“It is hard to mesh all this”: Invoking attitude, persona and argument organisation

Abstract

This paper focuses on the nature of invoked attitude as part of the Appraisal framework under Systemic Functional Linguistics, offering an amended set of diagnostic tools for tracking the ways that attitude may be invoked or implied in written discourse. Under the appraisal framework, categories of attitude are located in a system network, suggesting that there are a set of features that can determine whether wordings belong to one or another category. However, in the many cases where such attitudes are invoked, rather than being directly activated through the use of highly stable lexical items, such categorisation becomes problematic. Categorisation of attitude is dependent almost entirely on discourse semantic features of text and on interpretive probabilities, even in the case of explicit or ‘inscribed’ Attitudes, since the value (negative or positive) of many lexical items can be ‘flipped’ in the co-text in which such wordings appear. This paper presents what is termed a Spectrum of potential invoked attitudinal activators, arguing that these textual devices and resources provide a more fine-grained approach to analyses where invoked attitudes have been identified in texts. The paper argues that intertextuality, in its broadest sense, needs to be taken into account when making claims that attitudinal values have been identified in texts. As such, the paper presents a somewhat different theoretical approach to that offered by Martin and White (2005). The proposals offered are illustrated with excerpts from a study based on a close analysis of a corpus of texts of written interaction comprising over 80,000 running words, texts in which invoked attitude was more frequent than inscribed attitudes.

Introduction

This paper is concerned with the nature of resources for activating evaluative meanings in discourse, and specifically with language which indirectly or implicitly activates positively or negatively attitudinal assessments, what is referred to in the appraisal literature as ‘invoked’ attitudes. These concerns are related in turn to the study of identity in text–termed here ‘textual persona’, with a similar approach having been previously applied to the study of identity in online written conversations (see for example Don 2007, 2008, 2011). The notions of textual identity, and textual persona are inter-related but refer to different perspectives on text, discussed further below. This approach to the study of textual identity and its relations with textual persona employs the appraisal framework for analysing writers’ use of evaluative language, and their means for adopting stances towards others, objects, and propositions. The appraisal framework, based in Systemic Functional Linguistics (White 1997, Martin 2000, White 2002, Martin & Rose 2003, Martin & White 2005, interalia), identifies three types or categories of evaluative meaning making: Attitude (meanings related to positive/negative assessments or responses), Engagement (dialogistic positioning of the writer/speaker in relations with other voices, viewpoints and potential respondents), and Graduation (adjustments of the force of utterances or of the boundaries of semantic categories).

The purpose of this paper is thus to offer a theoretical discussion, arguing that 1) all evaluation is inevitably intertextual in nature, and that 2) intertextual references position both reader and writer as ‘identities’. Because invoked attitude figures so prominently in the analysis of textual identity, the paper proposes an amended version of what has been proposed by Martin and White (2005: 67) in the form of a set of overlapping 'categories' designed to fill gaps in the interpretive paradigm, and to call for a more precise means of tracking and identifying how attitude is invoked in texts. For this purpose, the paper provides examples of previous analyses of attitude as illustration of how this might work in practice.

Rather than proposing a taxonomy of attitudinal invocation which is significantly different from that of Martin and White, the following discussion proposes that analysts attend to a wider array of discursive features and ‘strategies’ than has previously been taken into account. While the challenges facing analysts using the appraisal framework and the subjective nature of much of the categorisation have been addressed by, for example Thompson (2014), Macken-Horarik & Isaac (2014), Fuoli (in press), and Hood & Martin (2007), the ‘cline of invocation’ itself has not been substantially reviewed, and the ways in which such implications of attitude also work to call on the assumed knowledge of addressees is yet to be given detailed attention.

The paper begins by outlining the challenges and issues that face appraisal analysts when using the framework to look at evaluation that is implied rather than directly inscribed. This is followed by a brief outline of the proposed ‘spectrum’ of invocations, after which the use of invoked attitude is linked to notions of textual identity and persona. The subsequent section extends the discussion of the spectrum of invocations, which is then followed by a series of examples drawn from an online discussion in order to illustrate the types of textual mechanisms offered in the paper, showing how they work to invoke both attitude and textual persona.

Attitude, invocation and associations

Within the Appraisal framework, indirectly attitudinal meanings are termed ‘invocations’ or ‘tokens’ of attitude and are said to ‘invoke’ positive or negative assessments and responses (e.g. see Martin and White 2005: 65). On the other hand, wordings which are typically viewed as explicitly conveying positive or negative assessments – i.e. lexis which has a largely stable attitudinal value across different contexts of use – are termed “inscribed” attitude in the appraisal framework (for example, lexical items such as beautiful/ugly, honestly/dishonestly, love/hate). The indirectly attitudinal meanings, or ‘invocations’ pose significant challenges for analysts employing the Appraisal framework. They often raise questions of interpretation and categorisation both with respect to the type of attitude being activated, and with respect to the target of the attitudinal assessment–i.e. who/what the reader is being positioned to view positively or negatively. Once again, although some of these issues have been attended to previously (see for example Don 2007, Hommerberg & Don 2015, Hood and Martin 2007, Hood 2006, Thompson 2014, Macken-Horarik & Isaac 2014), and apart from that offered in Hood & Martin (2007) issues regarding the means by which such attitudes are implicated in texts and in their social contexts have not been addressed to any great extent. Accordingly, this paper focuses on clarifying how discourse analysts might deal with such invocations of attitude systematically and in a theoretically-principled manner.

