An Examination of the Discursive Devices Employed in Fact Construction A Discourse Analysis of Ghost Stories

This is examined with particular reference to accounts that orientate around paranormal phenomena, and builds on previous research in this area Wooffitt, 1992: Wooffitt & Allistone, 2005

An Examination of the Discursive Devices Employed in Fact Construction:

A Discourse Analysis of Ghost Stories

Name – Timothy Reese

Supervisor – Dr Peter Lamont

AbstractThe current study examines the discursive strategies that are used when peopleattempt to establish an account as factual This is examined with particular reference

to accounts that orientate around paranormal phenomena, and builds on previous research in this area (Wooffitt, 1992: Wooffitt & Allistone, 2005: Lamont, 2007) by examining accounts of ostensible ghost encounters A discursive strategy of

‘disaffirming the stereotype’ is identified, whereby writers tend to emphasise either aspects of the surroundings, the ghost, or their emotional state, during their account, which are at odds with what can be thought of as a stereotypical ghost story A

stereotypical ghost story can be considered one whereby events occur during the night, where an anxious individual, primed by eerie surroundings, believes they have seen or heard a ghost, and subsequently becomes frightened It is argued that

disaffirming the stereotypical ghost story is designed to serve two functions; firstly it

is designed to disassociate the writer’s account from the easily debunked stereotype, and secondly it is designed to create a sense of saliency, which makes the account more memorable via the Von Restorff effect In addition methodological issues concerning discourse analysis and parapsychology are discussed

IntroductionDiscourse Analysis & Fact Construction

Discourse is an ambiguous term (van Dijk, 1997) but can loosely be defined as

a continuous stretch of language containing more than one sentence, which can occur

in either conversational or narrative form (Blackburn, 1994) In turn discourse

analysis (DA) can be defined as a method of qualitative research devoted to

understanding all forms of spoken interaction and written text (Colman, 2001) DA is qualitative in that it does not focus on numerical data, rather the data which it focuses

on is language itself Therefore in order to understand DA and its critical perspective,

it is first necessary to understand DA’s stance towards language According to Bhatia

et al (2008) there are two crucial factors that dissociate DA from other lines of

linguistic analysis Firstly DA emphasises that language can be analysed on the level

of the text, as well as at a structural level, i.e the broader semantic meaning of a text can be analysed, as well as finer grained details such as syntax and phonetics

Secondly DA emphasises that language ought to be considered as a tool for social action, and is not merely a passive referential medium through which we reflect the state of the world The origins of this view of language are perhaps best traced back toLudwig Wittgenstein’s influential ‘Philosophical Investigations’ where he stated that language can be conceived of as a series of games in which people are competing to construct different versions of reality, which ought to be beneficial to the speaker or writer in one way or another In other words Wittgenstein highlighted that language can be used to manipulate reality as opposed to merely reflect it At first this may not seem a particularly radical claim, as the potential of language to manipulate has long been known, the teaching of rhetoric for example, the art of using language to

persuade or influence others, can be traced back to medieval universities (Blackburn,

1994) However, DA argues that all discourse are socially orientated this way, and, as noted by Austin (1962), the distinction between discourses that ‘do things’, and discourses that merely ‘state things’ is redundant.

In its relatively brief history DA has been incorporated into a number of different disciplines, using different approaches, and has focused on different kinds of data The study of DA within psychology (discursive psychology) has built upon earlywork from conversation analysis (CA), and has focused on how mental states, social events, and material objects are established in discourse (Bhatia et al, 2008) CA is a field of study that is concerned with the norms that underlie the organization all forms

of verbal social interaction, ranging from everyday interactions between friends to quite ‘monologic’ forms of interaction (e.g speech making), and asserts that

interaction is both highly organised and orderly (Drew & Curl, 2008)

The present study focuses on the process of fact construction in discourse This has been examined by various researchers (e.g Gilbert & Mulkay, 1984; Potter

