1 Implicatures as Forms of Argument

The notion of implicature was introduced by Grice in The Causal Theory of Perception (Grice 1961) and de eloped in his further essays (Grice 19!")#1 The theory is meant to describe the relationship between what is said and what is meant in cases in which the con entional meanin$ commonly associated with the sentence (%what is said&) does not correspond with the spea'ers communicati e intention( or his)her intended effect on the audience (%what is meant&) (Grice 19!"* Grice 19+9( ,,-* .e inson 19+/( 9!)# 0mplicatures are therefore conse1uence of a spea'ers saying a specific sentence# 2n Grices iew( what a man says needs to be considered in the context of the expectations and presumptions of the community of speakers he belongs to (Grice 1975, 47). Grice collected such presumptions and expectations under general categories conceived as communicative norms (Sperber and Wilson 1986, 162) or presumptions. If an utterance superficially appears not to conform to the presumptions that the speaker is cooperative and is speaking truthfully, informatively, relevantly, perspicuously, and otherwise appropriately, the hearer tries to e3plain this failure by reconstructin$ a new uncon entional meanin$ for the sentence (4ach ,--/( 1"")# Grice concei ed these presumptions as $eneral and conte3t5independent communicati e principles( and they were reduced to principle of relevance by 6perber and 7ilson (Sperber and Wilson 1986, 382; Wilson and Sperber 2004, 608) and applied to several forms of communication. Grices proposal for analyzing implicit meaning in terms of a process of reasoning tri$$ered by a breach of dialectical e3pectations is based on three basic principles8 the meetin$ of the parties mind (or understandin$ of the intentions)( the e3istence of shared rules of communication( and the ability to reconstruct %what is meant& from %what is said&# These three principles ha e been studied from co$niti e perspecti es( transferrin$ the concept of meanin$ to the field of intentions and beliefs (Grice 1969* Thomason 199-) and applyin$ uni ersal principles (cooperation and rele ance) to e3plain all inferential processes# 9owe er( such ma3ims and principles are not uni ersal( but simply characteri:e some conte3ts of dialo$ue# 0n le$al dialo$ues or ne$otiations such ma3ims cannot hold (Goodwin 2001)# ;oreo er the crucial step between what is said and what is meant has been ne er described# The purpose of this paper is to in esti$ate the relationship between reasonin$ and implicitness# 7hat is the process of reasonin$ lin'in$ a sentence with an intention< 7hy do different people with different bac'$round 'nowled$e draw the same inferences in different circumstances< 7hy does the process of retrie in$ the implicit meanin$ ary accordin$ to the conte3ts of dialo$ue< 2ur proposal is to interpret the $rounds of Grices theory usin$ ar$umentation schemes in a dialectical framewor'( in order to show how implicatures wor' in this structure as implicit ar$uments# 1# Presumptions and meanin$ Grices theory of implicatures relies on the hearers ability to understand the intended communicative purpose of an utterance, that is, its meaning. The mechanism the speaker activates to retrieve the communicative intention is triggered by a failure of communicative expectations that are crystallized in four maxims or in one super-maxim.
1

=or an o er iew of the roots of Grices theory of implicature( see (>a is ,--!)#

2
Conceiving such presumptions as abstract norms, Grice reduced all communication to a some unspecified but evidently particular type of dialogue in which such rules are taken to hold. This ideal dialogue, however, hides the complexity of human conversation, whose regulating principles are context dependent. For instance, the most basic principle, the cooperative principle, does not hold in all contexts of dialogue. In legal crossexamination, the most important rules of the cross examination is that the lawyer needs to avoid possible evasions (Bronston v. U.S., 409 U.S. 359 (1973); Goodwin 2001). For instance, in this case, the lawyer asked a question, and retrieved the answer by resorting to implicatures.
Case 1: Presumption of evasion

"Q. Do you have any bank accounts in Swiss banks, Mr. Bronston?" "A. No, sir." "Q. Have you ever?" "A. The company had an account there for about six months, in Zurich."

The witness actually held a bank account in a Swiss bank, but was found to have testified truthfully, as he never stated the contrary. In this type of dialogue, the Gricean maxims, including the relevance maxim cannot hold. What is needed to account for such conversational differences is a weaker requirement. Instead of considering maxims and supermaxims as principles, we maintain that they are dialogical presumptions that depend on the type of dialogical purpose and dialogue type. In ordinary conversation the interlocutor is presumed to be cooperative and therefore willing to contribute to the dialogue. In other contexts of dialogue he is presumed to aim at winning the discussion, hiding some information, or trying to make the other party lose. In order to retrieve the intended meaning of what is said, or rather the purpose of a dialogical move, it is necessary first to understand the purpose of the dialogue the interlocutors are engaged in. Grice (19!"( ?") underscored this first re1uirement of interpretation introducin$ the notion of %direction& of the dialo$ue( which imposes conditions on the possible con ersational mo es8
2ur tal' e3chan$es do not normally consist of a succession of disconnected remar's( and would not he rational if they did# They are characteristically( to some de$ree at least( cooperati e efforts* and each participant reco$ni:es in them( to some e3tent( a common purpose or set of purposes( or at least a mutually accepted direction# This purpose or direction may he fi3ed from the start (e#$#( by an initial proposal of a 1uestion for discussion)( or it may e ol e durin$ the e3chan$e* it may he fairly definite( or it may he so indefinite as to lea e ery considerable latitude to the participants (as in a casual con ersation)# 4ut at each sta$e( 62;@ possible con ersational mo es would be e3cluded as con ersationally unsuitable

Aristotle first noticed that dialo$ical intentions can be broadly classified into types( dependin$ on the institutional and con ersational settin$ (Topics( B000( ")#
Inasmuch as no rules are laid down for those who argue for the sake of training and of examination - and the aim of those engaged in teaching or learning is quite different from that of those engaged in a competition; as is the latter from that of those who discuss things together in the spirit of inquiry. For a learner should always state what he thinks: for no one is

3
even trying to teach him what is false, whereas in a competition the business of the questioner is to appear by all means to produce an effect upon the other, while that of the answerer is to appear unaffected by him. On the other hand, in an assembly of disputants discussing in the spirit not of a competition but of an examination and inquiry, there are as yet no articulate rules about what the answerer should aim at, and what kind of things he should and should not grant for the correct or incorrect defense of his position. Inasmuch, then, as we have no tradition bequeathed to us by others, let us try to say something upon the matter for ourselves.

2n his iew( the participants institutional settin$ helps to understand their purposes and intentions in spea'in$( pro idin$ a blueprint for retrie in$ their specific dialectical $oals# 7alton (1989, 1990; further developed in Walton and Krabbe 1995; Walton 1998) pro ided a typolo$y of dialo$ue types (ne$otiation( information see'in$( eristic( persuasion dialo$ue( deliberation and in1uiry) in the form of a set of prototypical dialectical conte3ts defined by the interlocutors roles( includin$ their reciprocal 'nowled$e based on the dialectical settin$ (such as wor'( family( etc) and institutional roles (i#e# father( professor( businessman( etc)# A type of dialogue therefore can be considered as a normative framework in which there is an exchange of arguments between two speech partners reasoning together in turn-taking sequence aimed at moving toward the fulfillment of a collective conversational goal (Walton 1998, 30). On this perspective (7alton ,-1-( 1?) the global goal of the dialogue determines the local goal of the interlocutors specific moves made as speech acts, like asking a question or putting forward an argument (Macagno 2008). The argumentation that takes place in connected moves at this local level can be modeled by profiles of dialogue (Walton 1989, 65-71). For instance, in a decision-making dialogue about choosing a restaurant, the participants are presumed to bring forward arguments in favor of their preferences, express their opinions, and ask for more information. It would be taken to be unreasonable in such a setting to resort to threats, or to demand uselessly detailed information while putting off making a timely decision. Moreover, the interpretation of each move will be influenced and determined by the global purpose of the dialogue. The same discourse move I like eating at Ginos will be taken as a proposal and an implicit argument in a deliberation dialogue, as a report of personal tastes in an information-seeking dialogue, as a premise of a value judgment in a persuasion dialogue, as the denunciation of a personal offence in an eristic dialogue, or as the putting forth of an intention or goal in a negotiation. If we want to analyze the structure of the reasoning underlying the mechanism of interpretation, we need to change perspective, and conceive interlocutors communicative intentions as a coherence relation of the dialogue. Grimes (Grimes 1975, 209ff) interpreted abstract intentions as rhetorical predicates, later named logical-semantic connective (Crothers 1979) or coherence relations (Hobbs 1979, 68; Hobbs 1985; Rigotti 2005), such as explanation, alternative, support, etc. Such predicates, or relations, connect discourse sequences in two similar fashions, through subordination or coordination. In the first case the predicate is explicit and imposes a set of coherence conditions, or pragmatic presuppositions (Bander e'en ,--,( ?!) Bach 2003, 163), on its arguments (Grimes 1975, 162). In the second case, such a predicate hides a deeper relationship (Ballard, Conrad, and .on$acre 19!1) that needs to be reconstructed in order to understand the role of the discourse se$ments or se1uences# =or instance( coordination can e3press temporal( causal( e3planation

4
relations# =rom a pra$matic perspecti e( such relations can be considered as hi$h5le el speech acts (Grice 19+9( /6,* Carston ,--,( 1-!51-+)( indicatin$ the role of the first le el speech acts( or rather( its felicity conditions (Vanderveken 2002, 28)# If we structure the language used in terms of predicates, we can conceive the common goal as a high-level predicate (the discourse purpose), which assigns a role to each dialectical move, or discourse segments (Grosz and Sidnert 1986, 178; Walton 1989, 68). This relationship can be visualized in the structure shown below:
Case 2: Discourse coherence

B: Where shall we go for dinner? A: We can go to Bobs. It is good and really cheap. A*: I am going to an expensive restaurant. Go wherever you want.

Figure 1: purpose of a dialogue as a high-level predicate

Move 1* does not fulfill the role and the presuppositions of the global dialectical goal, as the purpose of advancing a proposal is not met by the discourse move. The pragmatic presuppositions of the dialogical predicate fail and the move can be described as an interruption of a dialogue, as opposed to a move that contributes to the dialogues properly moving along toward its goal. 2. Interpreting discourse relations: Predicate conditions and presumptions The collective purpose of the dialogue imposes some conditions on the possible dialogical moves. However, the interlocutors need to instantiate such generic conditions, and in order to do this they need to know what the interlocutor shares. For instance, if I want to make a proposal such as Lets go to Bobs, I take for granted that my interlocutor shares the meaning of the words uttered, the purpose usually conveyed by an imperative sentence, what is Bobs. However, we cannot know what our interlocutor knows now. All what we can tell is that imperatives can convey proposals, that speakers usually know the ordinary meaning of the words, and that usually people remember

5
where they have been before (see also Perelman and Olbrechts-Tyteca 1951). All what we can do is to presume such information. The concept of presumption is used to describe a particular type of inference based on general accepted principles stating how things usually happen. They are defeasible generalizations, and hold as true until the contrary is proven (see Rescher 1977, 26). Presumptions are forms of inference used in conditions of lack of knowledge. For instance, if it is not known that a person is guilty, he is presumed to be innocent because a person is normally innocent, unless he or she commits some crime. If something has happened in a certain place the previous day, such place can be searched for evidence because it is presumed not have changed meanwhile. Such inferences are therefore strategies to fill the gap of incomplete knowledge shifting to the other party the burden of providing the missing contrary evidence. Such principles of inference need to be shared and based on the ordinary course of events; in particular, the presumed fact needs to be more likely than not to flow from the proved fact supporting it. Discourse is based on presumptions. The speaker adopts a conversational behaviour because he presumes that the interlocutor cooperates or does not cooperate. He conveys an intention through an action because such an action is presumed to be associated with an intention. He uses a certain word to convey a specific meaning because it is presumed to have that meaning. Speakers can also breach some presumptions, because they presume that the hearer will rely on a higher-level of presumptions. For instance, indirect speech acts or tautologies are based on the same mechanism of breaching presumptions related to the relationship between speech acts and intentions, and words and meanings. On this perspective, interpretation is an argumentative activity that is carried out based on presumptions and breaches, or rather clashes, of presumptions. While the first type of argumentative process is simply grounded on presumptive, or defeasible modus ponens, (Lascarides and Asher 1991, 57) the second process, usually referred to as implicature, hides complex reasoning patterns aimed at explaining presumptive inconsistencies. On our perspective, implicatures are indirect speech acts of a kind (Bach 1994, 13), whose presumptive meaning differs from the intended one. Such a discrepancy, caused by a conflict of presumptions, need to be solved through a process of explanation. The first crucial aspect is how an implicature is triggered by presumptions. Consider the following interro$ati e sentences#
Case 3

1# Can you pass the salt< ,# 9a e you the physical ability to pass the salt<

6uch sentences can be both used to in1uire about a physical ability if uttered to a physically impaired person( or an interlocutor whose arm is in plaster# 9owe er( in a conte3t in which it is not e ident that such impairment is the case (1) would normally be considered as a polite re1uest( and (,) as a rebu'e# 4oth interro$ati e sentences are presumed to con ey the speech act of as'in$ a 1uestion( based on the intention of obtainin$ needed information# 9owe er( when as'ed in an ordinary conte3t( such presumption clashes with the stron$er one that usually people can perform ordinary actions# The presupposition of the presumed speech act (see Bander e'en and 6earle 19+"( 6656!) conflicts with the encyclopedic presumption and therefore the speech act cannot be felicitous# =rom such an interpretati e failure a need for e3planation arises(

6
which can be pro ided by interpretin$ the sentence as uttered to perform a non5 prototypical act# 0n such a case( the dialo$ical relation or predicate is reinterpreted# 7e can represent this process as follows8

Figure 2: Presuppositions of the indirect speech act

The spea'er( by utterin$ the interro$ati e sentence( e3ploits a presumption which is commonly shared( namely that interro$ati e sentences are used to as' 1uestions# 9owe er( such a mo e presupposes some conditions( and therefore ad ances some presumptions that are in conflict with other stron$er presumptions (a person in normal conditions is able to perform ordinary acts)# The e3planation therefore corrects the wea'er presumption (the spea'er intended to obtain information) with the most reasonable possibility (ad ance a re1uest)# 0n (,) the e3planation is more comple3( as the precision of the re1uested information enters in conflict with a conte3tual factor (e idence of the interlocutors ability)# The e3planation would be that the spea'er intended to communicate that the only acceptable reason why the hearer does not pass the salt can be that he is impaired# This process of reinterpretation of the discourse mo e (Asher and .ascarides ,--6) can be $uided by dialo$ical relations# =or instance( we can consider the followin$ cases of particulari:ed con ersational implicatures8
Case 4

A# 7ould you li'e to $o to the theatre toni$ht< 4# 0 ha e to study#

Case 5

A# >o you want another piece of chocolate ca'e< 4# 0s the pope Catholic<

These e3amples are forms of particulari:ed con ersational implicatures( but at the same time are indirect speech acts of a 'ind# 0n the first case the hearer( 4( uses an asserti e( but not to inform the spea'er that he has to study# The purpose of his mo e is to refuse

7
a proposal# 0n the second case( 4 does not want to as' a 1uestion in order to ac1uire some missin$ information( but to accept an offer# 0n such cases the interpretation of the non5paradi$matic mo e is $uided by the role imposed by the dialo$ical relation8 in the aforementioned cases( a proposal or a offer re1uire an acceptance or a refusal( and therefore 4s mo e is presumed to be a reply to the specific proposal or offer# 6ometimes interpretation is $uided by dialo$ical relations drawn from conte3tual information that specifies further the function of the dialo$ical mo e# =or instance( we can consider the followin$ e3chan$e between a sea5captain and his first mate (=ischer 19!-( ,!,)#
Case 6: Drunkard captain

The captain wrote in the shipCs lo$8 DThe first5mate was drun' all dayD# 7hen the first5mate read the lo$( he confronted the captain# The captain replied8 D7ell( it was true( wasnCt it<D# The followin$ day the first5mate( whose normal duties include writin$ up the shipCs lo$( $ot his re en$e# 9e wrote in the shipCs lo$8 DThe captain was sober all dayD#

0n this case( the implicature is guided by the dialogical predicate drawn from the contextual information that the sentence has been written on a logbook, and the presumption that on a logbook only exceptional events are recorded. We can represent the structure of the interpretation procedure as follows:

Figure 3: Explaining conflicting presumptions

This account of implicature shows a crucial relationship between interpretation and dialo$ue theory in two 'ey respects# =irst( implicatures need to be e3plained in terms of dialectical rele ance# And second( they need to be analy:ed as implicit ar$uments( in ol in$ a pattern of reasonin$ leadin$ from a specific premise to a conclusion# 6uch pra$matic and lin$uistic phenomena can be therefore inte$rated and de eloped within dialectical ar$umentation theory( and can be a startin$ point for de elopin$ ar$umentation theory into a theory of te3tual interpretation# 3. Conversational implicatures as implicit arguments 2n Grices iew( two types of implicatures are distin$uished8 con entional implicatures( dependent on the con entional meanin$ that certain words ha e( and $enerali:ed and particulari:ed con ersational implicatures( dependent on the use of a word or a sentence# The mechanism of implicature tri$$in$ and interpretation $uide set out abo e constitutes the bac'$round for analy:in$ the reasonin$ structure of particulari:ed implicatures# As seen abo e( dialectical predicates impose precise re1uirements on the se1uences of dialo$ue* when some inconsistencies arise between dialo$ical presumptions and encyclopedic presumptions( a process of reasonin$ is tri$$ered to reconstruct the specific purpose of the mo e( pro idin$ a specific

9
interpretation of the mo e# This process of reinterpretation is $rounded on specific reasonin$ patterns# /#1# Eeasonin$ from best e3planation

0n case 6 abo e the conflict between the presumed e3ceptionality of the captains soberness conflicts with the encyclopedic presumption that captains are usually sober on duty# This conflict is e3plained by retrie in$ the information that the captain is a drun'ard* howe er( in other conte3ts other inferences could ha e been drawn# =or instance( if a party was thrown on the ship( the captain would ha e been praised by that comment# 0f brou$ht before a court( the statement would ha e been interpreted literally( without pro idin$ any further e3planations# The reason of these possibilities can be found in the reasonin$ from best e3planation# Any conflict of presumptions needs to be e3plained( and possible solutions need to be found# 6uch solutions( howe er( conflict with presumptions drawn from the circumstances of the case# 2n this perspecti e( the stron$est( or best e3planation is the e3planation that in a $i en conte3t can be more hardly reFected by a counter5presumption# The abstract scheme can be represented as an instance of abducti e reasonin$( or inference to the best e3planation (7alton ,--,( ??)8
Argumentation Scheme 1: Reasoning from best explanation

F is a findin$ or $i en set of facts# E is a satisfactory e3planation of F# Go alternati e e3planation EC $i en so far is as satisfactory as E# Therefore( E is a plausible hypothesis( based on what is 'nown so far#

The ar$umentati e structure underlyin$ the particulari:ed implicature in case 6 can be represented as follows8

10
ASSERTION The captain was sober all day#
CONTEXT .o$boo'#

0G=2E; T9@ 9@AE@E of an exceptional fact#

Captains are usually sober8 the fact is not e3ceptional#

Pres.: The crew is usually sober on duty. Pres.: Captains are usually sober on duty. Pres.: Captains are usually healthy; alcohol is not a medicine Pres.: This captain as a captain is sober.

EXPLANATIONS 1# 2n such day all the crew was drun'* ,# Captains are usually drun'* /# The captain was sic' and he needed to drin'* ?# The captain was an alcoholic#

EXPLANATIONS 1# The 6pea'er is mista'en# ,# The fact that the captain was sober is e3ceptional#

Pres.: Speakers are rarely mistaken.

Pres.: Captains are usually sober on duty.

CONCLUSION The Captain was an alcoholic

Figure 4: Argumentative structure of the Drunkard Captain

This mechanism is based on the basic presumption that the spea'er is usin$ his discourse mo e for a specific purpose( which in some conte3ts is stron$er than the e3planation that he is simply mista'en# >ependin$ on the conte3t and the presumptions of dialo$ue( the process of e3planation ends at the first e3planatory step or proceeds further# A similar e3ample of this process can be found in the followin$ famous case of particulari:ed implicature (Grice 19+9( //)#
Case 7: Recommendation Letter

A professor supposed to write a recommendation letter for his student( writes %>ear 6ir( ;r# Hs command of @n$lish is e3cellent( and his attendance at tutorials has been re$ular# Iours#&

The described 1ualities cannot fulfill the role of %bein$ an e3ceptional 1uality for the position&* therefore( this conflict between dialo$ical presumption and encyclopedic presumption needs to be sol ed( and the best e3planation is that no other 1ualities can be found in the student# The reasonin$ can be represented as follows8

11
ASSERTION ;r# HCs command of @n$lish is e3cellent( and his attendance at tutorials has been re$ular#
CONTEXT Eecommendation letter

0G=2E; T9@ 9@AE@E of Hs e3ceptional s'ills for an academic position#

Command of @n$lish and attendance is not an e3ceptional s'ill#

Pres.: Speakers are reasonable. Pres.: Usually attendance is not an exceptional achievement.

Pres.: Everyone can attend classes.

Known.: Attendance is not a re uirement. Pres.: !ecommentation letters support the student"s application.

EXPLANATIONS 1# H was the only one able to attend the classes# ,# Attendance is an essential re1uirement for posts# /# H has no other academic 1ualities#

EXPLANATIONS 1# The 6pea'er is unreasonable# ,# The fact that H attended the classes is e3ceptional#

CONCLUSION H is not fit for the position#

Figure 5: Reasoning from best explanation Recommendation Letter

Eeasonin$ from best e3planation is not the only pattern of reasonin$ used in interpretation( but is the one that is directly tri$$ered by the conflict of presumptions# 2ther types of reasonin$ howe er may inter ene at the second le el of meanin$ reconstruction# /#,# Ar$ument from analo$y The second le el of the reasonin$ underlyin$ the reconstruction (or construction) of the meanin$ can be based on an ar$ument from analo$y# =or instance( we can analy:e the followin$ e3amples (Iule ,--+( ?/5??)8
Case 8

4ert8 >o you li'e ice5cream< @rnie8 0s the Pope Catholic<

Case 9

4ert8 >o e$etarians eat hambur$er< @rnie8 >o chic'ens ha e lips<

12
0n both cases( @rnies replies to ob ious 1uestions not pointin$ out or attac'in$ the tri iality of the answer( but ad ancin$ an implicit ar$ument that we can represent as follows (7alton 199"( 1/"51/6)8
Argumentation Scheme 2: Argument from analogy

;AJ2E PE@;06@8 Generally( case C1 is similar to case C2# ;0G2E PE@;06@8 Proposition A is true (false) in case C1# C2GC.K602G8 Proposition A is true (false) in case C2#

As ar$ued in (;aca$no and 7alton ,--9)( the proposition true in both cases is abstracted from the possible properties of the two compared cases# 7e can consider it as a functional $enus( an abstract cate$ory under which the two instances fall not absolutely( but only for the purpose of the ar$ument# =or instance we can consider the followin$ e3ample (Grimes 19!"( ,1!)8
Tryin$ to do lin$uistics without any reference to meanin$ would be li'e $oin$ into battle with one hand tied behind your bac'#

9ere the conclusion( correspondin$ to the implicit meanin$( or con ersational purpose( of the utterance( is that doin$ lin$uistics without considerin$ meanin$ is useless( or limitati e# 6uch a conclusion is drawn from abstractin$ a $eneral property that can be shared by the two parallel cases# 0n both cases + and 9( the meanin$s of the answers (ob iously yes)no) are abstracted from the analo$y between the two 1uestions* howe er( the implicit abstraction from the speech acts ad anced as analo$ous leads to other possible reconstructions# =or instance( in the first case the 1uestions can be $athered under the same cate$ory of %Luestions whose answer is 'nown to e eryone&( while in the second case under the $enus %6tupid 1uestions&# 0n both cases an implicit alue Fud$ment on the act of as'in$ the 1uestion can be drawn# 7e can represent the implicit ar$umentation of case + as follows8

13
UESTION A# 0s the Pope Catholic<
CONTEXT

Answer to the 1uestion Q# %>o you li'e ice5cream<&

E@P.I T2 T9@ 2==@E of an ice5cream#

A 1uestion is not an acceptance or a refusal (no reply to an offer)#

Pres.: Speakers are reasonable. Pres.: Usually interro#ative statements are not used to accept or re$use an o$$er.

ANALO!" 1# A is a 1uestion that is ob ious and its positi e answer is 'nown by e eryone (it should not e en be as'ed)# ,# B is a 1uestion and its answer is predictable (usually people 5 or the interlocutor 5 li'e ice5cream)# /# B is an 1uestion whose positi e answer is ob ious and it should not e en be as'ed#

EXPLANATIONS 1# The 6pea'er is unreasonable# ,# The spea'er uses an interro$ati e statement as a reply because it has a similar answer to the offer#

CONCLUSION Iou should not e en as' it (ob iously yes)#

Figure 6: Reasoning from analogy

2b iously this analysis is simply an appro3imation of a more comple3 mechanism# 4ehind e ery e3planation there is an abducti e reasonin$ (see .ascarides and Asher 1991) leadin$ from pieces of e idence (an interro$ati e sentence is used as a reply) to an e3planation (the two speech acts are similar in some respects)# /#/# Practical reasonin$ and other pra$matic ar$uments

6ometimes asserti e statements are used not to inform( but to lead the interlocutor to a specific action# 0n such cases( the inade1uacy of the assertion to perform the prototypical action it is associated with can be e3plained in terms of implicit pra$matic ar$uments( in which the stated sentence is the only e3plicit element# =or instance( we can consider the followin$ case (Grice 19!"( "1)8
Case 10

A8 0 am out of petrol# 48 There is a $ara$e around the corner#

The statement %0 am out of petrol& contradicts a presupposition of the dialectical predicate to inform the hearer( the fact that the hearer may be interested in the state information# The spea'er cannot presume that the hearer is interested in his condition(

14
or that the stated information is of any interest for him# Therefore( a mechanism of reinterpretation is tri$$ered8 lac' of petrol is a condition of need (causal relationship)( and as people are usually presumed to be interested in tryin$ to help who is need when it is possible( the conclusion is that the hearer should be interested in helpin$ the spea'er find some petrol# This conclusion( howe er( does not fulfill the re1uirement of a proper act of informin$( but only the re1uisite of an act of re1uestin$# The pattern of reasonin$ on which the lin' between need and re1uest of help (or appeal to pity) is $rounded can be represented usin$ the ar$ument scheme for appeal to pity (7alton 199!( 1-")8
Argumentation Scheme 3: Appeal to pity

0ndi idual x is in distress (is sufferin$)# 0f y brin$s about A( it will relie e or help to relie e this distress# Therefore( y ou$ht to brin$ about A#

0n this case( the re1uest for help aries accordin$ to the conte3t8 dependin$ on whether the interlocutor can pro ide some petrol( dri e the spea'er to a $as station( or help in other fashions( the precise interpretation of the relational predicate aries# 2n the other hand( 4s reply is an in itation to brin$ about a specific action( based on As needs# 9is information is therefore an implicit ar$ument from practical reasonin$( in which the needed premise (%in $ara$es it is possible to find some petrol&) and the conclusion (%you should $o to the $ara$e&) are missin$# The scheme is the followin$ (7alton 199,( +959-)8
Argumentation Scheme 4: Practical reasoning

;y $oal is to brin$ about A (Goal Premise)# 0 reasonably consider on the $i en information that brin$in$ about at least one of MB-(B1(###(BnN is necessary to brin$ about A (Alternati es Premise)# 0 ha e selected one member Bi as an acceptable( or as the most acceptable necessary condition for A (6election Premise)# Gothin$ unchan$eable pre ents me from brin$in$ about Bi as far as 0 'now (Practicality Premise)# 4rin$in$ about A is more acceptable to me than not brin$in$ about Bi (6ide @ffects Premise)# Therefore( it is re1uired that 0 brin$ about Bi (Conclusion)#

These two schemes connect information with actions( and therefore can e3plain se eral forms of implicatures in which asserti e sentences are interpreted as directi es# /#?# Causal ar$uments

The relationship between the presumed purpose of a discourse mo e( as imposed by the dialo$ical predicate( and the role such a mo e is presumed to play can be sometimes based on causal ar$uments# 7e can consider the followin$ implicature (.e inson 19+/( 1,6)8
Case 11

A8 7hat on earth has happened to the roast beef<

15
48 The do$ is loo'in$ ery happy#

The assertion on the do$s happiness is presumed not to fulfill the role of %pro idin$ information on the roast beef&# 9owe er( the relationship between %happiness of the do$& and %information on the roast beef& can be retrie ed throu$h an ar$ument from si$n (7alton ,--,( ?,)#
Argumentation Scheme 5: Argument from sign

Generally( if this type of indicator is found in a $i en case( it means that such5and5such a type of e ent has occurred( or that the presence of such5and5such a property may be inferred# This type of indicator has been found in this case 6uch5and5such a type of e ent has occurred( or that the presence of such5and5such a property may be inferred( in this case

The e3pression of the do$ is an indication of satisfaction* such satisfaction can be specified by the reference to a food( and therefore can be interpreted as the result of eatin$ the roast beef# The same relation from si$n can e3plain the followin$ implicature#
Case 12

A8 7hat time is it< 48 6ome of the $uests are already lea in$

9ere there are se eral possible reasons for $uests to lea e* howe er( the function of the discourse mo e is to pro ide a time indication# The lea in$ of the $uests is therefore presented as an indication of time( based on the presumption that late time causes $uests to lea e# 0n the aforementioned cases the information $uidin$ the interpretation and specification of the ar$ument scheme is pro ided by the dialo$ical information (conte3t)# 6uch a specification can be also pro ided by the shared information# 7e can consider the followin$ case (Carston ,--,( 1-9)(
Case 13

4ill $oes up to 6cotland e ery wee'end#

As Carston points out (Carston ,--,( 11-)( %in different specific conte3ts this could implicate C4illCs mother is illC( 4ill has a $irlfriend in 6cotlandC( C4ill $ets as far away from .ondon as he can when he canC( C4ill still hasnCt $ot o er his obsession with the .och Gess monsterC( etc#&# 9ere the si$n is pro ided by information about 4ill and specific presumptions related to specific beha iors# =or instance( the information that 4ill has a mother in 6cotland( or that he has not any $irlfriend in .ondon( or that he is obsessed with the 6cottish monster tri$$ers the presumption that %people tend to stay closer to what they lo e or care (or lo e or necessity cause people to desire to see each other)&* the fact that 4ill hates .ondon the presumption that %people tend to escape what they hate (or bad conditions cause people to lea e)&#

16
Ar$ument from si$n can be concei ed as an instance of causal ar$uments( lin'in$ an e ent to its effects# The $eneral scheme of ar$ument from cause can be represented as follows (7alton 199"( 1?-)8
Argumentation Scheme 6: Argument from cause

;AJ2E PE@;06@8 Generally( if A occurs( then B will (mi$ht) occur# ;0G2E PE@;06@8 0n this case( A occurs (mi$ht occur)# C2GC.K602G8 Therefore in this case( B will (mi$ht) occur#

A clear e3ample can be case ? abo e( in which the interlocutor instead of answerin$ ne$ati ely to the proposal of $oin$ to the theatre replies that he %has to study&# 9ere( the $oal of the utterance is not to inform the interlocutor about spea'ers duties( but to lead him to draw a conclusion from cause to effect# 6tudy is presumed to be incompatible with $oin$ to the theatre8 if someone studies( he needs to stay at home( and if someone stays at home( he cannot at the same time see a mo ie# 6uch latter causal relationship is presented as an alternati e8 either A or B* not A* therefore B# This representation of implicatures can e3plain both the reasonin$ mechanism underlyin$ reinterpretation of the discourse relation (such as in case 6( ! and 1-) or the reinterpretation of the role of the discourse mo e (or se1uence)# 0n all the aforementioned cases the reasonin$ depends on specific information pro ided by the conte3t and the co5te3t# 6ometimes( howe er( the ar$umentati e procedures are conte3t5independent and are simply related to the lin$uistic structure# Con#entional implicatures as implicit arguments 0f we analy:e Grices $enerali:ed and con entional implicatures in this framewor'( we can notice that they represent two types of implicitness( which we will refer to as implicature within predicates( or presuppositions( and implicatures within paradi$ms( or implicit ar$uments from alternati es# Grice notices that connectors li'e therefore con ey an implicit meanin$ which is within the word itself (4ach 1999)# =or instance we can consider the followin$ sentence (Grice 19+9( ,")#

11

9e is an @n$lishman* he is( therefore( bra e#

0t would be infelicitous to say %9e is an @n$lishman* it is( therefore( sunny outside&# Grice reali:es that therefore con eys the implicit information that there is a relationship between bein$ an @n$lishman and bein$ bra e# Consider8
11 4en is nice but he dri es a =ord Capri#

0n , there is a contrast between %4ob is nice& and %4ob dri es a =ord Capri& is ta'en for $ranted (Carston ,--,( 1-+)# 6uch con entional implicature can be e3plained as a presupposition of the predicate (see >ucrot 196+* 19!,* 4ar'er ,--/( ")( that is( a condition or re1uirement of the predicate affectin$ the felicity of the speech act# Grice (Grice 19+9( /!) pointed out another type of implicit content con eyed by some utterances# 9e analy:ed sentences of the followin$ 'ind#

17
?# "# 6# !# The fla$ is white X is meetin$ a woman this e enin$ X went into a house yesterday and found a tortoise inside the front door 6ome $uests are already lea in$

Grice (Grice 19+9( "15",) noted that such sentences( when uttered in any type of conte3t( always con ey the same implicit information( namely8
?a# The fla$ is not white and other colors ) the fla$ is white e en thou$h stained with blood# "a# X is meetin$ a woman that the interlocutor does not 'now# 6a# X went into a house that is not specific (not his house)# !a# Got all the $uests are already lea in$#

These inferences do not only depend on the semantic features of the 1uantifiers( but also on the paradi$m of possible alternati es# The inference is not simply drawn from an implicit content of the word( but from the structure of the word selection# 0f we consider lan$ua$e as a matter of choices between lin$uistic items in order to con ey a meanin$( dependin$ on the type of information we need to con ey we need to select a possibility while e3cludin$ others# =or instance( if we consider the followin$ sentence#
The computer is hea y#

7e ha e made a choice amon$ other predicates that can be attributed to computer( and con ey a specific type of information( namely information about its wei$ht# 7e can represent such a choice as shown in fi$ure +#

Figure 7: System of choices in language

The affirmation of a predicate( on such a perspecti e( is an implicit ne$ation of the other possible alternati es (see Pap 196-( "/) fallin$ within the same $enus( namely the same semantic characteristic which is conte3tually identified# Con entional

18
implicatures can be therefore described as a particular type of disFuncti e syllo$ism applied to natural lan$ua$e( in which an element is presented as an alternati e to a set of possible choices fallin$ within the same semantic cate$ory( or $enus# This type of reasonin$ is $rounded on Aristotles semantic principle expressed in his Topics (Topics IV, 1), stating that the genus cannot predicated of an entity unless one of its species is predicated of it too. 7e can represent the reasonin$ as follows8
Argumentation Scheme 7: Argument from alternatives

x is A A belon$s to the $enus O 0f somethin$ is O( it is either A( or B( or CP Therefore x is not B (not C( etc#)

0dentifyin$ the semantic $enus is crucial for the process of interpretation( because it also determines and specifies the e3act semantic feature of the noun to which the predicate is referred# =or instance( if 0 say that a computer is fast( 0 am considerin$ the computer as an electronic de ice( whereas if 0 describe it as hea y 0 am considerin$ it merely as an obFect# This structured framewor' can account for some other 'inds of implicatures# =or instance( if we consider the sentence %the fla$ is white& we implicate that the fla$ is only white( and not( for instance( white and red( e en thou$h it does not e3clude that it may be dirty (stained with blood) (.e inson 199"( 9!)# The semantic paradi$m is the symbolic color( and therefore the sentence specifies the fla$ as the most e ident part of the emblem( and not as fabrics( nor as the frame# The semantic paradi$m symbolic color not only specifies the paradi$m of predicates that can be attributed to the fla$( but also the concept of fla$# A fla$ cannot be transparent( nor can the colors be mi3ed in order to be identified as such# 0t is as an emblem that the fla$ is white( but it is as a piece of fabric that it is red with blood#

19

Figure 8: Semantic paradigms and the specification of aspects of meaning

6emantic paradi$ms therefore wor' by specifyin$ the semantic features of the obFect of the predication( and the possible choices# This structured process is a form of interpretation( since establishin$ the semantic predicate $o ernin$ the paradi$m means retrie in$ the meanin$ of the sentence# >ependin$ on the type of paradi$m( and therefore the $o ernin$ semantic predicate( or paradi$matic $enus( the meanin$ aries# Got only can the meanin$ of predicates be determined in this fashion( but also the interpretation of 1uantifiers# .ets reconsider cases " and 6# The indefinite article con eys the information that the noun which is determined is not 'nown by the other party( or is not specific for the spea'er himself (for a treatment of articles and their paradi$m of specificity see 9aspelmath 199!)# 0t contrasts with the definite article( which is used to represent the assumption that the entity is part of the domain of the hearers 'nowled$e or can be retrie ed by him# 0n the semantic paradi$m characteri:ed by the semantic trait of %uni1ueness&( only two alternati es are possible( namely the specific and non5specific article# The choice of the second e3cludes that the entity is 'nown( or 'nowable by the hearer# 6imilarly( considerin$ the more $eneric paradi$m of 1uantifiers( in ! some is contrasted with all( none and one# Therefore( the spea'er( by ma'in$ such choice( e3cludes that only one $uest is lea in$( that it is not true that no $uests are lea in$( and that not all $uest are lea in$# 6uch mechanisms of specification allow one to e3plain se eral types of processes of interpretation( from prepositions to 1uantifiers or predicates# Conclusion Grices theory of implicatures has been used in this paper to brin$ out aspects of the pre iously unclear relationship between reasonin$ and meanin$# 7e ha e shown how

20
con ersational implicatures represent implicit meanin$ tri$$ered by the use of a sentence and how they can be considered interpretations of the meanin$ of a word or a speech act# 2n this perspecti e( con ersational implicatures are tri$$ered by conflicts of dialo$ical and epistemic presumptions that are resol ed by a process of best e3planation( which in its turn are based ar$umentation schemes such as inference to the best e3planation( practical reasonin$( ar$ument from si$n( appeal to pity and analo$y# >ependin$ on the conte3t( the presumptions on which the process of e3planation is based on ary( and therefore the conclusions of the implicit ar$uments can be different# The ar$umentati e structure underlyin$ con entional implicatures stems from a different reasonin$ process( based on a lin$uistic structure# 0n this case( the interpretation of a words meanin$ is carried out throu$h the e3clusion of its alternati es within its semantic paradi$m( or $enus# References Aristotle# 1969# Topics# 0n The or!s o" Aristotle# e$. %illiam &oss. 23ford8 23ford Kni ersity Press# Asher( Gicholas( and Ale3 .ascarides# ,--6# 0ndirect 6peech acts# 'ynthese 1,+8 1+/5 ,,+# 4ach# Qent# 199?# ;eanin$( 6peech Acts( and Communication# 0n Basic Topics in the (hilosophy o" )anguage( ed# Eobert 9arnish( /5,1# @n$lewood Cliffs( GJ8 Prentice5 9all# 4ach# Qent# 1999# The myth of con entional implicature# )inguistics an$ (hilosophy ,,8 ,/!5/66# 4ach# Qent# ,--/# 'peech acts an$ pragmatics# 0n 4lac'well# Guide to the Philosophy of .an$ua$e( eds# ;ichael >e itt and Eichard 9anley( 1?!516!# 23ford8 4lac'well# 4allard .ee( Eobert Conrad( and Eobert .on$acre# 19!1# The >eep and 6urface Grammar of 0nterclausal Eelations# Foun$ations o" )anguage !( 18 !-511+# Barker, Stephen. 2003. Truth and Conventional Implicature. Mind 112, 445:1-33. Carston( Eobyn# ,--,# Thoughts an$ *tterances# ;alden8 4lac'well# Crothers( @dward# 19!9# (aragraph 'tructure +n"erence. Gorwood 8 Able3# >a is( 7ayne# ,--!# 9ow normati e is implicature# Journal of Pra$matics /9816""R 16!,# Ducrot, Oswald. 1968. Le structuralisme en linguistique. In Qu'est-ce que le structuralisme?, eds. Oswald Ducrot, Tzvetan. Todorov, Dan Sperber, Moustafa Safouan et Franois Wahl, 13-96. Paris : Seuil. Ducrot, Oswald. 1972. De Saussure la philosophie du langage, preface to John Searle, Les actes de langage, 7-34. Paris: Hermann. Fischer, David. 1970. Historians' fallacies. New York: Harper & Row. Goodwin( Jean# ,--1# The noncooperati e pra$matics of ar$uin$# 0n (ragmatics in 2,,,'electe$ papers "rom the .th +nternational (ragmatics Con"erence# vol. ++# e$. Eni!/ 0emeth# 21322... Ant erp- +nternational (ragmatics Association. Grice( Paul# 1961# The causal theory of perception# (rocee$ings o" the Aristotelian 'ociety( 6upplementary ol# /"8 1,1R1",# Grice( Paul# 19!"# .o$ic and Con ersation# 0n The .o$ic of Grammar( >onald >a idson and Gilbert 9arman( 6?5!"# @ncino( California8 >ic'enson#

21
Grice( Paul# 19+9# 'tu$ies in the ay o" or$s# Cambrid$e( ;ass8 9ar ard Kni ersity Press# Grimes( Joseph# 19!"# The threa$ o" $iscourse# The 9a$ue8 ;outon# Gros:( 4arbara( and Candice 6idnert# 19+6# Attention( intentions( and the structure of discourse# Computational .in$uistics 1,( /8 1!"5,-?# 9aspelmath( ;artin# 199!# +n$e"inite pronouns# 23ford8 23ford Kni ersity Press# 9obbs( Jerry# 19!9# Coherence and coreference# Cognitive science / 8 6!59-# 9obbs( Jerry# 19+"# 2n the Coherence an$ 'tructure o" 3iscourse. Eeport Go# C6.05+"5 /!( Center for the 6tudy of .an$ua$e and 0nformation( 6tanford Kni ersity# .ascarides( Ale3( and Gicholas Asher# 1991# >iscourse Eelations and >efeasible Qnowled$e# 0n Proceedin$s to the ,9th Annual ;eetin$ of the Association of Computational .in$uistics (AC.91)( Ber!eley *'A( ""56/# 6troudsbur$( PA8 Association for Computational .in$uistics# .e inson( 6tephen# 199"# Three le els of meanin$# 0n Grammar and meanin$( ed# =ran' Palmer( 9-511"# Cambrid$e8 Cambrid$e Kni ersity Press# .e inson( 6tephen# 19+/# (ragmatics( Cambrid$e8 Cambrid$e Kni ersity Press# ;aca$no( =abri:io# ,--+# >ialectical Eele ance and >ialectical Conte3t in 7altons Pra$matic Theory# +n"ormal )ogic ,+ (,)8 1-,51,+# ;aca$no( =abri:io( and >ou$las 7alton# ,--9# Ar$ument from Analo$y in .aw( the Classical Tradition( and Eecent Theories# (hilosophy an$ &hetoric ?,( ,8 1"?51+,# Pap( Arthur# 196-# Types and ;eanin$lessness# 4in$ 69( ,!/8 ?15"?# Perelman( Chaim( and .ucie 2lbrechts5Tyteca# 19"1# Act and Person in Ar$ument# Ethics 61( ?8 ,"15,69# Eescher( Gicholas# 19!!# 3ialectics- a controversy2oriente$ approach to the theory o" !no le$ge# Albany8 6tate Kni ersity of Gew Ior' Press# Ei$otti( @ddo# ,--"# Con$ruity Theory and Ar$umentation# 'tu$ies in Communication 'ciences( 6pecial 0ssue8 !"596# 6perber( >an( and >eidre 7ilson# 19+6# &elevance- Communication an$ cognition# 23ford8 4lac'well# Thomason( Eichmond# 199-# Accommodation( meanin$( and implicature8 0nterdisciplinary foundations for pra$matics# 0n 0ntentions in Communication( eds# Philip Cohen( Jerry ;or$an( and ;artha Pollac'( /,"R/6/# Cambrid$e( ;A8 ;0T Press# Bander e'en( >aniel( and John 6earle# 19+"# Foun$ations o" illocutionary logic# Cambrid$e8 Cambrid$e Kni ersity Press# Bander e'en( >aniel# ,--,# Kni ersal Grammar and 6peech act theory# 0n @ssays in 6peech Act Theory( eds# >aniel Bander e'en and 6usumu Qubo( ,"56,# Amsterdam8 John 4enFamins# 7alton( >ou$las( and @ri' Qrabbe# 199"# Commitment in >ialo$ue# Albany8 6tate Kni ersity of Gew Ior' Press# 7alton( >ou$las# 19+9# +n"ormal )ogic# Gew Ior'8 Cambrid$e Kni ersity Press# 7alton( >ou$las# 199-# 7hat is Eeasonin$< 7hat is an Ar$ument< 5ournal o" (hilosophy +!8 /995?19# 7alton( >ou$las# 199,# 'lippery 'lope Arguments# Gewport Gews8 Bale Press# 7alton( >ou$las# 199"# A Pragmatic Theory of Fallacy. Tuscaloosa and London: The University of Alabama Press.

22
7alton( >ou$las# 199!# Appeal to Pity. Albany: State University of New York Press. 7alton( >ou$las# 199+# The 0e 3ialectic. Conversational Contexts o" Argument# Kni ersity of Toronto Press( Toronto# 7alton( >ou$las# ,--,# )egal Argumentation an$ Evi$ence# Kni ersity Par'8 The Pennsyl ania 6tate Kni ersity Press# 7alton( >ou$las# ,-1-# Types of >ialo$ue and 4urden of Proof( 0n Computational ;odels of Ar$ument8 (rocee$ings o" C644A 2,1,# e$. Pietro 4aroni( =rederico Cerutti( ;assimilano Giacomin( and Guillermo 6imari( 1/5,?# Amsterdam8 026 Press# 7ilson( >eirdre( and >an 6perber# ,--?# Eele ance theory# 0n 7an$8oo! o" (ragmatics( ed# .aurence 9orn and Gre$ory 7ard( 6-!R6/,# 23ford8 4lac'well# Iule( Geor$e# ,--+# (ragmatics# 23ford8 23ford Kni ersity Press#

Abstract
0n this paper( we use concepts( structure and tools from ar$umentation theory to show how con ersational implicatures are tri$$ered by conflicts of presumptions# Presumptive implicatures are shown to be based on defeasible forms of inference used in conditions of lack of knowledge( includin$ analo$ical reasonin$( inference to the best e3planation( practical reasonin$( appeal to pity( and ar$ument from cause# Such inferences are modelled as communicative strategies to knowledge gaps that shifting the burden of providing the missing contrary evidence to the other party in a dialogue. Through a series of illustrative examples, we show how such principles of inference are based on common knowledge about the ordinary course of events shared by participants in a structured dialogue setting in which they take turns putting forward and responding to speech acts.