While invocations of attitude typically involve some signalling or flagging of an attitudinal value in the co-textual environment (Martin & White 2005: 66, Hood & Martin 2007) categorisation under the attitude framework is challenging on account of the fact that meanings are inevitably social, and thus also rely on intertextual references and shared assumptions from outside the text. One purpose of this paper is to argue that instances of attitudinal stance in discourse are dependent on ‘associations’ attaching to phrases and other linguistic signs due to the way these signs have been used, and are typically used, in other texts. These associations and the attitudinal positions they invoke are thus ultimately a function of intertextuality, in the broadest sense of the term–i.e. in the sense which underlies Bakhtin’s much quoted dictum that ‘Each utterance is filled with echoes and reverberations of other utterances to which it is related by the communality of the sphere of speech communication’ (Bakhtin 1986: 91). Intertextuality is, of course, a complex notion, related in the literature to a number of different objects of study, but ultimately it is intertextuality, or ‘transtextuality’ to adopt Genette's (1997) more technical term–which is central to the concerns of this paper and the analysis of invoked attitude.

Instances and readings

While all types of evaluation are of interest here (c.f. Hunston & Thompson 2000, Ch1), my primary focus is on the attitude framework, and specifically on how wordings which are implicitly rather than explicitly attitudinal may be interpreted during a ‘reading’ of a text.¹ Reference to a ‘reading’ implies that texts activate potentially different meanings dependent on the context, including the social practices or cultural frames that actual readers bring to each reading event. The challenges this brings to analysis in terms of reliability and replicability are now being raised in the literature (Fuoli in press, Macken-Horarik & Isaac 2014). Reference to the reading event also implies that appraisal analysis is primarily concerned with the instance (rather than delicacy c.f. Halliday & Matthiessen 1999: 15). With reference to the instance, Martin & White (2005: 164) provide a ‘Cline of instantiation-evaluation’, the endpoint of which they label ‘reaction (reading)’ which they gloss as:

The take-up of evaluative meanings in a text according to the listener/reader’s subjectively determined reading position; the attitudinal positions activated by the reader as a result of their interaction with the text (2005: 164).

This means that analysts often need to account for their own readings of attitudinal categories when conducting research of evaluative stances, and thus the concerns of the present paper centre on how categories of attitude can be more precisely justified. In addressing the need for precision under these conditions, Fuoli (in press) proposes a step-wise approach to any appraisal analysis, noting that:

Classifying evaluative expressions into the categories provided by the model is also a difficult and subjective task. In many cases, multiple interpretations for textual items are possible and the boundaries between the categories are not always clear-cut. (in press: 2)

The appraisal framework provides options from its taxonomy of attitudinal sub-types involving the analyst in firstly determining whether the attitudinal value being referenced is an instance of Affect (positive or negative emotional responses), Judgement (positive/negative assessments of human behaviour and character by reference to social norms) or Appreciation (positive/negative assessments of entities, processes and situation in terms of their aesthetic properties or social value). Each of these broad categories is associated in turn with its own more delicate taxonomy of sub-types (i.e. sub-types of Judgement and so on) for a particular span of text. When the span of text in question indirectly ‘invokes’ an attitudinal value, I argue that it is important that analysts do not depend entirely on intuition when deciding on a classification according to the attitude framework, and that they avoid offering ad hoc criteria by which a particular classification has been made. Similar issues regarding the actual targets of attitude occur in texts where the targets of attitudinal invocations – i.e. whom or what is being evaluated – are not mentioned in the co-text at all. Since some of the recognition criteria for appraisal categories involve the nature of the target being identified (e.g. whether the target is ‘conscious’, or ‘human behaviour’, or ‘artefact’, or proposition, etc.), this poses further challenges for analysts claiming theoretical robustness for the framework as a whole, and for their own interpretation of the specific texts under analysis using the framework.

A proposed amended cline: a spectrum or ‘array’ of textual strategies

As a way to address these and related problems – i.e. to provide a more systematic means for accounting for the interpretation of evaluative stance and attitudinal typing in discourse – this paper offers a proposal for enhancement of the account of invocation offered by Martin and White (2005: 67). Martin and White propose that invocations can be broadly divided into three types: (1) those where the attitudinal stance is activated by unevaluated experiential tokens (termed ‘afforded’ attitude), (2) those where it is signalled by evaluative meanings such as intensification (graduation) or counter-expectancy (termed ‘flagged’ attitude), and (3) those where lexical metaphor activates the positive or negative assessment (termed ‘provoked’ attitude). While the set of features offered here follows Martin and White in recognising these as key strategies or textual arrangements by which attitudinal stance may be indirectly indicated or put into play, it is important to recognise that other mechanisms or discursive strategies may also play a role when a span of text has the potential to invoke an attitudinal value. The following discussion proposes that analysts attend to a wider array of discursive features and ‘strategies’ than has previously been taken into account, justifying decisions as to invoked attitude categories. At the same time, a similar ‘cline’ of attitudinal invocation as proposed by Martin and White is the basis for the discussion, which argues for several amendments.

The proposed array includes discursive strategies² which may not of themselves activate an attitudinal meaning but which may nevertheless play a role in conjunction with other signals of attitude. In what I term an invocation spectrum, these various discursive strategies are matched below (Fig. 1) with a slightly modified version of Martin and White’s taxonomy of invocation (afforded, flagged and provoked attitude), in order to show the original gross groupings of resources determined to have invoked attitude in the corpus of texts. The aim was to provide a more finely-grained account of how attitudinal values may be typically and potentially activated indirectly in discourse. A key purpose of this paper, then, is to set out this spectrum of invocation, with a view to providing the means by which analyses of attitudinal invocation may be made more explicit. The present paper offers a revised version (Fig. 2) of this earlier proposal, which will be discussed and exemplified in more detail below.