& Edwards, 1990; Smith, 1990; Wooffitt, 1992; Collins & Pinch, 1993; Lamont, 2007), and focuses on the discursive devices that people employ in their attempts to establish their version of a contentious issues as factual A discursive device can be defined as any phrase or feature of a discourse that is intended to influence the nature

of an interaction or account in favour of the speaker or writer In the process of fact construction discursive devices are used in the hope that they will help to establish thefactual nature of the speaker or writer’s account Potter (1996) provides a

comprehensive review of the discursive devices used in the process of fact

construction, and claims that there are two main strategies which can be employed to help establish the facticity of an account; firstly the speaker or writer needs to

establish that they are a reliable source of information, and secondly they need to

establish that the incident which the account revolves around occurred independently

of them Broadly speaking establishing that the speaker or writer is a reliable source

of information can be done in three ways; by emphasising that they are of sound mind, by emphasising that that they are an expert in the particular field at hand, or by establishing that they have no motivation to lie The first two factors can be

accomplished by the discursive device of category entitlement Category entitlement

is a strategy that plays upon the belief that certain categories of people are entitled to have more or less reliable knowledge about a certain issue than most people Anyone belonging to the category ‘psychologists’ for example, is by expected many people to have reliable knowledge about body language, and therefore in an account about bodylanguage it may be beneficial for the speaker or writer to align them self with the category psychologist Conversely there are also categories that speakers or writers should attempt to distance themselves from, for example ‘wishful-thinking’ or ‘not of sound mind’, as their accounts may be undermined by association with these

categories The third factor can be established using the discursive device of stake inoculation The concept of stake inoculation is perhaps best understood by

considering what Edwards & Potter (1992) refer to as the ‘dilemma of stake’, the dilemma being that any account can be undermined as a product of the speaker or writer’s stake or self interest Stake inoculation negates this criticism by explicitly stating that the account is not driven by such motivations, for example, “the

prosecution witnesses, many of them passers-by with no conceivable axe to grind, were articulate and plausible” (Potter, 1996: p.110) In the preceding example the emphasis of the witnesses as having “no conceivable axe to grind” is designed to highlight the objectivity of the witnesses and their accounts, emphasising that they arenot products of stake

The second broad strategy that the producer of an account may employ to establish the factual nature of an account is to establish that the events which the account orientates around occurred independently of the speaker or writer Potter (1996) identified three strategies for achieving this; empiricist discourse,

corroboration, and vivid detail Empiricist discourse makes use of what has come to

be known as empiricist repertoire This is a style of discourse first identified by Gilbert & Mulkay (1984) in their study of scientific disputes, which minimises the involvement of the producer of the account and instead treats data as the primary actor This is largely accomplished through a grammatical strategy, whereby verbs that are usually applied to human action are in fact applied to the data, for example,

“the data implies that…” Corroboration is a very persuasive strategy as it suggests that an account is not simply a lie or a figment of the producer’s imagination This does not rule out the possibility of collusion however, but the probability of collusion

is usually considered highly unlikely, and therefore the probability of a corroborated account being untrue is far less likely than an uncorroborated account The use of vivid detail in accounts is intended to prove that the producer was present, as only a genuine observer will have been able to capture the particulars of a scene

So far the nature of DA and its role in fact construction have been discussed However in order to appreciate the significance of the phrase ‘fact construction’, it is necessary to understand DA’s social constructionist stance towards reality, as facts arestatements that ostensibly reflect reality

Discourse & Reality

“When Mr N N dies one says that the bearer of the name dies, not that the meaning dies” (Wittgenstein, 1953: p.20) This quote eloquently draws attention to what Potter (1996) refers to as the ‘interactional space’ between discourse and reality

Reality is an ambiguous term, but can simply be defined as “that which there is” (Blackburn, 1994: p.320) However the reason that reality is an ambiguous term is that there is disagreement as to what ‘is’ and ‘is not’, for example whether or not numbers or colours should be included is disputed (Blackburn, 1994) Parker (1992) makes a useful distinction between three different levels of reality, or as he phrases them object statuses He named these; ontological object status, epistemological object status, and moral/political object status Ontological status refers to an object’s physiology and may be considered the commonsensical notion of reality, e.g the physical entity that we might use the word ‘chair’ to refer to Epistemological status refers to the meaningful status that an object acquires through discourse, e.g the status that an object acquires after it is denoted as belonging to the category chair Lastly moral/political status refers to conceptual objects that have epistemological status but no ontological status, and therefore can be called into being, e.g attitudes orintelligence.

DA can be considered a methodology which falls under the wider ideological sphere of social constructionism Burr (1995) claimed that it is impossible to define social constructionism per se, but that any epistemological approach which adopts one

or more of the following characteristics at its basis could be considered social

constructionism; a critical stance towards taken-for-granted knowledge, the view that the way we understand the world is culturally and historically specific, the view that knowledge is sustained by social processes, and that descriptions (constructions of theworld) encourage some social actions, and discourage others “The major thrust of social constructionism is the claim that human beings, and all the other ‘things’ consciously present to us, are socially, and discursively produced” (Burr, 1995: p.89) This is well illustrated by an example first brought to light by Ferdinand de Saussure

(Potter, 1996) De Saussure claimed that objects derive their meaning from

comparisons with other similar objects, rather than reflecting a natural order of the world For example he cited the English word ‘river’ as deriving its meaning from contrasts with alternatives such as ‘stream’ Here the distinction is clearly a size basedone However in French a different distinction is made between a ‘fleuve’, which flows into the sea, and a ‘rivière’, which flows into a lake or river The point is that the concepts of ‘river’ and ‘stream’ are not a naturally occurring phenomena waiting

to be realised and reflected in language In fact these different concepts (‘river’,

‘stream, ‘fleuve’ and ‘rivière’) were constructed, and are maintained by different societies on the basis of the differing comparisons that are made during verbal social interactions, these objects therefore are discursively produced

It is important to emphasise that when social constructionists claim that objects are discursively produced and maintained, they are not claiming that an object’s ontological status is socially produced, but that it’s epistemological and moral/political statuses are In the past social constructionism has been criticised (e.g Nightingale & Cromby, 1999) as people have wrongly ascribed it as an ontological theory However as Edley (2001) notes “it is actually very difficult to find occasions where social constructionists have explicitly denied the existence of an extra-

discursive realm….it is an opinion that others (i.e realists) have attributed to them.” (p.437)

In conclusion DA views Parker’s (1992) epistemological and moral/political realms of reality as socially constructed and maintained phenomena This has

significant implications for the issue of fact construction as facts are presented within these realms and consequentially must be constructions themselves Therefore it is the

role of DA to identify the discursive devices that are used in the construction,

maintenance and deconstruction, of ostensible facts

Relevance of DA to Mainstream Psychology

The most significant factor concerning DA and mainstream psychology is that

DA challenges the assumptions that underpin cognitivism Cognitivism can be defined

as a positivist theoretical approach, which claims that all aspects of the mind and behaviour can theoretically become fully explained through observation and

experimentation According to cognitivism mental activity is the process of acquiring and processing information, performed by discrete mental entities, and is normally conceptualised in terms of rules and representations, defined in symbolic or

propositional form (Still & Costall, 1991) Behaviour is seen by cognitivism as the output of these information processing systems In the latter half of the twentieth century cognitivism replaced behaviourism as the dominant influence in mainstream psychology, and remains a highly influential theory for a large amount of current research Therefore any theory that stands to undermine cognitivism has potentially detrimental consequences for much contemporary psychological research In a review

by Edwards and Potter (1992) both cognitivism and psychology’s reliance on

cognitivism as an all explaining theory were heavily criticized (cited Willig, 2001) These critiques can be summarised into five unfounded assumptions that cognitivism makes, which will now be examined in turn:

i) Firstly cognitivism assumes that talk about people’s beliefs and attitudes can

be used as a direct route to their cognitions However, as discussed in the first section

of the introduction, talk cannot be considered to be an accurate reflection of what people are actually thinking, as socially orientated

ii) Cognitivism also assumes that cognitions are based on perceptions, i.e that cognitions are representations of real objects that occur in the world However DA argues that objects themselves are constructed, at least on an epistemological level through language Consequentially rather than objects giving rise to cognitions, DA argues that it is discourse and social interactions that give rise to objects as we are capable of thinking of them, hence language itself should be the focal point of

research

iii) Thirdly cognitivism assumes that an objective perception of reality, upon which cognitions are based, is theoretically possible, and that resource saving

heuristics account for the differences between people’s perceptions However if, as

DA states, language constructs reality rather than reflecting it, there cannot possibly

be an objective perception of reality as each individual constructs their own unique version of reality

iv) Furthermore cognitivism assumes that the events and objects, which peopleform their attitudes towards, are one and the same However, DA states that all events and objects are relative to the individual Therefore differences in expressed attitude towards what is ontologically the same object or event can be attributed to the fact that people are forming attitudes towards epistemologically different objects or events, rather than the difference being due to fundamental differences between the individuals For example the defendant’s perception of a court case will vary greatly from the victim’s perception of the same case

v) Finally, cognitivism assumes that our cognitions are relatively enduring, stable, and therefore predictable However DA’s interpretation of language highlights the importance of the social context in which accounts are given DA argues that the same person may well express significantly different attitudes about the same issue

when in a different social situation Therefore all we can really infer from accounts is what people are doing with their words, rather than inferring information about their cognitions.

These criticisms of cognitivism challenge the validity of conclusions drawn from experiments which have been conducted within a cognitivist framework, which accounts for a large amount of psychological research past and present Due to these criticisms DA should be seen as a critique of mainstream psychology, rather than just

an alternative methodology, as advocated by Willig (2001) This point is iterated by Still & Costall (1991), whose main criticism of cognitivism is that it ignores the fact that human behaviour takes place with and within a complex and dynamic

environment This leads to what Still & Costall refer to as ‘the frame problem’, whereby after a rule has been developed to explain a particular behaviour, it is

unlikely that this rule will apply in all situations, and therefore it is difficult to know when the rule should apply and when it should not

In conclusion if we are to accept the main tenets of DA and social

constructionism, which seem difficult to reject, we are forced to question the validity

of cognitivism as a research method This can be attributed to two main factors Firstly DA conceptualizes language in a radically different way to cognitivism, and consequently challenges cognitivism’s reliance on language as a way of inferring people’s cognitions about ostensibly the same objects and events Secondly DA places

an emphasis on ethnomethodological concerns, which severely distort an individual’s behaviour Cognitivism almost entirely ignores such concerns, and conceptualises humans as disconnected units Therefore the relevance of DA to psychology is that it brings into question the validity of cognitivism, and consequently criticises

psychology’s reliance on this theoretical approach

Relevance of DA to Parapsychology

Parapsychology can simply be defined as the study of psi phenomena

(Colman, 2001) The term psi denotes anomalous processes of information or energy transfer that cannot be explained by known physical or biological mechanisms (Bem

& Honorton, 1994) Psi can be split into two categories; extra-sensory perception (ESP) and psychokinesis (PK) ESP can be defined as the experience of, or response

to, stimuli without conventional sensory contact This can be divided into three subcategories of; clairvoyance, telepathy, and precognition Clairvoyance is the ability

to see things beyond the limits of orthodox sight (e.g through walls), telepathy is the ability to perceive another person’s mental processes, and precognition is the ability toperceive objects or events in the future On the other hand PK can be defined as the movement or change of physical objects as a result of mental processes, i.e

influencing the result of a dice roll using only the mind Parapsychology had its beginnings in psychical research at the end of the nineteenth century (Utts, 1991) Psychical research concerned itself with anecdotal reports of ostensibly paranormal phenomena In contrast parapsychology has attempted to model itself on the natural sciences, and hence has conducted experiments in laboratories, involving thousands oftrials with ordinary subjects, the results from which are analysed using statistical techniques (Wooffitt & Allistone, 2005) The origins of this method of investigation into the paranormal are usually attributed to J.B Rhine and his associates at Duke University who developed forced choice experiments in order to test for the existence

of psi In particular they examined ESP with the use of Zener cards (a deck of 25 cards with 5 each of 5 different symbols) whereby participants had to either guess the order of a randomly shuffled pack of cards (testing clairvoyance), or infer the identity

of a card from another person’s mind (testing telepathy), or guess the order before the

pack was shuffled (testing precognition) The results obtained from these experiments were positive, i.e participants were guessing the correct cards more often than chancewould suggest, and were published in Rhine’s 1934 book Extra-Sensory Perception However largely due to the controversial nature of these results, in that they defied conventional scientific knowledge, they were ignored by most scientists, and

attributed to either statistical error, coincidence or deception (Inglis, 1992)

Since parapsychology’s beginnings at Duke University alternative techniques have been developed to test for the existence of psi, and in particular ESP The most prominent example is the ganzfeld technique which came about after a surge in interest into altered states of consciousness during the 1960’s (Schmeidler, 1994) In atypical ganzfeld a receiver, while lying in a comfortable chair, and in a soundproof room, wears earphones that play white noise, and goggles that only let unpatterned light enter the eyes These conditions are intended to create a homogenous sensory state, which is alleged to enhance psychic functioning (Storm & Ertel, 2001)

Meanwhile a sender in a separate room is shown either a static or dynamic visual target which they are instructed to attempt to ‘send’ telepathically to the receiver During this period the receiver is asked to speak all thought aloud After this session either the receiver or a judge, who has heard the receiver’s articulated thoughts, is asked to pick the target from four possibilities In accordance with previous work, these experiments have found above chance hit rates At a meeting of the

Parapsychological Association in 1982 the significance of these results was discussed,which lead to a lack of accordance between the members, most notably Ray Hyman and Charles Honorton (Utts, 1991) After both publishing separate papers on the issue the pair later collaborated (Hyman & Honorton, 1986), agreeing that there was an overall unexplained effect, but disagreed as to whether this constituted evidence for

the existence of psi In this paper they also outlined criteria for which future ganzfeld studies should meet, for example strict security precautions against sensory leakage Despite these improvements subsequent meta-analyses of ganzfeld studies have continued to produce positive results (e.g Ben & Honorton, 1994: Storm & Ertel, 2001)

However despite consistent positive results from early Zener card tests, and ganzfeld studies, as well as PK experiments (e.g Radin & Ferrari, 1991) there has stillbeen a lack of widespread acceptance of the validity of these results and

consequentially parapsychology’s scientific status (Allison, 1979) Wooffitt &

Allistone (2005) argue that this is perhaps not surprising as controversial knowledge

is unlikely to be accepted on the back of single experiments, as sociological studies have shown

An alternative line of study within parapsychology has been to look at the differences between believers and non-believers in paranormal phenomena These can

be referred to as sheep-goat studies, whereby the sheep are believers and goats are non-believers In these studies the existence of psi is not at issue, and the fact that so many people believe in these phenomena is seen as justification enough to warrant investigation (Irwin, 1993) Typically this line of research has sought to account for people’s belief in the paranormal in terms of some form of deficit, and over the years belief has been linked to; childhood trauma (Ross & Joshi, 1992), fantasy proneness (Irwin, 1991), social marginality (Bainbridge, 1978: cited Irwin, 1993), inability to critically analyse (Gray & Mill, 1990), and probability misjudgement (Musch & Ehrenberg, 2002) to name but a few However results from this line of analyses have been largely inconclusive This has frequently been attributed to the fact that there is not a satisfactory way of measuring people’s level of paranormal belief (Wiseman &

Watt, 2006), largely stemming from the fact that it is difficult to decide what

constitutes paranormal, as opposed to mysterious or unexplained (Coelho, 2005: cited Lamont, 2007)

As is hopefully apparent traditional research methods in parapsychology are struggling to be productive Positive results in favour of the existence of psi are not being accepted due to their controversial nature (Wooffitt & Allistone, 2005), and sheep-goat studies are failing to find consistent results due to the inherent difficulties

in defining and measuring paranormal belief Therefore an alternative method of enquiry could be particularly beneficial for parapsychology Wooffitt & Allistone (2005) argue that DA can be a very productive line of enquiry for parapsychology They argue that not only can DA shed light on the verbal descriptions offered in ganzfeld studies and in accounts of spontaneous experiences, but also that this work could be helpful for qualitative research on conscious experience This line of

research, which aims to supplement quantitative research on consciousness, focuses

on vocal reports of what consciousness is like Therefore in addition to providing insights within the parapsychological field, DA could allow parapsychology to

establish common ground with the study of consciousness, as well as discursive psychology This may have the benefit of creating greater acceptance of the scientific status of parapsychology, and it’s as yet unexplained positive findings into the

existence of psi

Previous DA’s of Paranormal Encounters

Previous DA’s that have orientated around ostensibly paranormal experiences have been relatively few Much of the work in this area has come from Robin Wooffittwho has built upon previous work by Harvey Sacks (Potter, 1996) Sacks examined the way that people reported extraordinary experiences (i.e hijackings, assassinations,

plane crashes) The theoretical link between these circumstances is that both accounts

of extraordinary events and paranormal events should be designed to resist

undermining, as the speaker or writer anticipates that the account’s factual status will

be contested These types of accounts can be undermined by three basic strategies; identifying the speaker as not of sound mind, by identifying motivation for the

speaker to lie, or by identifying a lack of corroboration for the speaker’s account

Sacks (1984) identified a discursive device regularly used to resist the

undermining accusation of the speaker not being of sound mind, namely the “I was just doing X, when Y” structure Here X stands for a very mundane activity (i.e doingthe washing up) and Y refers to the exceptional event (i.e a plane crash) Wooffitt (1992) demonstrated that the “I was just doing X, when Y” structure is frequently used in accounts of paranormal experiences and He argued that this structure helps the speaker to establish a sense of their own normality, by emphasising the normality

of their behaviour In the same publication Wooffitt also demonstrated that active voicing, the inclusion of quoted speech, is used to help construct the factual nature of paranormal accounts by creating a sense of corroboration For example, “and they said ‘Oh, that wasn’t too strange an experience’ because they had heard it before” (Wooffitt, 1992: p.158) Attributing the quote to a vague group of people, as opposed

to one particular individual, can be seen as a form of systematic vagueness, a

discursive device whereby the use of detail, which might otherwise be examined to reveal contradictions and confusions in the account, is avoided Therefore the use of the pronoun ‘they’ in the above example, is not only intended to create a sense of corroboration between multiple parties, but it also attempts to negate specific

questions which might undermine the source of the quote

In a more recent study by Lamont (2007), the significance of avowals of prior scepticism was discussed The avowal of prior scepticism could be said to take the format, “at first I was a sceptic, but then Y”, where Y denotes a paranormal

phenomenon This can be seen as analogous to the structure “at first I thought X, but then I realised Y” first officially identified by Gail Jefferson, again in relation to exceptional rather than paranormal events However Jefferson stresses that the matter

of authorship is problematic due to the significant contributions of Harvey Sacks (Jefferson, 2004: p.164) In this structure X denotes a common explanation (i.e a car backfiring to explain a load bang) and Y is the extraordinary explanation (i.e the load bang was due to gunfire) Avowals of prior scepticism can be interpreted in two ways Firstly the avowal can be interpreted as an example of what Potter phrases stake inoculation This is a discursive device applied when the speaker anticipates that their account will be undermined as the product of expectation or peer-group pressure The explicit statements of initial scepticism implicates that the facts are so strong that theyovercame the speaker’s initial scepticism and expectations (Potter, 1996), and allow the events to be presented as factual Secondly the avowal can be interpreted as an example of category entitlement, whereby the speaker or writer attempts to align themselves with the category ‘sceptic’ The significance of this is that scepticism is the socially expected stance towards paranormal phenomena and entitlement to the category of ‘sceptic’ indicates that the individual is a critical thinker and of sound mind (Lamont, 2007)

To conclude, there have been a limited number of DA’s of accounts of

ostensibly paranormal phenomena However those conducted thus far have identified three major strategies; the “at first I thought X, but then I realised Y” structure gives the speaker the category entitlement of ‘sane person’, active voicing allows the

speaker to create a sense of corroboration, and an avowal of prior scepticism serves tocounters the criticism of stake, as well as enabling the speaker or writer to align themselves with the category ‘sceptic’.

The Current StudyThe Current Data

The current data are a collection of accounts of ostensible encounters with ghosts, compiled and edited by Vivienne Rae-Ellis (1990) The accounts come from British citizens, who responded to an appeal from Rae-Ellis in the late 1980’s for people to send her details of any ghost encounters which they may have had The appeal was driven by advertisments published in various newspapers and journals The nature of these encounters varies markedly, and includes; apparitions, ghosts heard, ghosts felt, ghosts smelt, as well as poltergeists, and cases of possession The accounts appear in their original form unless it has proved necessary to make

clarifications, avoid repetitions or shorten the text Edited parts of the accounts are punctuated

The current study therefore focuses on accounts which do not occur during talk interactions Wooffitt (2005) claims that, ‘talk-in-interaction’ present more for theresearcher to study than accounts independent of interaction Wooffitt argues that CA studies offer the researcher the opportunity to examine the sophisticated and intricate practises that occur during conversation, and that the study of independent accounts,

“invites the analyst to disattend to the interactional frameworks in which such

discourses, etc, are claimed to be present” (Wooffitt, 2005: p.179) However, it can also be argued that interactions that occur during conversation are so complex that conclusions drawn from them may be undermined by unknown variables Goodwin