the argumentative contributions they make to discourse. More particularly, our exploration of this litotic function, like the other contributions to this special issue, contributes to the enterprise of computational rhetoric, the enterprise of plumbing such functions with the powerful machines and perceptive algorithms of our age. Because litotes is a trope, it implicates directly the semantics of discourse, making it especially valuable for such meaning-probing tasks as text summarization and argument mining. But because of its schemic elements – its morpholexical-syntactic signatures of negation – litotes is also more tractable computationally than most semantic figures.

Among the diverse goals we seek in this essay, we would be most pleased to find it has sparked more and deeper studies into litotes, to begin redressing the great disparity between the scanty research and universality of litotes. Viva litotes!