Fig. 1

Mapping of M&W 2005 with early version of spectrum of invocations

Attitude and textual persona

A secondary purpose of the paper is to observe that the deployment of certain styles of invoked Attitudes can be related to the textual persona of the text’s voice, since the stance (s) adopted by that persona is dependent on what the writer (s) assumes is or is not already known or acknowledged by interlocutors and other audience members alike. Thus, instances of invoked attitude have a key role in construing a textual identity, relying as they do on authorial assumptions about the knowledge, beliefs, values and expectations of the addressee (s) and readers. Here I make a distinction between ‘textual identity’ as the overall patterning or voice of the text itself, which takes into account all metafunctional elements, experiential, interpersonal and textual, and is closely related to style, while ‘textual persona’ is a projection of who the writer might be in terms of values and co-positioning, taking into account stances towards objects, ideas and others – something which is more focussed on the tenor relations being set up in the text – and very dependent on indicators of engagement, as well as attitude, in the process.

The data from which examples below have been taken consisted of highly interactive and argumentative texts contributed to an electronic discussion forum during the late 1990s and early 2000s. Contributors to this online discussion were not previously known to each other, and so group norms and the matter of identity and status were always under negotiation. In such a context, the advancing of attitudinal positions was always a delicate matter, and much of the positioning of writers and their audience entailed the use of implication and what Wigboldus et al. (1999) in their study of group dynamics termed ‘low linguistic abstraction’ (what the appraisal framework would term ‘invoked’ attitude). More specifically they found that, when describing the desirable or positively assessed behaviour of out-group members, speakers would consistently use what they termed ‘low linguistic abstraction’, but ‘high linguistic abstraction’ (‘inscribed’ attitude) when negatively evaluating their behaviour. The reverse was true for the description of in-group members’ behaviour: their desirable behaviour was realised by high linguistic abstraction (inscribed attitude), while their undesirable behaviour was described by means of low linguistic abstraction (invoked attitude). Thus, what Wigboldus et al. termed ‘low linguistic abstraction’ and what the appraisal literature terms ‘invoked attitude’ was strongly associated with the positive assessment of people outside the group and with the negative assessment of people in the group. My own studies (Don 2007) confirm the findings of Wigboldus et al, at least with respect to also indicating a strong correlation between the use of invoked attitude and negative assessments of in-group members–i.e. discussion group members were quite often critical of other group members but strongly preferred to use attitudinal invocation when doing so. As a consequence this data provided a rich source of this type of attitude.

As indicated previously, invoked attitude refers to the ways in which tokens of attitude- that is, implications of attitudinal stances as opposed to relatively more explicitly attitudinal acts- can be made potential for readers in any text. Such invocation is related to the ‘construction’ of textual persona as an effect of discourse and co-text, and is thus not concerned necessarily with a ‘real’ or individual personal identity, but with the ways that writers /speakers employ the resources of meaning-making in their texts. In so doing writers seem to either expect readers/listeners to supply associations which activate attitudinal stances, or to miss such implications altogether. From another perspective, such textual persona may be enacted by means of projecting such shared assumptions on the readership. This is not to discount the real-life experiences of any actor in enacting a persona, but it is the social semiosis of the discourse and the resources employed for any such performance which is the focus here. In this sense then, textual persona can be traced as both an effect of the organisation of an overall argument or the use of persuasive strategies in general (in particular, the co-positioning of writers and addressees) - as well as an effect of the implied associations carried by the phenomena which are attitudinally evaluated (i.e. targets of attitude) – perhaps across multiple texts. Thus, one issue addressed by looking at the means for invoking attitudes is related to how these rhetorical strategies are also implicated in the construction of textual persona (see for example Bucholtz & Hall 2005, Don 2011). The importance of invoked rather than explicit attitude in projecting persona through association will become evident during the course of the discussion, but in general I argue that it is via the use of resources of attitude that a speaker/writer’s social background and value system becomes visible.

As already noted, the analysis of invoked Attitude can be problematic not only with respect to decisions as to the type or sub-type of Attitude involved, but also with respect to identification of the target of the attitudinal assessment. For example, one instance of potential invocation may be ambiguous as to exactly who or what is being evaluated, or there may be grounds for seeing a single attitudinal expression as having multiple targets or sub-types. The argument I advance here is that a writer’s choice as to whom or what s/he targets for attitudinal assessment can be just as significant regarding the nature of the textual persona being constructed as the type/sub-type of attitude being advanced. In other words, what one chooses to discuss, and to evaluate, is indicative of one’s concerns and social values, even in texts where evaluative language is under-represented. Thus, issues with respect both to type of attitude and the target (s) of that attitude are relevant to explorations of the textual performance of persona. This also has implications regarding audience ‘projection’, since the positioning of self in any one textual act, is at the same time a positioning of audience members, thus acting to construe their personae as well. Recognition of shared assumptions, shared knowledge, and reliance on intertextual reference is also an essential component of the implication of solidarity or a high degree of Affiliation (Don 2011, Knight 2010) in such instances, and so the prevalence of invoked attitude in the texts used in the study was not surprising. In other words, the nature of online communication in the corpus on which the original study was based brings with it a heightened sense of social space, group allegiances, and the need to maintain virtual ‘face’. Of course, it is not only textual allusions and verbal intertextual associations which rely on the recognition of affiliations and allegiances. Other graphic resources such as colour, typography, and layout itself are also deployed in social media sites to manage identity as well (c.f. Kress and van Leeuwen 2002, van Leeuwen 2005, Djonov & van Leeuwen 2012), invoking attitudes towards certain images and ideas in less precise ways than made available through verbal language.

Approach and sources of examples

The data for the study from which this spectrum was drawn was comprised of a corpus of around 80,000 words of chronologically-ordered, unfolding instances of written texts. These were manually annotated for attitude using the UAM systemic coder (O’Donnell 2002, 4.5) and selected for inclusion on the basis of their contribution to an ongoing ‘thread’ or discussion. This means that the unit of analysis needs to be considered from the perspective of whole text, as meanings accumulate or build up over the duration of the ‘text-time’, as well as extra-textually over the course of several contributions (or ‘posts’) to the online conversation. For this reason, online discussions provide a much more convenient source of data on continued negotiation and attitudinal positioning on related topics over longer periods of time than has been previously available in other dialogic modes such as conversation. At the same time, short segments and phases of these texts will be used here to illustrate some of the issues attending analysis of invoked attitude in longer argumentative texts–whose attitudes and evaluative stance may need to be accounted for by reference to their location in longer arguments–sometimes consisting of several contributions made to discussions and occasionally stretching over several months. This can be considered as akin to what Halliday has termed the phylogenetic process by which languages (and their meaning making potential) develop over time within a particular community (e.g. Halliday and Matthiessen 1999: 17–18, Halliday & Matthiessen 2014: 68). In this case, a study of the development of meaning-making potential is within the circumscribed community of the email list, a group within which ‘intertextual’ references to other conversations within the same group may only be (easily) grasped by those same group members.

It has already been noted that each reading of a text is regarded as a new ‘response’ to the potential meanings involved, and highlights the fact that the interpretation or ‘reading’ of attitudes in any text changes slightly as new information/knowledge comes to hand, and is dependent therefore on the ontogenetic position-in-time of the reader/ addressee/ audience members–i.e. each person’s individual experience of meaning making as this develops over time through multiple, cumulative communicative experiences. In terms of the reading of the whole text, its linearity, or the logogenesis of meanings, is at once also dependent on what was above attributed to the process of phylogenenis, a process pertaining to wider as well as more local cultural ‘reservoirs’ of meaning built up in languages by its users over time. In this regard, ontogenesis also plays a part, especially when individual ‘repertoires’ of meaning making and interpretation (c.f. Bernstein 1996, Martin 2010) are not privy to the experiential and interpersonal histories and knowledge demanded by the text for its reception. This applies especially to any reading of attitude (i.e. evaluation of a target) in any texts. One obvious example of how individual repertoires affect the interpretation of attitude occurs regularly in movie theatres where the film may be subtitled for the local population, but where the only person laughing at the dialogue is the native speaker: the translation may be entirely adequate, but the nuances and intertextual references of the original language and its cultural assumptions become ‘lost in translation’.

The spectrum of invocations

The spectrum or ‘array’ of resources associated with attitudinal invocation presented below sets out some of the linguistic arrangements which may invoke attitudes in text, or may be associated with that invocation. A number of differences should be noted between this and Martin and White’s presentation of their taxonomy invocation subtypes in their Language of Evaluation monograph (c.f. 2005: 67), Firstly, for reasons which will become clear below, the proposed spectrum of invocation operates with a simpler and flatter taxonomy than that presented by Martin and White. Specifically, following White (1998) and Don (2007, 2008) it recognises only two broad sub-types of invoked attitude, given the labels ‘evoked’ and ‘provoked’ attitude, with the ‘evoked’ category coterminous with Martin and White’s ‘afford’ sub-type (i.e. what they term ‘experiential’ tokens) and the ‘provoked’ category combining Martin and White’s ‘flag’ and ‘provoke’ sub-types. This means that prominence is given to the distinction between invocations which rely entirely on assumed (e.g. cultural, intertextual) knowledge or values for the attitudinal inference to arise (i.e. ‘evoked’) and those which, in addition to relying on assumed cultural knowledge, also involve local co-textual signals or other in-text indicators that an attitudinal value is at stake (i.e. ‘provoked’ in the current taxonomy, combining ‘flagged’ and ‘provoked’ in Martin and White 2005). This two-way taxonomy gives prominence to this particular distinction (experiential meanings/cultural framing, versus textual signalling), a distinction which is obscured to some degree in Martin and White. In addition, while they render their taxonomy of invocations strategies as a system, a system network was not viewed as an appropriate means of illustrating a set of strategies or discursive elements that combine in many cases to invoke an attitude for the reader. Hence what is proposed is named here a ‘spectrum’, intimating that the lines between these strategies are blurry, and do not of themselves define an attitudinal value. Similar arguments are advanced by Macken-Horarik & Isaac (2014: 81) when they observe that “…this kind of appraisal [i.e. to do with radiating prosodies and cultural frames] however, resists enclosure in analytical boxes and frustrates the ‘either-or’ distinctions that are central to the system network.” Their work sets out similar concerns to those argued here, discussing how invoked attitudes can be viewed as operating along a ‘cline of implicitness’ (Macken-Horarik & Isaac 2014: 89), where different environments of appraisal, from the level of the word, through wording, to phase, text pattern and culture, act to constrain choices from both text-creating and reading perspectives. In this paper, the perspective taken is rather of an array or ‘spectrum’ of resources of interpretation, in order to highlight the fact that the model focuses on mechanisms for invoking attitudes rather than levels along a cline of instantiation.

The spectrum presented here was originally devised in 2007, and was mapped onto elements of the cline proposed in Martin and White (2005: 67). This original model was shown in the previous Fig. 1, in order to show how the two perspectives are related. This diagram has since been updated and expanded rendering as Fig. 2 below. The following Fig. 2 shows the amended and revised proposed spectrum with several annotations and simplifications. Previous literature on the nature of evaluative mechanisms that are most closely related to some of these areas of invocation are also noted in the following diagram.

Fig. 2

The spectrum of invocation – evoke versus provoke

Figure 2 above has been designed to indicate a primary differentiation among mechanisms of invocation according to whether they are (1) dependent on high affiliation in terms of assumed knowledge and shared experience of the world (‘evoke’) or (2) they are more reliant on elements of the text itself and the explicit lexico-grammatical means for signalling an evaluative stance towards the current proposition. So that, what is considered as acting to ‘Flag’ an Attitude in Martin and White is subsumed under the first division of ‘Provoke’, as shown in Fig. 2 above. This is because such flagging may be easily identified in the co-text whereas ‘Afford’ type invocations (for which I use the label ‘Evoke) instead rely primarily (although not exclusively) on assumptions or projections of shared knowledge between interlocutors.

Further, with respect to differences between my model and that of Martin and White, a wider repertoire of ‘signals’ of invoked attitude have been included in this account, although many of those noted here are subsumed within engagement signals under their scheme. Perhaps most notably, contra Martin and White 2005, various forms of attribution have been included (items numbered 4 and 5 in Fig. 2 above), along with what the appraisal framework terms ‘Denial’ (negation), assessments of counter-expectedness, the logico-semantic categories of contrast/comparison and causality and certain intra-textual reference relations (where, for example an experiential token is linked cataphorically or anaphorically with lexis elsewhere in this text which is attitudinally inscribing -c.f. also ‘propagation’ Hood 2006). It should be noted that the numbered items in the figure are not presented as sub-types in a taxonomy of invoked attitude (hence the use of curved brackets to avoid confusion with the type of bracketing used in system networks) but rather are offered as some of the key mechanisms and textual features by which attitude may be indirectly ‘evoked’ or ‘provoked’. They are arranged from top to bottom in a loose array connoting that those at the ‘top’ are closer to the explicit end, whereas those at the ‘bottom’ rely much more on implication.

Intertextuality

It should also be noted that my proposed spectrum differs from Martin and White’s model in the attention it pays to the role of intertextuality - both narrowly and broadly defined – in the invocation of attitude. A ‘narrow’ intertextuality relates closely to what Fairclough (1992) refers to as ‘manifest’ intertexuality, that is, when it is obvious that a wording or phrase is almost directly lifted or quoted from another recognisable source. On the other hand, a more broadly defined intertextuality is inherent in all language via phylogenetic processes that imbue wordings and even generic structures with associations that activate meanings for readers and speakers of the language. Lemke (1985: 283) refers to the difference between these dimensions of intertextuality as distinctions between “relatively global rather than local patternings of language use” (see also Thibault 1989, Bakhtin 1986, Genette 1997).

With reference to Fig. 2 above, intertextuality of one form or another is obviously involved in items numbered 3 through 5, but is also most fundamental to items 1 and 2 (material referencing cultural doxa and in-group allusions). Due to the mode of communication from which the data for the original study was obtained – computer-mediated, member-subscription discussion list – the reliance on ‘membership’, and the comprehension of intertextual reference on which invocations depend becomes salient in interpretation. The unit of analysis here needed to be considered as one contribution to the list: a whole message sent or ‘posted’ to the group discussion in chronological sequence, since meanings in general and attitude in particular were often dependent on logogenetic development of an argument (text-time) as well as their chronological ordering as response (‘real’-time). Such posts include technological information, or ‘meta-data’. So that, in its original context, the term ‘post’ labels the ‘utterance’, including all its meta-data and time-sequenced reading, while ‘text’ refers to any re-presentation of the posted matter for analysis, and this is usually stripped of its meta-data: information that other members would use as what (Genette 1997) refers to as paratext. These extra elements of intertextuality, while not available in this analysis, nevertheless need to be mentioned as forming a distinctive ‘layer’ of information for those actual persons who originally made contributions to the discussion online (Don 2007). And, while whole posts/texts need to be broken down into smaller stages and sub-units so that positioning strategies and attitudinal stances of writers may be analysed, at the same time, the original studies (Don 2007, 2008) revealed such sub-units also appear to be marked by changes and shifts in the attitudinal positioning–which, in turn, is the focus of this type of analysis.

Thus intertextuality inevitably has a role to play in accounting for the possible interpretations of each unit of text as an instance of a wider set of potential meanings: all text-units are part of a larger set of texts, either as a class of similar texts, or as part of a chain of on-going textual events. This Bakhtinian perspective informs much of the literature in appraisal (c.f. Martin & White 2005: 92, Don 2007, Tann 2011). This also means that analysts need to take the role of ethnographer and the reading position of participant-observer, since such knowledge may be either intensely local, or generalised within the language and its users. It is intertextuality of this type which is implicated when an experiential token references some attitudinally-loaded cultural frame (Fig. 2, item 2, termed ‘cultural doxa’ in the spectrum set out there) or via an in-group allusion (item 1 in Fig. 2) which then gives rise to an attitudinal inference (c.f. White 2006). Thus the experiential token has this attitudinal potential only when it is associated with some socially recognised value in prior texts, texts which circulate either generally in the culture, or more narrowly in the communications of the in-group.

As well, as indicated by items 3, 4, and 5 in Fig. 2 above, other types of intertextual effects are also implicated in the invocation of attitude. As already outlined, for the labels for items numbered 3 and 4 (manifest and constitutive intertextuality), I rely on Fairclough (1992) who proposes ‘manifest’ and ‘constitutive’ intertextuality as sub-types of the broader phenomenon of ‘inter-discursivity’. The differences are significant in terms of the mechanisms by which attitudinal inferences may be triggered in a text. “Manifest intertextuality” applies when specific other texts are drawn upon in a text, typically via explicit quotation in the form of direct or indirect speech. Here the quoting or citing of material from another text has attitudinal potential when, for example, the quoted source is an authoritative one or is favoured in some way (what Tann 2010: 5.2 terms a ‘scripture’) or when it is assumed the quoted beliefs or observations will inevitably be regarded as untoward in some way (speakers ‘damned out of their own mouths’). ‘Constitutive intertextuality’, on the other hand, employs a type of generic re-constitution, in which a discourse type is constituted through a combination of elements of ‘orders of discourse’, serving to imitate a style, but to keep hidden explicit pointers to any originary text (s). Appraisal analysis by itself is unable to cope with this type of intertextuality, and none of the groupings on the proposed spectrum of invocations proposed above are, on their own, able to capture such generic imitation in the service of attitude. This type of intertextuality (Genette 1997: 4) defines as architextuality, the most “abstract and most implicit of all”. It is also noteworthy that his definitions of such types of his broader ‘transtextuality’ occur in the course of his explanation of the mechanisms of parody. At the extreme end of the spectrum of invocations, parody may be identified as one of the means by which writers (speakers) can effect or imply attitudinal positioning. In turn, the recognition of any text as parody is dependent on readers having knowledge of the register of the text being parodied.

Other theorists within the SFL tradition have approached the matter of constitutive intertextuality, that is, the expectations afforded by texts mirroring to some degree some aspect of other similar texts, by reference to ‘abstract thematic formations’ (e.g. Lemke 1985, 1995, 2002) or ‘semantic-thematic formations’ (Thibault 1989, 1991, 2002). Thibault (1989: 190) for example, argues that “two or more texts, which are in no way structurally related to each other, may operate or participate in a shared system of thematic-semantic relations”. In this way, attitudes become potentially available in dialogue through participants’ recognition of such relationships.

Here I am proposing that signals located towards the ‘top’ of the spectrum represented in Fig. 2 above should be understood as more directly activating attitudinal meanings - that is, the role of the author in positioning the reader to supply an attitudinal inference is more obvious, and its valeur generally available to the broader language community – and those towards the ‘base’ of the figure as less directly activating an attitudinal assessment, to a narrower audience, with intertextuality and the use of parodic cross-thematic formations representing the most extreme end of the spectrum and thus being more open to interpretation. Potential readings of irony also figure in the less explicit end of the spectrum, for example, Alba-Juez and Attardo (2014: 93) argue that “the concept of evaluation is crucially attached to that of verbal irony”, and other theorists such as Louw (1993), Clift (1999), and Partington (2007) have all argued for the relationship between use of irony and potential evaluative stances. Most of these mechanisms will be raised and exemplified in the forthcoming discussion, but space prevents a detailed exposition of the mechanisms of irony here.

It is perhaps necessary to stress again that it is not proposed that all the mechanisms included in the above figure are necessarily able, of themselves, to invoke attitude without other accompanying signals in the local co-text. As will be seen in the examples presented below, it is typical for a number of these types of invocatory mechanisms to reinforce or help activate each other in any segment of text to provoke readings of attitudes. Thus these divisions need to be regarded as groupings of elements or resources observed to occur in the service of invoking attitudes in texts, and not as definitive bases for any particular attitudinal value.

Discussion and exemplification

In this section, I turn to a discussion of the way in which the resources outlined in Fig. 2 above work as invocations by reference to text extracts drawn from the dataset of contributions to the discussion list which was described above. Example 1 below serves to illustrate some of these groupings of resources (mechanisms or ‘strategies’), and the challenges which face the analyst attempting to account for the evaluative stance invoked in any one section of a longer piece. The example is excerpted from a lengthy post to the discussion (originally 34 sentences), in which each orthographically signalled paragraph of the text makes separate arguments that are nevertheless interrelated, and thus many of the attitudinal meanings of the piece are dependent on their location in logogenetic text-time. In this first excerpt, and those that follow, it is possible to observe what Geoff Thompson (2014) described as the ‘Russian doll’ effect in analysing spans of wordings invoking an attitude – which in turn is embedded in, or fused with, longer stretches of text to invoke further attitudinal values. In Don (2007) this type of over-layering of attitudes was referred to as 'attitudinal density'.

Invocations and attitudinal density

The following excerpt appeared in sentence 5, and its purpose early in the piece was to set out the basis for the arguments to follow – basically a complaint by the poster, who was new to the group, that she hadn’t been appropriately ‘welcomed’ as a new member and that her posts were not being suitably responded to by established group members. In the following extract she intimates that the other group members have responded to her, as a new group member, with ‘fear’, ‘suspicion’ and ‘doubt’–which have nevertheless not been openly voiced.

Example 1

[..I] would rather hear your fear, suspicion or doubt directly than to hear their echoes in all of our exchanges or in the poverty of our exchange. [sally4] ³

What first strikes the appraisal analyst here is the way in which the attitudinal density (Don 2007: 177) of the excerpt (in which attitudes are nominalised/presumed rather than argued) functions as a means to deploy these as targets of the attitude at the clause rank. This ‘strategy’ operates to position readers as having no argument with the assessment, and allows the assessment to be assessed itself. For example, the first target of attitude [Affect: Inclination positive] I would rather + [[target] hear directly] + [target] also functions to introduce an ‘observed’ or non-authorial Affect: your fear, suspicion or doubt. The ‘fear, suspicion or doubt’ is, in some sense, ‘irrealis’: it is imputed to the generalised addressee, the readers of the piece, rather than observed as having occurred or being associated with any specified emoter. The attitudes of fear, suspicion and doubt held on the part of these addressees was analysed as instantiating Affect: Security: negative, and it is framed not only by Affect: Inclination (would rather hear (these things) directly), but also as part of a comparative construction: rather [+verbal group + noun group] than [+noun group]. The comparative helps to underwrite the negative values implied, and relates to the resources summarised as item (6d) in the array of invocations outlined above. If the imputation of ‘fear, suspicion and doubt’ were not already able to invoke a negative Judgement: Propriety towards the holders of these emotions (c.f. item (7) in Fig. 2), then the comparative co-text helps to pin it down, or ‘activate’ it.

The writer implies that such attitudes would then be rendered as ‘echoes’, or realised in ‘the poverty of our [future] exchange [s]’. It is this location in a comparative construction which stabilises the invocation of attitude towards a ‘real’ target, the addressees, via strategies of attitude tokens (item (7), Fig. 2) and local signals such as comparison (item (6d) Fig. 2), without explicitly mentioning those addressees, except in the deictic possessive, your. The addressees remain present only as potential holders of fear, suspicion and doubt, as potential dissemblers of their feelings, and as potential poor interlocutors. The local (i.e. within the text span) targets, however – those boldened in the excerpt above – remain at the level of implied possessions of the ‘real’ target: your fear, suspicion or doubt (to which the writer expresses positive Inclination about hearing directly), and the poverty of our exchange (to which the writer expresses negative Inclination about hearing their echoes).

The threat of negative Judgement

It will be noted that the noun group ‘the poverty of our exchange’ which functions as target for authorial Affect, also works to potentially negatively judge interlocutors of the future, via negative Appreciation: Valuation (item (7), Fig. 2). The token of negative Judgement: Propriety in this clause complex is read as provoked by this final value of Appreciation the poverty our exchange, and this interpretation depends in turn on the realisation of the target as not, grammatically-speaking, human behaviour, but as a nominalisation: our exchange. The negative evaluation is thus realised as a nominalised possession of a possible future condition, rather than an evaluation of any actual 'exchange' – such strategies can be seen to constitute a type of veiled threat or warning regarding behaviour, and are thus related to control of social boundaries and an attempt at the formation of group norms. In turn, this relates directly to the notion of identity, stance-taking and the realisation of textual persona, since such norms tell participants what they are allowed to do or say. According to Fairclough (2003: 41, citing Bernstein 1996) the ‘norms’ of interaction can be seen as “a moral order [which] are oriented to and interpreted differently by different social actors, and these differences are negotiated” (c.f. Don 2007, 2011). Thus, one strategy for regulating behaviour may be effected through the use of invoking potential negative judgement. In terms of what Halliday’s semantic network model (1973, reproduced by Hasan 2001) offers for rhetorical strategies of this kind, this appears to function as what he terms a ‘warning’. Under this model, he provides options used in disciplining a child, and given that one of the recurrent tropes of the text involves the notion of ‘family’, this suggests the potential for the activation of a context of control, the register on which Halliday based the semantic network model referred to here. Hasan (1996: 114) notes that Bernstein’s theory considered this register as critical to the process of socialisation. In this sense then, the writer of this text could be seen as invoking attitude as a way of exerting control over the group by the threat of negative evaluation, or what Hasan (2001) describes as “condition implicit”.

It is this threat of negative evaluation that points to the implied attitudes here, and relates to those contexts in which [modulation: obligation] is commonly found. That is, in contexts where directives are made through the use of declaratives featuring modal finites such as should, ought, had better, must, etc., the implication is often that failure to carry out what should be done will result in negative Judgement: Propriety being levelled. The invocations in these cases are irrealis–merely threatened.

The next section exemplifies ‘tokens’ at the less explicit end of the spectrum and hence focuses on the more intertextually-reliant ways of invoking Attitude.

Evoking (Affording) attitude and assumed cultural values

At the ‘bottom’ end of the spectrum (see Fig. 2 above), the groupings distinguished for Evoking (Affording) Attitude have fuzzy or overlapping boundaries, and indeed it is more appropriate to consider them, not as separate ‘categories’, but instead as attempts to describe resources for evoking attitudes at another level of delicacy. In many instances of text segments where attitudes are invoked – or interpreted as being invoked – different degrees of intertextuality are involved: (1) entirely local meanings or in-group allusions may be involved, as well as (2) the use of such grammatical resources as nominalisation or presupposition (c.f. also Example 1 above), in addition to (3) more generalised reference to cultural mores, quotations of other texts, places, or people, i.e. ‘associations’ in order to imply evaluative stances.

Example 2

Let ’em have the WTC , and the Space Needle . The Pentagon by all means, I wouldn’t mind lending a hand. But if they touch the Chrysler building …. [gen02.2/rob]

Example 2 above constitutes the whole of the body of a responding text, in which attitude is invoked at several levels. Firstly, the only arguably explicit attitude is realised by an instance of ‘generic behaviour’: to lend a hand, which is generally appraised positively as Judgement: Propriety (as ethically good) for those who lend a hand – perhaps, again arguably invoked via positive Affect towards that behaviour: I wouldn’t mind ~. In this case, the writer ‘wouldn’t mind lending a hand’ towards the target (s) [letting them have] the WTC, and the Space Needle [and] The Pentagon.

At this level, a reader not familiar with the context might interpret this as a positive evaluation of the three buildings mentioned, but when it is understood to have been posted in October of 2001, and that WTC is an abbreviation of the World Trade Center, then their appraisal in terms of negative Appreciation: Valuation (i.e. the buildings assessed as having low social value) becomes more apparent. Of course, the phrase let ‘em have is so intertextually charged with an attitude of not caring about the targets of the having, that the counter (but if they touch..) is almost expected. This negative attitude towards the three named buildings is thus underscored by the counter-expectation which follows, which at the same time acts to positively evaluate the target the Chrysler Building. Here we have a discourse structure of [directive + counter + [hypothetical-real]] (Hoey 1991, 1996, 2013, Winter 1994) which sets up the evaluative positioning and hence the attitude towards the target buildings:

“let them [verbal] + but + [if they [verbal] [-then …]]”.

The ‘real’ in this case, i.e. the consequence of the hypothetical situation, is here left as an implied threat of reprisal. This again relates to item (6d) in the spectrum of invocations set out in Fig. 2, where rhetorical patterns can be used as resources for implying an attitude, as well as item (3) where so-called constitutive intertextuality, or what (Lemke 1995) refers to as compositional practices within thematic formations, allow comparison-contrast and evaluative stances to be fore-fronted.

Note also the relevance of Graduation in the segment (item (6a) in Fig. 2) – the upscaling of ‘them having’ to ‘them touching’ the target building is also expected in this matching-relation: counter structure.

What is of interest, however, is the way in which intertextuality figures as part of accounting for attitude in this segment. For this text, any invocation of attitude is dependent on our membership of the group of people who recognise the reference to the attack on the World Trade Center and the Pentagon–or “9/11” as Americans refer to it. The writer also groups these two buildings together with another, “the Space Needle”, and through such clustering or text-level association, implies that these buildings are all similarly flawed and no loss to society. For the device to work, the reader needs to also recognise and negatively appreciate the Space Needle, and at the same time, recognise too the implication of an off-the-cuff, careless remark which is signalled by the representation of casual speech: e.g. the use of ‘em instead of them, and the trailing dots at the end. It also depends for some of its punch on what Maton (2014) describes as the ‘axiological charge’, i.e. the cultural-level associations (item (2) in Fig. 2) accruing to the two respective groups of buildings–something that does not work if the reader is not familiar with the Chrysler Building, and does not accept its status as iconic of the (positively appraised) architecture of New York City.

In summary, this short piece depends on a variety of invocations, the most obvious being intertextuality i.e. reference to specific pieces of American architecture which the writer expects addressees to be familiar with, as well as allusion to the recent historical events after which this post was originally written. At the same time, such references do not work on their own so to speak, and the text is also ‘flagged’ by its being framed in a matching relation (item (6d), Fig. 2) (see Hoey 1991, 1996, 2013, Don 2008) of counter expectation (6c) and other local signals of preference acting to signal an attitude: let ‘em have, I wouldn’t mind, but if as well as the use of Graduation (item (6a), Fig. 2), in the contrast between ‘letting them have’ the dismissed buildings versus the implied raised stakes if ‘they touch’ the Chrysler Building.

Associations of assumed cultural values

The following excerpt is similarly dependent on a range of intertextual references and assumptions regarding the values attaching to them. The excerpt originally appeared at the end of a longer post, and marked a shift in both ideational content and interpersonal orientation. In this extract, the poster reports on a recent trip to Seattle in the US and in particular what he presents as a negative experience of some coffee shops there. Again, one of the assumptions underlying the attitude implied relies on a clustering of values–in this instance the values are associated with certain musical styles, where Frank Sinatra and the Three Tenors are associated, and then contrasted with “cutting-edge rock”.

Example 3

BTW these Seattle coffee shops seem to favor as background music Frank Sinatra and ‘The Three Tenors Sing Broadway Show Tunes’ -- where’s the cutting-edge rock to be found?[sft11.5/matt]

Context and local references are salient here: the fact that the music is being played in Seattle coffee shops in the mid-nineties is significant for the reading of negative attitude toward these coffee shops via counter-expectation (item (6c), Fig. 2), which is not explicitly signalled but is rather an effect of comparison, or contrast. Again, strategies of direct reference to assumed shared knowledge are used–whether this is part of the wider lexicon in the form of knowledge of Frank Sinatra (item (2), Fig. 2: cultural doxa), or dependent more on local references privy only to those addressees who know what style of music Seattle was famous for at that time, and also in fact, actual experience of “Seattle” coffee shops whose background music is recognisable in this description, i.e. Starbucks (item (1) in Fig. 2: in-group allusion). The actual attitudinal trigger is crafted as a rhetorical question incorporating an implied comparison (item (6c), Fig. 2) teamed with an explicit token of Attitude in the form of lexical metaphor, ‘cutting edge’: Where’s the cutting edge rock to be found? The question implies that cutting edge rock was expected, but was not evident in ‘these Seattle coffee shops’. Instead, it is implied that the playlist included music which is considered passé. Because ‘cutting edge’ rock inscribes a positive Appreciation: Valuation of that target, it also implies by contrast that the music of Frank Sinatra et al. are evaluated negatively, and that the target of the evaluation is ultimately those Seattle coffee shops who do not provide the positively valorised musical material.

In this excerpt then, Seattle coffee shops (and likely Starbucks in particular) are the primary target of negative Appreciation via counter-expectation, a typical engagement strategy. However, this stance cannot be located in any one element of the phase. Instead, counter-expectation is activated through a number of interwoven elements. The same applies with respect to an analysis of the ultimate (or ‘real’) attitudinal target of this extract. We note that nowhere in this extract are Seattle coffee shops explicitly targeted for evaluation. Nowhere are they explicitly “appreciated”, for example. It’s only as a result of the interweaving of these multiple evaluations, and specifically of multiple invocations, that the coffee shops–and ‘Starbucks’ specifically–emerge as this ultimate target.

To take the analysis one step further, such interpretations of writer attitude toward specific targets, and targets associated with particular clusters of ideas and objects imbued with axiological (either/or ‘epistemological’) charges (Maton 2014), also highlights both the writer’s association with particular ideological stances, and the writer’s positioning of their addressees. The expectation that Seattle coffee shops should be playing cutting edge rock shows that the writer also expects his addressees to be aware of Seattle’s status in terms of musical genre, and that these addressees will also find it amusing or ironic to hear that Frank Sinatra and the Three Tenors are instead in favour there. The audience members are thus positioned via association as cognoscenti, or in-group members in this case, and as favouring the cutting edge rock that the writer also favours–or, alternatively as not in favour of Starbucks for this reason.

Provoking (Flagging) Attitude

Strategies by which writers frame utterances in such a way as to overtly signal they hold a particular attitudinal stance, but nevertheless refrain from anything directly attitudinal are covered in Fig. 2 by items (5) through to (8).

Ventriloquy (item 5) involves the writer purporting to quote or cite an external source but doing so in such a way that suggests he/she is ‘putting words in the source’s mouth’–i.e. actually misrepresenting them in some way, or at least offering an exaggerated version of what they might say or believe. Graduation will often be involved in signalling this exaggeration. The purpose of such ventriloquism is typically to cast the source and/or their purported viewpoint in a negative light, although it is possibly that it might also be used to position the reader to view the source positively.

Figure 2, items (6a) through to (6e) (e.g. