ISSA Proceedings 2002 – From Argument Analysis To Cultural Keywords (And Back Again)

1. Introduction
The present investigation aims at bridging recent research on cultural keywords (i.e. words that are particularly revealing of the values of a culture) carried out in various areas of linguistics with the logical and rhetorical analysis of arguments. It will be shown that between these two scientific endeavours there can be a fruitful two-way influence. On the one hand, considerations from argumentation theory can help significantly in the complex task of  hypothesising and testing candidates to the status of keywords in a given culture. On the other hand, our understanding of the functioning in argumentative discourse of endoxa and topoi (as culturally shared values and beliefs and culturally shared rules of inference respectively) can greatly benefit from explicit semantic analyses of cultural keywords. In the article a strategy for this interaction is outlined, motivated and briefly exemplified.

2. Keywords and cultural keywords
What is a keyword? A keyword in the sense the term acquired in the fields of Library Science and Internet search engines, is, as the key metaphor suggests, a means of access to digitally stored information. Apparently, keywords can be used so because they are in some sense representative of a whole body of knowledge to which they are associated. Likewise, the notion of cultural keywords, which introduces a further layer of metaphor, suggests the, admittedly vague, idea of words that are particularly revealing of a culture, that can give access to the inner workings of a culture as a whole, to its fundamental beliefs, values, institutions and customs. In short, of words that explain a culture.
The notion of cultural keyword is often associated to the name of Raymond Williams and to his influential pocket dictionary Keywords: A Vocabulary of Culture and Society (Williams 1976). His study, based for linguistic data on the Oxford English Dictionary, methodologically belongs to a broadly humanist tradition of scholarship, falling somewhere between the history of ideas and what is sometimes called the “external history of language”. In the choice of entries it largely reflects the author’s concerns for social organisation and sometimes his interest for Marxist social theorising: alienation, bourgeois, capitalism, dialectic, hegemony, revolution.
While his contribution to cultural analysis is broadly relevant for the understanding of the cultural and ideological backdrop of a number of contemporary argumentative practices, in what follows we will adopt a much narrower focus, restricting ourselves to the contribution of linguistics proper and, more specifically, to approaches that emphasise the use linguistic semantic methods and theoretical tools, in order to examine how these tools can be brought to bear on the tasks of reconstruction and evaluation of natural language arguments.

Linguistic semantic research on cultural keywords can be seen as but one aspect of the use of linguistic methods to investigate culture. In the USA a rich tradition of anthropological linguistics was developed  by scholars such as Franz Boas, Edward Sapir and Benjamin Lee Whorf. In this tradition language, both grammar and lexis, is seen as a “symbolic guide to culture” (Sapir 1949: 162)(i). Moving from this classic anthropological linguistics tradition, the work of Anna Wierzbicka on cultural keywords concentrates on the semantic analysis of areas of the lexicon where highly language specific distinctions reflect specific ways of living as well as “ways of thinking” that historically shaped a community and, at the same time, help to perpetuate these ways. According to Wierzbicka (1997: 22), linguistic semantics provides a rigorous methodology for decoding culture specific meanings and, consequently, for elucidating the tacit assumptions which are linked with them. The domains covered by Wierzbicka’s analyses range from social and political values, to ethics, folk-psychology and ethnic identity, which she examines with respect to a number of European and extra-European languages.
What motivates the choice of a lexical item as a keyword? In fact, one difficulty with keyword research is that, as Wierzbicka (1997) puts it, “there is no objective discovery procedure for identifying key words in a culture”. A series of  clues may direct the investigator towards a particular word: sheer frequency of occurrence, frequency of occurrence in a particular domain, frequency of occurrence in book titles, songs, proverbs, sayings, richness of the phraseological patterns in which the word occurs. However, in order to justify the claim that a lexical item is indeed a keyword in a given culture, a researcher has to show that a thorough semantic analysis of this item leads to interesting insights on that culture. One such result may be the discovery that the word is a focal point around which an entire cultural domain is organised, and that the concept it denotes is the basis of a whole array of  more or less tacit cultural rules of interaction, or “cultural scripts” (cf. for instance the Japanese omoyari discussed in Wierzbicka 1997 and in Goddard &Wierzbicka 1997).
The present paper argues that looking at the role played by words in argumentative texts researchers in cultural keywords can find, if not an “objective discovery procedure”, certainly a significant testbed.

