‘I don’t know how it is best to put this thing down’: Uncooperative narration in Ford Madox Ford’s The Good Soldier

This article offers a pragmatic stylistic analysis of Ford Madox Ford’s 1915 novel The Good Soldier, assessing the technical notion of ‘cooperation’ developed in Gricean theory as a means of explaining its distinctive narrative style. Critics have generally agreed this text offers a salient example of unreliable narration, but have differed significantly in how they judge the character of the narrator himself, and even in how they interpret the events of the novel. Grice characterises communicative exchanges as typically rational pieces of behaviour, in which a mutually accepted set of purposes guarantees that individual contributions are suitable and constructive, but he does also detail ways in which speakers may fail to fulfil these expectations. This article will analyse examples from the novel in which the relationship between narrator and reader might be characterised as uncooperative in specifically Gricean terms; the narrator faces a clash of conversational maxims, opts out of the maxims or violates them, rather than either adhering to them or flouting them for communicative effect. This suggests a way in which pragmatic literary stylistics might be able to offer a principled explanation of the idiosyncratic style of this canonical work of modernist fiction, and by extension of the effects of this style of narration on the reader.


Introduction
The main aim of this article is to offer a pragmatic stylistic analysis of Ford Madox Ford's 1915 novel The Good Soldier (Ford, 1915, henceforth TGS).The narrator of this text has widely been classified as 'unreliable', to the extent that the reader is likely to experience considerable confusion.This confusion is mirrored in the variety of interpretations offered by critics of the novel, who disagree as to whether the narrator is deluded, deceived or deceitful, and even as to what actually happens in the story he is telling.
The discussion of the narrative style of TGS offered here draws on Gricean pragmatics.Grice's theory of conversation is centred on the conception of communication as a cooperative enterprise, and his principle and maxims offer a detailed account of the expectations concerning speaker behaviour which contribute to communicative success.The theory also takes account of reasons why speakers may act in apparent or actual contravention of these expectations.Speakers may flout a particular maxim or maxims, with the intention of conveying an implicated meaning.In these cases, the overall cooperative nature of the interaction is maintained.But sometimes a speaker may choose to opt out of one or more of the maxims, may be faced with an irreconcilable clash between two maxims, or may simply violate a maxim by breaking it, possibly with an intention to deceive.In such cases the speaker might be said to be behaving in a manner which is either noticeably or surreptitiously uncooperative.This paper is concerned with whether the narrator in TGS can be characterised as behaving uncooperatively in this precise Gricean understanding of the term.
Pragmatic theory has of course been added to in many significant and important ways since the first publication of Grice's theory of conversation.Others have built on, adapted and critiqued his insights to develop, for instance, various versions of neo-Gricean pragmatics (e.g.Levinson, 2000;Horn 2007) and relevance theory (Sperber and Wilson, 1995).These developments have been reflected in the field of pragmatic literary stylistics, where stylisticians have utilised the work of Levinson (e.g. Pattison, 2019), Horn (e.g.Chapman, 2012) and Sperber and Wilson (e.g. Clark, 2009) to offer valuable insights into a range of literary texts.The focus in this article on the 'original' Gricean model, is in no way intended as a rejection of these developments or of their utility to stylistic analysis.As I have argued elsewhere, different pragmatic theories are not most usefully viewed as 'competing' with each other; rather, 'specific approaches adopted should be those which seem to provide insights into the specific phenomena being considered' (Chapman and Clark, 2014: 7).The specific phenomenon of cooperation, or lack of it, in the communication between narrator and reader seems particularly pertinent to a consideration of TGS.The Gricean approach to pragmatics, with its emphasis on this very notion, would seem to lend itself particularly to a stylistic treatment of this text.
In addition, there is an ongoing discussion concerning the utility of Grice's theory as an approach to the nature of literary discourse in general and to unreliable narration in particular.A secondary aim of this article is to contribute to this discussion.Some early participants in the discussion claimed that the Gricean notion of cooperation simply does not apply in the case of literature, which has its own separate set of pragmatic rules (e.g.van Dijk, 1976).Others maintained that Grice's account applies as much to the interaction between a literary text and its readers as it does to that between conversational partners, but that the narrators of some literary texts may display uncooperative tropes of the type identified by Grice (e.g. Pratt 1977).In such cases, there is usually some clue or signal to the reader as to the nature of and the reasons for the narrator's uncooperative behaviour.The approach taken here is broadly in accord with this view.However, it is often difficult or impossible for the reader of TGS to establish with any certainty the reasons why the narrator is behaving in an uncooperative manner, which may explain the confused and sometimes contradictory responses the novel has elicited.This article explores the possibility of using Grice's theory of conversation as a tool for the description and explanation of the distinctive narration of TGS.In more general terms, it contributes a supportive case study to the claim that the theory can be used to offer a principled account of the phenomenon of unreliable narration in literary texts.
Previous discussions of unreliable narratives in relation to Gricean pragmatics are considered in section 3, followed in section 4 by a qualitative study of TGS using this framework.Before that, section 2 offers a brief overview of the novel, and a survey of some of the critical response it has attracted.

The Good Soldier
The timing and even the order of the fictional events described in TGS are difficult to pin down with any certainty.The narrative concerns the lives of a group of protagonists in the early years of the 20th century, extending up to and perhaps even beyond the book's publication in March 1915 (see Cheng, 1986, for a valiant attempt to reconstruct the chronology of TGS).The first person narrator is John Dowell, a wealthy and leisured American travelling in Europe with his wife Florence, to whom he is devoted.The Dowells strike up a friendship with an English and Irish couple, Edward and Leonora Ashburnham, who are accompanied on their European travels by Nancy, who is Leonora's niece and Edward's ward.Edward is the titular good soldier, a retired army officer whom Dowell admires for his good humour, physical prowess and generous nature.Florence and Edward purportedly suffer from weak hearts, which necessitate both couples undertaking prolonged stays in the spa town of Nauheim every summer.They continue this arrangement for 9 years, until the sudden death of Florence, apparently from a heart attack.The Ashburnhams return to England and Dowell to America, where he stays until Edward and Leonora separately but simultaneously send him telegrams asking him to join them.He arrives in England in time to witnesses the departure of Nancy, who is sent to India to live with her father.Shortly afterwards, a telegram arrives from India and Edward immediately commits suicide by slitting his own throat with a penknife.
There is a parallel version of events, which Dowell claims to learn from Leonora only after Edward's death.Both Florence and Edward had deceived Dowell, and Leonora was complicit in the deception.Florence had married him in order to be able to follow her lover, Jimmy, to Europe.The Ashburnhams' marriage was a sham; Edward was a serial adulterer and soon after their first meeting had begun an affair with Florence.He had subsequently fallen in love with Nancy and when Florence discovered this she committed suicide by poison.Edward had arranged for Nancy to be sent away for his peace of mind, but when he received her telegram indicating that she was carefree and happy he despaired and took his own life.By the end of the novel Leonora has remarried and Dowell has bought the Ashburnhams' country estate where he lives with Nancy, whom he brought back from India after she lost her sanity at the news of Edward's death.
Throughout TGS, the rambling, disjointed and often self-contradictory nature of the narrative make even the basic plot details hard to decipher, let alone issues relating to motivation, knowledge and understanding.As an example, here are the opening sentences of the novel.The reader is in danger of beginning the book confused as to whether this is a second hand story or a personal reminiscence, what the nature of the relationship was between the Ashburnhams and the Dowells, and even how the present tense commentary on 'today' relates temporally to the events of the 'story'.This is the saddest story I have every heard.
We had known the Ashburnhams for nine seasons in the town of Nauheim with an extreme intimacyor rather, with an acquaintanceship as loose and easy and yet as close as a good glove's with your hand.My wife and I knew Captain and Mrs Ashburnham as well as it was possible to know anybody and yet, in another sense, we knew nothing at all about them.This is, I believe, a state of things only possible with English people of whom till today, when I sit down to puzzle out what I know of this sad affair, I knew nothing whatever.Six months ago I had never been to England and, certainly, I had never sounded the depths of an English heart.I had known the shallows.
(TGS: 9) TGS has attracted wide and diverse critical attention.There is general agreement that the novel offers 'clinching evidence of the essential untrustworthiness of narration' (Smith, 1972: 345), or even 'the most perfectly deployed example of the unreliable narrator' (Barnes, 2008: 420).Many critics link this to Ford's commitment to literary impressionism.Ford argued that the business of literature was to present subjective understanding; human beings could not experience reality objectively, and as a result literature should focus on the perceiving mind, rather than attempting to describe an external reality that was inherently unknowable.Life did not narrate and therefore 'the general effect of a novel must be the general effect that life makes on mankind' (Ford, 1924: 290).Some critics have seen this as the implicit message of TGS.Compared to the omniscient narrative of 18th century novels, TGS 'implies a more limited and tentative conception of the way man knows' (Hynes, 1961: 327).The novel tells us that '[w]e are in a world of which it need to be said not that plural readings are possible (for this is true of all narrative) but that the illusion of the single right reading is possible no longer.'(Kermode, 1974: 353, original emphases).For some, the strange narrative style challenges the possibility of knowledge itself: 'Dowell's narration confounds the distinction on which knowledge is basedthe distinction between illusion and reality' (Goodheart, 1986: 384); TGS undermines any belief in 'our ability to know or perceive accurately' (Nigro, 1992: 383); '[t]he book foregrounds the difficulty of knowing and understanding' (Saunders, 1996: 401).
There is disagreement, however, about the reasons for the unreliable narration; critics have different views about Dowell as a character, and about how the reader is to respond to him.For some, Dowell's faults are grounded in some sort of psychiatric pathology.
Schorer refers to 'a mind not quite in balance' and the 'distorted understanding of the narrator' (Schorer, 1948: 324), while Meixner argues that Dowell displays 'a severely neurotic personality' (Meixner, 1962: 336).McCarthy suggests a more moderate diagnosis, but still one which focuses on inability of Dowell's part; he 'is not very imaginative or perspective; he has a plodding, meticulous mind' (McCarthy, 1997: 142).De Angelis has a different explanation; Dowell's narrative style can be explained by his powerful but supressed desire for Edward, which has left him 'closeted and confused' (De Angelis, 2007: 426).
Some critics judge Dowell leniently, despite his deficiencies as a narrator.Armstrong draws attention to the need for the reader to make a decision as to whether to trust or suspect Dowell, suggesting that '[b]oth are guesses about the hidden or as yet undisclosed', but deciding on balance to trust him (Armstrong, 1987: 397).Levenson sees Dowell as erratic in his narratorial behaviour, but basically morally laudable; he is 'the one character capable of selflessness' (Levenson, 1984: 372).Those critics who take a less charitable view of Dowell argue that his intent is to deceive, and question his version of what he knows and even of what takes place in the novel.For Griswold, in adopting a rambling style and ineffectual persona, 'Dowell is merely trying to hoodwink the reader' (Griswold, 2017: 156) and is concealing the fact that he murdered Florence.Nigro also has '[a] strong sense that Dowell is hiding somethingthat this habit of narrating in bits and pieces is simply a retelling of the story in half-truths' (Nigro, 1992: 385); he speculates that Dowell murdered both Florence and also Edward.
There is some consensus, then, that Dowell is an unreliable narrator and that Ford used this as a device to communicate his own views on the purposes of literature.But it is striking just how much room for disagreement the text of TGS leaves concerning the reasons for the unreliability, and the types of readings of the novel that these might prompt.The general question of how people interpret language presented to them in communicative encounters is fundamental to pragmatics in the Gricean tradition, and the issue of how readers respond to narratives is the central concern of pragmatic literary stylistics.The next section is concerned with what Grice said about how people behave in communicative encounters, particularly about behaviour which is for some reason uncooperative, and with how his ideas have previously been discussed in relation to unreliability in literary narration.Section 4 returns to TGS, testing out the possibility that Dowell is uncooperative in a manner modelled in Grice's theory of conversation.

Cooperative and uncooperative behaviour in the Gricean tradition
At the heart of Grice's theory of conversation is his conception of talk exchanges as characteristically cooperative efforts, summed up in his claim that participants are generally expected to observe the Cooperative Principle: '[m]ake your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged' (Grice, 1975: 26).Grice's maxims of conversation, which he groups under the headings of Quantity, Quality, Relation and Manner, provide detail as to how cooperation operates in practice.Hearers may legitimately assume that speakers are giving them sufficient information which is backed up by adequate evidence and which is germane to the topic of the conversation, and they may further assume that the information is being conveyed with appropriate clarity.On this basis, the hearer is entitled to supply information not explicitly given but necessary to support the assumption of cooperation.If A states 'I am out of petrol' and B replies 'There is a garage round the corner', B may legitimately infer that A is conversational implicating that the garage is open and has petrol; otherwise she would have been infringing the maxim 'be relevant'.
Almost immediately on introducing the conversational maxims, Grice details ways in which speakers may fail to fulfil them.He is under no illusion that conversations always proceed in the neat, well-ordered way that would result from strict adherence to the maxims.He sees these lapses from the ideal not as unfortunate exceptions or messy counterexamples, but as types of human behaviour which can be categorised and explained and which may have communicative functions of their own.The type of non-fulfilment which most interested Grice is flouting.The deliberate and ostentatious non-fulfilment of a maxim leads a hearer to seek an interpretation which would preserve the assumptions which underlie conversation.Importantly, these are not examples of uncooperative behaviour; Grice makes it quite clear that 'though some maxim is violated at the level of what is said, the hearer is entitled to assume that that maxim, or at least the overall Cooperative Principle, is observed at the level of what is implicated' (Grice, 1975: 33).A favourite example is the reference letter for an academic job in philosophy which reads: 'Dear Sir, Mr. X's command of English is excellent, and his attendance at tutorials has been regular.Yours, etc.'.On the face of it, by not providing adequate information to assess the candidate for the job, the referee has breached the first maxim of Quantity, which requires contributions to be sufficiently informative.The implicated meaning is, of course, that the referee thinks that Mr. X is not a good philosopher, and this is likely to be what the appointment committee will understand, based on their assumption that the referee is being cooperative in delicate circumstances.
Grice also considers what may occur when a speaker is faced with a clash between two maxims.A and B are planning a holiday in France, both knowing that A would like to visit C if possible.A asks 'Where does C live?' and B replies 'Somewhere in the South of France'.B's response is certainly suboptimal in relation to the first maxim of Quantity; it does not provide adequate information for A's purposes.But A may legitimately assume that the reason for the vagueness is that B cannot be more specific without going against the second maxim of Quality, which requires speakers only to make statements for which they have adequate evidence.The speaker's behaviour is still motivated by an awareness of the Cooperative Principle, and still conveys an implicature.But implicated meanings in such cases concern speakers, and their own states of ignorance or inability; 'B implicates that he does not know in which town C lives' (Grice, 1975: 33).
A less cooperative form of behaviour is when a speaker chooses to 'opt out' of the maxims and the Cooperative Principle.Instead of writing the under-informative reference, the philosophy tutor could simply not have written at all.In some cases, however, speakers may indicate awareness of the usual expectations of interaction by drawing explicit attention to what they are doing; 'he may say, indicate, or allow it to become plain that he is unwilling to cooperate in the way the maxim requires.He may say, for example, I cannot say more; my lips are sealed' (Grice, 1975: 30).That is, speakers may choose to indicate that, and perhaps why, they are not adhering to the usual expectations of cooperation.
The most uncooperative form of behaviour identified by Grice is the one about which he says least.A speaker 'may quietly and unostentatiously violate a maxim; if so, in some cases he will be liable to mislead' (Grice, 1975: 30, original emphasis).Unlike in the cases of flouting, facing a clash, or opting out, the hearer is unaware of what the speaker is doing, and is allowed to proceed in the belief that the speaker is being cooperative.Speakers who offer information which they believe to be false, for instance, are in direct contravention of expectations of Quality.Instances of unostentatious maxim violation could be seen as parasitic on normal expectations of cooperation; they will fulfil the speaker's intention to mislead precisely if the hearer is unaware of that intention.

Applications to literary analysis
Grice's ideas were soon recognised as potentially important to literary studies.They chimed with contemporary interest in explaining literature as a discourse, or an interactive relationship between text and reader.There was disagreement, however, as to the extent to which the Cooperative Principle and maxims could be taken to apply to literary discourse.For van Dijk, literary discourse is fundamentally different from other types of interactions, introducing the need for a separate, although analogous, set of maxims constituting the 'Construction Principle' of literary communication.It is mutual knowledge between 'speaker' and 'hearer' that the sentences uttered are not true in the factual world, so there is no intention to deceive, and indeed no expectation of cooperation from the speaker; 'the speaker has simply "opted out" (Grice) from the contextual principles of ordinary conversation' (van Dijk, 1976: 46).
Pratt argues that, on the contrary, the discourse of literature is fundamentally and essentially cooperative.In writing and publishing a novel, an author has volunteered an utterance; 'volunteering an utterance is always opting in, and one cannot opt in and out at once' (Pratt, 1977: 172).However, Pratt draws a distinction between the author, who has opted in and therefore guarantees a cooperative encounter, and the narrator, with whom the reader can expect no such contract.In this she prefigures ideas which were to become central to present day stylistics.In Short's 'discourse structure of fictional prose', the 'narrator-narratee' level of discourse is embedded within the overarching 'author-reader' level (Short, 1996: 256).Pratt observes that 20th century novels are in fact characterised by narrators who deviate in various ways from the expectations of cooperation.She notes that the types of nonfulfillment of a maxim which Grice discusses are variations of knowing or intentional nonfulfillment.To explain unreliable narration it is also necessary to consider unintentional nonfulfillment, which comes about 'for lack of verbal skill or for some other reason' (Pratt, 1977: 159).Unintentional nonfulfillment of a maxim might be described as a type of unostentatious violation, because the maxim is infringed without any intention that this infringement will be noticed, or that some implicated meaning will be recovered.However, it differs from the type of unostentatious violation which Grice identifies, in that the speaker or narrator is not deliberately setting out to deceive the audience.
An infringement of a maxim on the part of an unreliable narrator, functioning as an uncooperative violation, may at the same time be recognised as flouting in the separate, securely cooperative communication between author and reader.As such, it can lead to conversational implicatures being communicated by the author while none are intended by the narrator.In the case of the meandering and disorganised narrative of Sterne's Tristram Shandy, for instance, 'Sterne is immediately understood to be flouting the rules for narrative and the rules for writing and to be observing the CP at the level of what is implicated'.He may be implicating, 'a critique of contemporary stylistic norms or orientative information about Shandy's psychological make-up, which we will need to respond as desired to what follows' (Pratt, 1977: 165-6).If the narrator violates expectations of cooperation, it will be made clear in the discourse between author and reader why this is the case: '[i]n all cases of unintentional failure on the part of the fictional speaker, the author implicates the cause of the failure and calls our attention to the contrast between what the speaker said and what we might have expected him to say' (Pratt, 1977: 182).
Not all critics agree with Pratt's assessment that unreliability is always unintentional.Olson argues for a distinction between fallible and untrustworthy narrators.In the former case, the unreliability is caused by external circumstances rather than inherent characteristics; narrators may be 'mistaken about their judgments or perceptions or are biased' (Olson, 2003: 101).Untrustworthy narrators however, behave as they do because of internal factors such as behaviour traits or self-interest.Narrators can move between the two, but readers will 'prefer making more straightforward attributions of fallibility or untrustworthiness' (Olson, 2003: 104).Working in a Gricean framework, Heyd describes unreliability as 'a deviation from standard communication patterns' which can be described as 'a violation of the Cooperative Principle' (Heyd, 2011: 3).She recognises a scale concerning intentionality or self-consciousness rather than two discreet categories, and explains that: '[t]he scale runs from utterly intentional CP violations, which are particularly morally deviant, through semi-conscious unreliability, which is the most "life-like" rhetorical strategy, to unintentional unreliability, which amounts to a cancelling of the cooperative basis of communication' (Heyd, 2006: 233).
Other critics argue that in some texts different types of unreliability may not be easily distinguishable from each other, which may in part account for the difficulty readers encounter with those texts.Steiner (1998) offers a Gricean account of The Good Soldier Švejk and His Fortunes in the World War, by Jaroslav Hašek (a novel with a coincidentally similar title to TGS, published a short time later, in 1921-23).Steiner is analysing language used by a character in the novel rather than the narrative voice itself, but it is the language of a character who narrates many stories within the novel and who functions as 'a partial configuration of a larger novelistic structure' whose 'discourse often conveys the narrator's point of view' (Steiner, 1998: 494).The text 'does not furnish enough clues' to the cause or the nature of the various types of uncooperative behaviour it manifests.Readers 'are unable to pin down Švejk's nonfulfillment of conversational maxims as unintentional failures or as quiet, unostentatious violations intended to mislead' (Steiner, 1998: 493).
Kukkonen distinguishes between unreliable narrators and the more general category of uncooperative narrators.Unreliable narration 'violates the maxim of quality in that it is not truthful and evidenced' (Kukkonen, 2013: 208), while uncooperative narrators 'can violate, present clashes between, opt out from or flout any of the maxims' (Kukkonen, 2013: 207).The distinction is an instructive one, but Kukkonen's definition of the uncooperative narrator is problematic, since it includes any type of infringement of the maxims.Flouting maxims often occurs in cooperative communicative encounters, and is indeed a chief way in which a speaker or narrator may convey a conversational implicature.It might be expected that uncooperative narrators would be relatively unlikely to flout maxims for communicative purposes.They would, however, be likely to display behaviour resulting from clashes of maxims, explicitly obfuscatory examples of opting out and uncooperative violations.These expectations will be investigated with reference to TGS in the next section.

Uncooperative narration in The Good Soldier
In this section, selective but representative examples are used to assess whether the unreliable narration of TGS can be explained in relation to the varieties of uncooperative behaviour identified by Grice.Each of the types of maxim nonfulfillment will be considered in turn: facing a clash of maxims, opting out and violating.Flouting will also be considered, in relation both to the narration itself, and to the separate interaction between author and reader.As discussed above, flouting is not in itself an uncooperative form of behaviour, and it might be expected that it will occur less frequently than the other types of nonfulfillment in Dowell's narration.However, Pratt predicted that violations of maxims on the part of the narrator function as communicative floutings on the part of the author, who will make the reasons for the violation apparent to the reader.These claims will be considered in relation to TGS.

Facing a clash of maxims
A clash of maxims is likely to result from some deficiency on the part of the speaker, specifically a lack of knowledge.In Grice's example, B is unable to give a fully cooperative reply to A's question.His answer, 'Somewhere in the South of France', conveys an implicature to A that B does not know in which town C lives.Grice describes B's answer as 'less informative than is required to meet A's needs' and therefore as an 'infringement of the first maxim of Quantity' (Grice, 1975: 32).The implicature is triggered by the use of a relatively vague expression (introduced by 'Somewhere…') in a context in which a more specific expression (such as the name of a particular town) would be more useful.
On a number of occasions, Dowell uses a relatively vague term, indicating that he does not have access to the more definite information expected from a narrator.For instance, telling the story of a medieval French love triangle, which may or may not have a bearing on his main narrative, he explains: Las Tours of the Four Castles had for chatelaine Blanche Somebody-or-other who was called as a terms of commendation La Louvethe She-Wolf.… So Piere Vidal declared himself Emperor of Jerusalem or somewhere … And Piere set sail in a rowing boat with four companions to redeem the Holy Sepulchre.And they struck on a rock somewhere… (TGS: 18-19, my emphasis) Vague expressions tend particularly to occur when Dowell is quoting someone else, or offering an external authority.For instance: Someone has said that the death of a mouse from cancer is the whole sack of Rome by the Goths... (TGS: 11, my emphasis) So here, as somebody says somewhere, was a special case.
(TGS: 106, my emphasis) Dowell reinforces some of these implicatures with an explicit statement of his ignorance: And then the Ashburnhams were moved somewhere up towards a place or a district called Chitral.I am not good at geography of the Indian Empire.
(TGS: 119, my emphasis) [Nancy] determined that the rest of her life must be spent in acting as Leonora's handmaidensweeping, tending, embroidering, like some Deborah, some medieval saint -I am not unfortunately up in Catholic hagiologybut I know that she pictured herself as some personage… (TGS: 150, my emphasis) In these examples, the expectation on Dowell as narrator to be sufficiently informative is incompatible with his ability.Whether or not he reinforces the implicature with explicit statement, Dowell is implicating to the reader his own deficiencies as a narrator.

Opting out
Grice explains opting out as an explicit or tacit acknowledgement that a type of conversational behaviour required by the Cooperative Principle is not going to be forthcoming.Grice glosses this as an 'unwillingness' on the part of the speaker, but one of his examples is 'I cannot say more', suggesting that inability may be an alternative reason for opting out.Dowell opts out of his conversational obligations as a narrator throughout TGS, mainly because of his inability to supply information that might be considered necessary and relevant.Near the beginning of the novel, summarising the friendship between the Ashburnhams and the Dowells, he muses: And yet, if you come to think of it, isn't it a little odd that the physical rottenness of at least two pillars of our four-square house never presented itself to my mind as a menace to its security?It doesn't so present itself now though the two of them are actually dead.I don't know… I know nothingnothing in the worldof the hearts of men.
(TGS: 12) A narrator who comments on his own ignorance of the story he has undertaken to tell is opting out of the 'common purpose or set of purposes' (Grice, 1975: 26)  The definite noun phrase 'the girl' might be taken to indicate that the reader already has information about its referent.But in fact, this is the first mention of Nancy, and as such it is radically under-informative.Later, while describing Leonora's tight control on the couple's finances, Dowell mentions that at one point she threatened to take control of the bank account away from Edwards and adds: I guess that made him cut his throat.
(TGS: 132) The manner of Edward's death is introduced in a subordinate clause which, again, indicates that the reader might be expected already to have information about the event, but no information has been provided.
The requirements of the single maxim of Relation, 'be relevant' are very close to those of the second maxim of Quantity; to provide more information than is required is necessarily to provide information that is not relevant.Grice himself comments on the overlap, noting of the second maxim of Quantity that 'its effect will be secured' by the maxim of Relation (Grice, 1975: 27).Examples in which Dowell gives minute details about events and circumstances which are of only marginal significance to the story are violations of both maxims.On the first page he establishes that the friendship between the two couples began 9 years before the point at which he is telling the story; on the second page he gives the following information: When we all first met, Captain Ashburnham, home on sick leave from an India to which he was never to return, was thirty-three; Mrs. Ashburnham -Leonorawas thirty-one.I was thirty-six and poor Florence thirty.Thus today Florence would have been thirty-nine and Captain Ashburnham forty-two; whereas I am forty-five and Leonora forty.You will perceive therefore that our friendship has been a young-middle-aged affair... Grice suggests that the maxims of Manner could be subsumed under the general supermaxim 'Be perspicuous' (Grice, 1975: 27).The requirement to offer a clear and lucid account of events might seem to underpin any attempt at narrative, but Dowell is often less than perspicuous.This is difficult to illustrate using short quotations; it involves, for instance, the many complexities in the timeframe of the narrative, which jumps back and forth without apparent method and often without signposting.Dowell also varies the manner in which he refers to the main characters, for instance alternating between 'Edward' and 'Ashburnham' or 'Captain Ashburnham' and between 'Leonora' and 'Mrs Ashburnham', without apparently signalling any differences in his relationship or attitude to the characters in question.
Dowell's many violations of Quantity, Relation and Manner provide a context in which the reader must try to decide how to assess his relationship to Quality: in effect, whether to believe him or not.The second maxim of Quality requires speakers not to say anything for which they 'lack adequate evidence' (Grice, 1975: 27) Dowell at least appears to commit himself to this.He claims on a number of occasions that he will only narrate what he has witnessed or been told, and will not report on conversations in which he did not take part because 'I am not going to make up speeches' (TGS: 142).But despite this, he does represent many conversations at which he was not present.He also describes in detail the thoughts of characters to which he could not have had access.For instance, he says of Edward's attitude to Nancy: He thought that she only pretended to hate him in order to save her face and he thought that her quite atrocious telegram from Brindisi was only another attempt to do that.
(TGS: 162) Yet at the end of the novel Dowell comments that when Edward received the telegram from Brindisi he 'whispered something that I did not catch' (TGS: 169) and said no more about it before killing himself, so he could not have adequate evidence for Edward's thoughts.Similarly, at one point Dowell mentions that he does not know how Florence and Edward got rid of her earlier lover, Jimmy, but offers the elaborate speculation that: I fancy that fat and disreputable raven must have had his six golden front teeth knocked out by Edward one morning whist I had gone out to buy some flowers in the Rue de Paix… (TGS: 67) Later, he refers to a visit which Edward made to Florence and himself in December 1904, and suggests that: It must have been during this visit that he knocked Mr. Jimmy's teeth down his throat.
(TGS: 72) Dowell made it clear that the incident with the teeth was pure speculation on his part, but later he comments confidently on its likely timing; he is making a statement for which he lacks adequate evidence.
Dowell's readiness to violate the second maxim of Quality leaves open the possibility that he is also capable of violating the first maxim: of saying things which he believes to be false.That is, Dowell would appear to be capable of the least cooperative type of behaviour a narrator could engage in, that of lying to his readers.It is perhaps little wonder that the critics of TGS have sometimes been reluctant to take Dowell at his word, and have accused him of everything from complicity in his wife's infidelity, to double murder.

Flouting
Unlike the other types of maxim nonfulfillment, flouting is not uncooperative; the apparent breach of a maxim is a deliberate and ostentatious conversational move to facilitate the communication of an implicature.Dowell does not flout maxims often in TGS, and when he does he tends to do so in an unsubtle manner, using figures of speech.When he first properly introduces the character of Nancy Rufford, he explains that: She had lived with the Ashburnhams ever since she had been of the age of thirteen when her mother was said to have committed suicide owing to the brutalities of her father.Yes, it is a cheerful story… (TGS: 70) In saying that Nancy's story is 'cheerful' Dowell is, of course, flouting the first maxim of Quality and sarcastically implicating the opposite of what he is apparently saying.
As outlined in section 3.1 above in relation to Short's (1996) discussion of prose fiction, the 'narrator-narratee' discourse can be seen as subordinate to the 'author-reader' discourse, and it is possible that within this superordinate discourse Ford Madox Ford may be flouting the expectations of cooperation in order to communicate implicitly.Pratt argued that the apparently flawed narration of Tristram Shandy might be a vehicle for the author to implicate 'a critique of contemporary stylistic norms' (Pratt, 1977: 165-6).Many critics have detected Ford's commitment to literary impressionism in TGS.The precepts of this school of thought are not, of course, explicitly stated in the text.But it might be that, in flouting the rules of cooperative narration, Ford implicates his belief that, for instance, life does not narrate, and that literature should be concerned only with subjective experience.
However, Pratt also argues that an author may use a disordered narrative to implicate matters concerning the psychology of the narrator, orientating the reader in relation to the text.She claims that whenever the narrator fails to behave cooperatively 'the author implicates the cause of the failure' (Pratt, 1977: 182).But in TGS it does not seem possible to reach any definite conclusion about the reasons for Dowell's rambling and contradictory narrative.It is impossible to choose with certainly between the suggestions offered by critics because there is no definitive indication in the text.What is implicated by the author about the reasons for the unreliability of the narrator seems to display the property of 'indeterminacy', which Grice suggests is a feature of many implicatures (Grice, 1975: 40).The reader is aware that there are problems with the narration, and might expect that the effect of these will be to implicate information about the character and credibility of Dowell, but is unable to decide what is being implicated.In this respect, TGS seems closer in nature to Jaroslav Hašek's The Good Soldier Švejk and His Fortunes in the World War, as discussed by Steiner (1998), than to Laurence Sterne's Tristram Shandy, as discussed by Pratt.Steiner commented that readers of Hašek's novel are given no assistance in determining whether to understand the narratorial nonfulfillment of maxims as 'unintentional failures or as quiet, unostentatious violations intended to mislead' (Steiner, 1998: 493), which may contribute to the difficulty they encounter with the text.Tellingly, Hašek's novel is close in date to Ford's while Sterne's is from a much earlier period in literary history.This raises a possibility, which may be worth exploring further, that an increase in indeterminacy of implicature is characteristic of the relationship between author and reader in Modernist literature.

Conclusions
Analysing the notoriously problematic narratorial style of TGS using the framework of Grice's theory of cooperative communication suggests that it is characterised by a high preponderance of examples in which one or more of the conversational maxims are not fulfilled.These examples range across most of the varieties of non-fulfilment identified by Grice: facing a clash of maxims, opting out of the maxims and simply violating the maxims.They do not, however, include a significant number of instances in which maxims are flouted, which is consistent with Grice's observation that the flouting of maxims, although apparently in breach of the cooperative principle at the level of what is said, in fact represent observation of that principle at the level of what is implicated.The forms of behaviour characteristic of Dowell as a narrator are therefore those which could be described as 'uncooperative' in Gricean terms.
This article therefore supports the general claim made in a number of previous studies that unreliable literary narrators might be explained using Grice's theory of conversation.But it questions some of the more specific claims which have been made, particularly in foundational work on the topic.Pratt (1977) argued that uncooperative behaviour on the part of a narrator can function as flouting of the cooperative principle by the author, in order to communicate with the reader by implicature.She indicated that the author may communicate implicit messages about the nature and function of narration itself, and this does seem to be the case in TGS, where the reader may recover beliefs about the unavailability of an objective understanding of reality which are consistent with Ford's commitment to literary impressionism, but which are not made explicit in the text.However, Pratt predicted that in any case of uncooperative behaviour on the part of a narrator, there would be information available to the reader by means of implicature as to the extent of and the reasons for that behaviour.This does not seem to be the case in TGS, where it can be difficult for the reader to ascertain exactly why Dowell is breaking a maxim, including whether he is doing so intentionally or unintentionally.This may explain the wide variety of different interpretations that critics of TGS have offered concerning the character of Dowell and even the central events of the novel.It raises the further, intriguing, possibility that the indeterminacy of authorial implicatures may be a particular feature of Modernist literature.
This article has offered a pragmatic literary stylistic analysis of TGS, drawing on Grice's theory of conversation and motivated by the emphasis this places on communication as an essentially cooperative encounter.In order to understand what makes a narrator problematic or specifically unreliable, it is necessary to have a clear sense of what would be expected in normal, successful linguistic interactions, and Grice offers a model of this.As narrator of TGS, Dowell draws on but subverts usual expectations of cooperative behaviour.
of the exchange.In fact, the phrase 'I don't know', and numerous variants, are repeated throughout TGS.As just a few examples: I don't much understand about these matters.(TGS: 20) Upon my word I have forgotten.(TGS: 22) I may have got some of the names wrong.(TGS: 38) I don't remember how she went on.(TGS: 39) I cannot keep these things in my head.(TGS: 88) I have not the least idea of what an English officer's wife's existence may be like.(TGS: 99) I don't know that I have got the financial details exactly right.(TGS: 114) I do not know much about English legal procedure -I cannot, I mean, give technical details of how they tied him up.(TGS: 115) I don't know how it was arranged; I cannot imagine how even a blackmailer can make his demands.(TGS: 118) I am only an ageing American with very little knowledge of life.(TGS: 162)

(
TGS: 10)    Over half of this passage is unnecessary to the task of establishing the age of the protagonists.In other examples Dowell repeats himself more or less verbatim without apparent informative intent:It was a great worry … Yes it was a great worry.(TGS:21) He hadn't had the least idea of saying that to Leonora.… He hadn't had the least idea of framing that speech.(TGS: 120) She seemed to have got hold of the clue to the riddle.Yes, Leonora seemed to have got hold of the clue to the riddle.(TGS: 127)