2. Corpora, “discourse” and argumentation
M. Stubbs in a series of recent publications (cf. Stubbs 1996  and 200)offered a quite different outlook on the issue of keywords, which on the one hand emphasises the usefulness of computer aided corpus analysis for a more systematic investigation of keywords, and on the other directly addresses the issue of the persuasive power of keywords. According to Stubbs, the analysis of cultural keywords should proceed mainly through the exam of the recurrent linguistic contexts in which these words occur (collocations). Typical collocations of keywords will provide evidence of their “cultural connotations”. The idea of using concordances to investigate culturally significant words, however, predates the advent of computers in linguistic research as it had been  already outlined by Firth (1935), who proposed a “systematic study” of the “contextual distribution of sociologically important words”, which Firth called focal or pivotal words. It is important to mention that for Firth this type of analysis wasn’t just a complement to the analysis of lexical units but was considered to be the core methodology of lexical semantics.
In carrying out with modern computer techniques the Firthian project, Stubbs places it within the study of discourse in the sense that Foucault and various postmodern social theorists give to this word: “In phrases such as ‘academic discourse’, and ‘racist discourse’, ‘discourse’ means recurrent formulations which circulate in a discourse community” (Stubbs 2001: 166). These recurrent patterns embody “shared meanings”, “particular social values and views of the world” (Stubbs 1996: 158). As Stubbs puts it “Such recurrent ways of talking do not determine thought, but they provide familiar and conventional representations of people and events, by filtering and crystallizing ideas, and by providing pre-fabricated means by which ideas can be easily conveyed and grasped” (ibid.).
It is particularly interesting, from our point of view, to look at what is according to Stubbs, the role played by cultural keywords and by recurrent patterns of discourse within argumentation.

In examining a series of speeches of British conservative politicians, Stubbs repeatedly observes how arguments, which are characterised by an “absurd logic” and if regarded rationally “are a sequence of non sequiturs” derive their force from being part of “a discourse which call up a set of linked key words, symbols and beliefs” and from the fact that they depend on a set of premises, which are unstated and probably unconscious” (Stubbs 1996: 162). Throughout his analysis Stubbs oppose a logical/ rational mode of argumentation to a mode of argumentation that does not “operate at this level” but is instead based on ‘keywords’.
If we consider what could be the contribution of this type of approach keywords to argumentation theory, and more specifically to the critical evaluation of arguments, we find that there are a few aspects of this notion that need to be clarified.
It is true that a large portion of the structure of natural language arguments remains implicit, as they are crucially dependent on unstated premises, which correspond very often to beliefs and values shared within a community. This type of picture emerges from the Aristotelian notions of enthymeme and endoxon, according to classic interpretations such as Bitzer (1959). If we take keywords simply as words that function as pointers to culturally shared beliefs and values (endoxa) or to culturally shared patterns of inference, their use doesn’t seem to entail a mode of argumentative functioning distinct and opposed to the logical one. The situation is completely different if we take argumentation by keywords as based on purely syntagmatic associations of words that derive their apparent naturalness in the mind of hearers only from repeated co-occurrence within a certain discourse. These two very different levels are not clearly distinguished in Stubbs’ analyses.
In fact, the author’s goal is not the description and critical evaluation of arguments – that is individual texts intentionally produced by authors in order to achieve certain (persuasive) goals – but rather the study of a process that takes place far beyond the sphere of conscious intention, in the realm of what Foucault calls “pure discourse without the knowing subject”. This process is the “reproduction of ideology”. According to this view, computer aided corpus analysis of large corpora is the ideal tool to uncover the processes that take place at the level of discourse because it treats texts “without regard to authors and their intentions” (Stubbs 1996: 194).
Given that this process takes place beyond the consciousness and, in a sense, regardless of the intentions of individual authors it becomes legitimate to argue, as Stubbs does, that the recurrence of collocations such as Jewish intellectuals and Marxist intellectuals is not innocent, even if  “there is nothing explicitly negative in such collocations and their negative force can easily be denied”, the word form intellectuals acquires nevertheless negative connotations because of its general frequency of co-occurence with words like contempt, hippie, ideology, activists and even with words such as students, young and dissident which “would be interpreted negatively in many circles” (Stubbs 1996: 188).
The problem we have with this type of analysis is that focusing one’s attention exclusively on quantitative patterns of lexical co-occurrence, regardless of the intention and structure of texts, results in a dangerously simplified image of culture and cultural reproduction. Inevitably, cultural reproduction has to pass through individual texts, which are characterised by specific intentions of the speaker. Four decades of research in pragmatics have shown that intentional behaviour plays an enormous role in the determination of the value of  particular linguistic occurrences within a text, at least as big as the role played by the past discursive practices of the speech community.
In our view, a natural language text, slippery and vague as it may be, is not a stone soup where words float free, tied only to their multiple associations within a Foucaultian discourse.
Simply put, a text is a coherent sequence of utterances, where coherence is not ensured by repetition of patterns, but by the congruity of the meaning of each utterance with the intended effect of the whole. The research strategy we propose here takes in to account how word meanings interact with the semantic-pragmatic structure of persuasive texts.

4. Keywords and topoi in the enthymematic structure of natural language arguments
Let us consider the following rather trivial example:
He’s a traitor. Therefore he deserves to be put to death.

One accessible interpretation of the above sequence is that the two asserted propositions are to be understood as manifesting a textual act of argumentation (Cf. van Eemeren & Grootendorst 1991), where q is the conclusion and p is a premise. In the approach to discourse semantics developed by Rigotti (1993) and Rigotti & Rocci (2001) such an interpretative hypothesis on the connection that the speaker establishes between the two asserted propositions is treated as an abstract relational predicate (Connective Predicate), which defines the roles of the two propositions relying them to the global intended effect of the text. In this specific case the construction of the interpretative hypothesis is facilitated by the use of  the explicit discourse connective therefore(ii).

According to the traditional, logical, view of the reconstruction of  enthymematic arguments, in order to have q follow from p, as in our interpretation, we have to supplement an adequate unstated premise, which is presupposed to be shared by the speaker and the hearer: one such premise is, for instance, Traitors deserve to be put to death. The argumentation is thus reduced to the following syllogistic form:
Major premise: Traitors deserve to be put to death (unstated)
Minor premise:  He is a traitor
Conclusion: He deserves to be put to death

We can notice that the word traitor plays an important role in the both in the logical and in the communicative structure of the argument. From a logical viewpoint, it appears in the subject of the major premise and in the predicate of the minor premise, playing the role of terminus medius in the structure of the syllogism. From a communicative viewpoint it plays an important role in the recovery of the unstated major premise.
It seems that the word traitor is associated with a number of culturally shared beliefs and values that confirm the plausibility of an unstated premise such as Traitors deserve to be put to death. Even if we do not share such values they remain easily accessible to us: the belief that some people may subscribe to such a set of values, or may have subscribed to them in the past, is part of our cultural endowment. This type of culturally shared values and beliefs can be identified with the Aristotelian notion of endoxon.
The discussion of the above example, simplistic as it may be, already allows us to present a working hypothesis for the discovery and testing ocultural keywords. We propose to consider as serious candidates to the status of cultural keywords the words that play the role of  terminus medius in an enthymematic argument, functioning at the same time as pointers to an endoxon or constellation of endoxa that are used directly or indirectly to supply an unstated major premise. More precisely, words that typically have this kind of function in public argumentation within a community are likely candidates to the status of keywords in the culture of that community.
Let us consider another, more concrete example, which also shows how such argumentative keywords can be used to label particular communities, or at least particular socially shared opinions. In the public debate on abortion the two opposing positions are often characterised as Pro-Life and Pro-Choice.
These labels are interesting because they are not directly descriptive of the standpoints argued for by the two opposing parties, rather they point of the values that are called forth in order to argue for the respective standpoints. One might present the two opposing lines of argument in quasi-syllogistic form as follows, as does Weigand (1997):
Major premise: The sanctity of life is an absolute value.
Minor premise: Abortion violates the sanctity of life.
Conclusion: Abortion violates an absolute value.

Major premise: Freedom of choice is an absolute value.
Minor premise: Laws prohibiting abortion violate freedom of choice.
Conclusion: Laws prohibiting abortion violate an absolute value.

One can observe that the termini medi of the two arguments are sanctity of life and freedom of choice respectively. One can conceive of the two labels Pro-Life and Pro-Choice as two condensed arguments, whose only explicit part consists in mentioning the keyword/ terminus medius(iii). In fact, when the standpoint being argued for or against is known (or presumed to be known) from the outset, as it is often the case in ongoing public debates, the mention of the keyword is sufficient to summon the major premise, which is then applied to the case at issue.
According to this view, a keyword is a predicate that plays a decisive role in the enthymematic structure of the argument, but not simply as a predicate but because it is bound to an endoxon, which is a proposition(iv). One recurrent characteristic of the endoxa connected to keywords is that they define a positive or negative orientation towards action: Life is to be preserved at any cost, Freedom of choice is most desirable, Treason is the worst crime and the like. The form of such endoxa may be as simple as the following:
For all x : if x has the property P then x is good/bad, desirable/undesirable.

The nature of these hypothesized endoxa, however may cast some doubts on the enthymematic reconstruction of the functioning of keywords in argumentation. Observing the use of words such as life, choice, freedom, democracy in persuasive discourse, one often has the impression that these words possess a persuasive power on their own. In the sense that what they denote simply appears to be considered good or bad, without any need of further motivating this value judgement.
Sometimes this special persuasive potency is considered to be part of the meaning of these words. Connotation as opposed to denotation is a term often employed to refer to this very special type of meaning. For Stubbs the negative or positive connotations attached to cultural keywords are both ‘reflected in’ and ‘generated by’ their “semantic prosodies” that is their repeated patterns of co-occurrence within a discourse, and are thought to be both conventionalised and at least independent from the denotative meanings.
The integration of the “intrinsic” argumentative power of words into linguistic semantics is one of the chief objectives of  the theory of “argumentation within language” (argumentation dans la langue) proposed by Anscombre & Ducrot (1983, 1989). The theoretical devices they employ might look similar to the idea of endoxa associated to keywords we have sketched above. They describe the semantics of a lexical predicate in terms of its virtual argumentative possibilities (the range of “virtual” conclusions it does or does not licence) rather than in terms of its truth-conditions. Let us consider, for example, the following argumentation:
It’s warm. Let’s go to the beach.

According to Anscombre & Ducrot (1989), the possibility of arguing from the warm weather to the decision for going to the beach is to be accounted directly by the linguistic semantics of the predicate warm. The semantics of a predicate is given solely in terms of a bundle of topoi (culturally shared argumentative rules) associated with it. The predicate work, for example, could be described in terms of the following:
The more work, the more success
The less work, the more relaxation
The more work, the more fatigue
The less work, the more happiness

This purely argumentative approach to semantics raises a number of  grave problems, both linguistic and epistemological, and has been the object of detailed criticism (cf. Iten 2000, for a fair but nevertheless lethal critique). Here we will limit ourselves to a few remarks more directly connected to developing an adequate approach to the functioning of keywords within argumentation. Perhaps the single greatest problem with this approach is that it really confines meaning and argumentation within the closed system of the langue: the meaning of a unit (e.g. work) is defined solely in terms of the topoi in which it appears and the meaning of  the other lexical units that appear in the topoi (e.g. happiness, fatigue, relaxation, success) is in its turn defined exclusively in terms of  other sets/ bundles of topoi (topical fields). This is, as Anscombre & Ducrot (1983) rightly claim, an approach in the purest structuralist vein: argumentation is reduced to nothing more than a linguistic connection between units in a system. It hasn’t anything to do with inference, let alone truth.

The fact that the topoi are the primitives of the theory also means that, in the end, there is no way to figure out why a unit licences a certain set of conclusions. The topical fields of Anscombre and Ducrot, considered as hypotheses on word meaning, are scarcely interesting, because they have no predictive power: nothing will tell us if a certain argument will be acceptable or not on the basis of topical fields, because we cannot practically make explicit the full sets, and the theory does not provide a principled way to generalise from a set of concrete topoi to a more abstract class of possible conclusions. In fact, it seems easier to think that if we are able to conclude (in certain contexts) from work to fatigue or to success it not because of the linguistic knowledge of the topical field, but because we have a rich background of world knowledge which includes beliefs about how our body works, how society works, etc.
Leaving aside “argumentation within language”, another less radical approach, which does not put into question denotative semantics, would be to include endoxa into the meaning of a keyword by treating them as a sort of analytical statements (compare For all x: if x is a triangle then x has three sides). This is what Wierzbicka (1997) does in practice, even if she does not address explicitly the issue, when she includes in her semantic definitions of cultural keywords clauses such as
– everyone thinks: this is good (from liberty: p. 136)
– people think: it is bad if someone does this (from whinge: p.216)

One general reason for keeping apart endoxa from the denotative meaning of words is that they are much less stable, more variable across particular communities and much easier to re-negotiate. An interesting illustration is offered by paradoxical statements and arguments.
A look at the behaviour of paradoxical arguments will help us to sketch an account of the persuasive power of keywords more true to the way they are established within communities through textual interaction. Let us consider the following two blatantly paradoxical arguments by Charles Baudelaire:
Le commerce est naturel, donc il est infâme (Mon coeur mis à nu, par. XLI, Baudelaire 1975-76: 703)
(Trade is natural, therefore it is vile)

La femme est naturelle, c’est-à-dire abominable (Mon coeur mis à nu, par. III, Baudelaire 1975-76: 677)
(The woman is natural, that is to say abominable)

While these argument strike us as paradoxical, they remain nevertheless perfectly comprehensible. The explicit discourse connective enable us to understand them as arguments, even out of context, that is to say to establish an argumentative Connective Predicate. In order to preserve the congruity of the connected propositions with the function assigned by the Connective Predicate we infer a suitable major premise, such as All that is natural is vile/ abominable, which is not a shared belief in our contemporary western culture, nor it was in Baudelaire’s times. The inference of the major premise, can be seen, in fact, as the accommodation of a presupposition imposed by the argumentative Connective Predicate. In interpreting Baudelaire’s arguments, however, we do not have to revise our notion of nature: we only have to hypothesise that Baudelaire subscribes to a very unusual set of values or entertains very peculiar beliefs about nature.

In fact, the above examples sound paradoxical only if compared to broadly shared present day assumptions about nature. If we consider how the theme of the artificial and the contempt of nature are developed in the Fleurs du Mal and the rest of Baudelaire’s works (cf. Cigada 1992), the statement All that is natural is vile/ abominable is no longer paradoxical, it becomes something which is shared within the community of the author and his understanding readers, once the poet has shaken from them the hypocritical veil of socially accepted mores and common sense. We have to recognise that texts can quite easily modify and even completely restructure the functioning of established keywords, or create their set of text-specific keywords. This may lead to the establishing of new cultural keywords: as the Fleurs du Mal  came to be considered a sort of “canonical”, foundational text of the Symbolist movement in France and abroad, its keywords (such as nature, artificial, ennui, spleen and sign/ symbol) became cultural keywords of the symbolist poetics and its values culturally shared values (at least as poetic values) within that symbolist culture.

If we examine closely the functioning of communication through texts we find that the establishing of new cultural keywords becomes much less a mysterious and impersonal process and that it is, at least in part, based on explicit argumentation.
For example, it is easy to agree that, after the tragic attacks of September 11 2001, terror terrorist(s), terrorism, etc. have become political keywords, in America as well in the rest of the Western World and in the Middle East, and have assumed an important argumentative role, in motivating political and military decisions as well as in justifying them in front of public opinion.
What is probably less obvious is how certain policy defining texts not only exploit terror as a keyword, but explicitly work to establish it and further motivate it through argumentation. This can be seen very clearly, for instance, in the speech on the Middle East crisis, U.S. President G.W. Bush delivered in the Rose Garden the 4th of April 2002. The role of the keyword terror is rather explicitly defined and motivated in on the first passages of the speech with the following argumentation:
Terror must be stopped. No nation can negotiate with terrorists. For there is no way to make peace with those whose only goal is death.

In this passage, terrorist are implicitly defined as people whose only goal is the death of their enemy: it follows that no negotiation, no compromise is possible with them: negotiation implies the exchange of goods, and we do not dispose of any good they desire that we can exchange. They have no price that we might be willing to pay, as no one can compromise on his own life. If we cannot negotiate we have (to use force) to stop them (Terror must be stopped).
In the rest of the speech the words terror and terrorists play the role of argumentative keywords, acting in many occasions as pointers to the endoxon Terror must be stopped established by the argumentation above. The structure of the argumentation can be partially implicit, as in the following example where the mention of terrorist networks that are killing its citizens acts a justification of Israel’s military operations:
Given his [Arafat’s] failure, the Israeli government feels it must strike at terrorist networks that are killing its citizens.

4. Reshaping meaning: reason as a keyword in Milton’s Areopagitica
In the preceding section we have briefly pointed to the way in which the endoxa attached to cultural keywords can be redefined through texts while their denotative meaning remains constant. There are however cases in which it is the denotative meaning of a word which is being redefined, or better reshaped by the textual context in which the word appears.
This operation of coercing a word into a new meaning has an important and delicate argumentative facet when the word in question is an important and prestigious cultural keyword, loaded with endoxa. The word, in its new meaning, may retain, wholly or in part, the persuasive power associated to the old meaning within the relevant culture(v). The example we want to examine here is a passage from Milton’s Areopagitica. A Speech for the Liberty of Unlicensed Printing to the Parliament of England (1644), where the word reason plays a fundamental argumentative role, and, at the same time, is subject to a subtle semantic shift:
“I deny not, but that it is of greatest concernment in the Church and Commonwealth, to have a vigilant eye how books demean themselves as well as men; and thereafter to confine, imprison, and do sharpest justice on them as malefactors. For books are not absolutely dead things, but do contain a potency of life in them to be as active as that soul was whose progeny they are; nay, they do preserve as in a vial the purest efficacy and extraction of that living intellect that bred them. I know they are as lively, and as vigorously productive, as those fabulous dragon’s teeth; and being sown up and down, may chance to spring up armed men.
And yet, on the other hand, unless wariness be used, as good almost kill a man as kill a good book. Who kills a man kills a reasonable creature, God’s image; but he who destroys a good book, kills reason itself, kills the image of God, as it were in the eye. Many a man lives a burden to the earth; but a good book is the precious life-blood of a master spirit, embalmed and treasured up on purpose to a life beyond life.”

The first paragraph of the passage, introduced by the quasi-performative clause I deny not is presented as concessive, while it plays a clear tactical role introducing the theme of the comparison between men and books, and the theme of the vitality, of the potency of life that characterises books. Both the general idea of the resemblance between men and book and the recognition of the potency in books are used as premises in the subsequent argumentation, and they both, in a sense, come for free as they appear to be conceded to the interlocutor.
The second paragraph presents the main standpoint as good almost kill a man as kill a good book, which will be later implicitly reinforced to the effect of claiming that destroying a book is at least as bad as to kill a man (and even worse). Here the words reason and reasonable and specific endoxa attached to them play a crucial argumentative role, and are connected through a topos to the main standpoint.
The relatively explicit mention of the endoxon according to which reason is what makes man an image of God (a reasonable creature, God’s image) enables Milton to connect reason to the ultimate culturally shared source of value: God. Successively, he argues that reason appears in books in its purest form, while men appear to be blessed with such a gift in widely varying degrees, so that many a man lives a burden to the earth. These arguments support the conclusion that destroying a book worse than killing a man through the application of a seemingly trivial topos, which has the following general form:
If x is a valuable substance the more of x the better, the less of x the worse.

Since books are likely to contain more reason than mortal men, the loss of a book is graver than the loss of a man.
The semantic congruity and logical consistency of the argument depend crucially on the attribution of a particular semantics of the word reason. We have to construe reason as a concrete noun denoting an uncountable substance, or, at least some sort of entity, and not as an abstract noun denoting a property of  the human being, a faculty (facultas) of the subject. As construed by Milton’s argument reason is something that can exist by  itself irrespectively of its support, it is like a liquid that can be put in different vessels. Human beings have value (that is are the image of God) inasmuch as they offer a suitable support to reason. In a sense, books offer a more suitable support for reason than human beings: for books store up reason as it is found in a master spirit, and preserve it to a life beyond life. Therefore books are more valuable than human lives themselves.

One can object to this analysis that envisaging reason as a substance is a metaphor, and that we are guilty of taking Milton’s poetic metaphors too literally. In fact, Milton in this and in nearby passages is attributing a number of predicates to reason all of them metaphorically.
We have to ask ourselves what is the point of these metaphorical predications. These predications have a number of entailments and Milton cannot be taken as communicating all of them. However the point in using metaphors is communicating some of these entailments, which are relevant. Suppose that a boy, Tim, is being told by his mother:
Your room is turning into a pigsty. You must clean it up immediately.
In order to interpret his mother’s metaphorical statement, Tim derives the following relevant implication:
A pigsty is a messy and dirty place.
Your room is turning into a pigsty. 
Your room is turning into a messy and dirty place.

This implication is relevant in the sense that it makes the first utterance an argument for the conclusion in the second utterance. While Tim’s mother cannot be held accountable of communicating the irrelevant (and implausible) implication Your room is going to be inhabited by pigs, she probably cannot deny having communicated Your room is turning into a messy and dirty place on the grounds that she was just using a metaphor.
Likewise, we cannot hold Milton accountable for saying that Reason is a liquid and can be put into vials, but we can and must maintain that he sees reason not as a human faculty but as something that has an existence independent from its accidental supports (be they men, books, chimps, robots or software agents). This is because that construal of reason is necessary to account for the congruity of Milton’s argumentation, just as hypothesising that Charles Baudelaire holds nature in contempt is necessary in order to account for the congruity of his paradoxical arguments.
Our hypotheses on Milton’s communicative intentions guide us in our search for the relevant entailments of the metaphors he uses. In a more technical linguistic parlance: we have to shift the semantic type of the noun reason in order to accommodate the presuppositions imposed by the specific argumentative Connective Predicate relying the sequences above into a unitary communicative intention.

The strategy of semantic analysis we have very informally outlined above does not provide an infallible method to evaluate the consistency of natural language arguments. All that we can do is to evaluate the congruency of two orders of hypotheses: hypotheses on the semantics of the word reason and hypotheses on the underlying logical-semantic structure of the text. But this limit, rather than being a shortcoming of the theory, is an inevitable condition, which is the consequence of two characteristics of natural language discourse. The first is that the semantics of lexical items can be, up to a certain extent, re-negotiated in the text, in order to accommodate the presuppositions imposed by Connective Predicates on discourse units. The second is that the semantic structure of a text is to a large extent implicit and has to be built inferentially as an ongoing hypothesis from various types of clues, which, of course, include the semantics of lexical items.

5. Conclusions
In this paper we have shown how semantic research on cultural keywords can be combined fruitfully with a classic enthymematic approach to argument analysis in order to provide a rationale for testing cultural keywords. Moreover, we have seen how this same approach can provide useful insights for a realistic treatment of the “persuasive power” of keywords.
It is worth noting that the interaction between lexical semantics and argumentation theory we sketched rests on a particular semantic-pragmatic theory of text as communicative action where the notion of semantic congruity plays an important theoretical role (cf. Rigotti & Rocci 2001). In this contribution we made but a very informal use of this notion. The future developments of this work will be devoted to spelling out its implications for argument analysis in a more explicit way.

NOTES
i. “Distinctions which seem inevitable to us may be utterly ignored in languages that reflect an entirely different type of culture, while these in turn insist on distinctions which are all but unintelligible to us” (Sapir 1949: 27).
ii. We can characterise the function of a lexical items such as therefore (q) as imposing a series of constraints on the proposition (p) and to its relationship on a number contextual propositions, one of which at least has usually to be available in a preceding utterance in the text, and finally on the intentions of the speaker in uttering q. A very informal explication of the meaning of therefore could be something along the following lines: [[Therefore (p)]] =  [[ The speaker presents p as inferable from evidence presented in the preceding text in conjunction with relevant knowledge shared by the speaker and the hearer]]. Given a favourable context the relation signalled by therefore can be further enriched by the hearer to derive a fuller interpretation (or Connective Predicate) according to which the speaker presents q as inferable from p, in order to argue for q.
iii. It is perhaps not entirely trivial to mention that the communicative effectiveness of the two labels depends on the fact that the keywords sanctity of  life (or at least deep respect for life) and freedom of choice are both values whose importance is widely agreed upon in contemporary Western societies, even by those who might not agree with the conclusions of one or the other argument.
iv. Note that what is believable or desirable is necessarily a state of affairs: the denotation of a predicate, that is a concept, cannot be believed or desired as such.
v. This phenomenon can be seen as a particularly subtle, not completely conscious instance of the fallacy of equivocation.  Diachronically such a shift may herald deep semantic and cultural changes. Synchronically, while it always appears as a pragmatic process of meaning coercion imposed by the semantic context to the word, this type of shift is not an ordinary phenomenon of contextual specification or metaphorical extension of meaning, for there is no clear awareness that a different concept is being expressed.

REFERENCES
Anscombre, Jean-Claude & Ducrot, Oswald (1983). L’argumentation dans la langue. Brussels: Mardaga
Anscombre, Jean-Claude & Ducrot, Oswald (1989). Argumentativity and Informativity. In: M. Meyer (Ed.) From Metaphysics to Rhetoric (pp. 71-87). Dordrecht: Kluwer Academic Publishers
Bitzer, Lloyd F. (1959). Aristotle’s Enthymeme Revisited. The Quarterly Journal of Speech. XLV. 4, 399-414
Cigada, Sergio (1992). Charles Baudelaire: antropologia e poetica. In: S. Cigada (Ed.) Il Simbolismo Francese (pp. 31-74). SugarCo Edizioni: Varese
Conley, T. M. (1984). The Enthymeme in Perspective. The Quarterly Journal of Speech. LXX.2, 168-187
van Eemeren, Frans H. & Grootendorst, Rob (1991). The study of Argumentation from a Speech Act Perspective. In: Jef Verschueren (Ed.) Pragmatics at Issue. Selected Papers of the International Pragmatics Conference, Antwerp, August 17-22, 1987, vol. I (pp. 151-170). Amsterdam/ Philadelphia: Benjamins
Firth, J.R. (1935). The Technique of Semantics. Transactions of the Philological Society, 36-72
Freese, J.H. (1947). Aristotle: The Art of Rhetoric. Cambridge-Mass./London.
Foucault (1972). The Archeology of Knowledge. Transl. A.M. Sheridan Smith. New York: Pantheon
Goddard, Cliff & Wierzbicka, Anna (1997). Discourse and Culture. In: T.A. van Dijk (Ed.) Discourse as Social Interaction. Discourse Studies: A Multidisciplinary Introduction. (pp. 231-259). London: Sage
Iten, Corinne (2000). The relevance of Argumentation Theory. Lingua, 110, 665-699
Rigotti, Eddo (1993). La sequenza testuale. Definizione e procedimenti di analisi con esemplificazione in lingue diverse. L’analisi linguistica e letteraria 1, II, 43-148.
Rigotti, Eddo & Rocci, Andrea (2001), Sens – non-sens – contresens. International Studies in Communication.1, 45-80
Sapir, Edward (1949). Selected Writings of Edward Sapir in Language, Culture and Personality. Ed. David Mandelbaum. Berkeley: University of California Press
Stubbs, Michael (1996). Text and Corpus Analysis. Computer-assisted Studies of Language and Culture. London: Blackwell
Stubbs, Michael (2001). Words and Phrases. Corpus Studies of Lexical Semantics. London: Blackwell
Weigand, Edda (1999). Rhetoric and Argumentation in a Dialogic Perspective. In: E. Rigotti & S. Cigada (Eds.) Rhetoric and Argumentation (pp. 53-69). Tübingen: Niemeyer
Wierzbicka, Anna (1997). Understanding Cultures through Their Key Words. Oxford: Oxford University Press
Wierzbicka, Anna (1999). Emotions across Languages and Cultures: Diversity and Universals. Cambridge: Cambridge University Press
Wierzbicka, Anna (1992). Semantics, Culture and Cognition: Universal Human Concepts in Culture-Specific Configurations. New York: Oxford University Press.
Williams, Raymond (1975). Keywords: A Vocabulary of Culture and Society. London: Fontana

Sources of Examples:
Baudelaire, Charles (1975-1976). Mon coeur mis à nu. Oeuvres Complètes. Ed. C. Pichot. Paris: Gallimard
Bush, George W. (2002). President to send Secretary Powell to Middle East. Office of The Press Secretary. The White House. April 4, 2002. Available at: www.whitehouse.gov (June 12, 2002).
Milton, John (1953-1982). Areopagitica. A Speech for the Liberty of Unlicensed Printing to the Parliament of England. The Complete Prose Works of John Milton. Ed. D.M. Wolfe. (pp. 486-570). London/ New Haven: Yale University Press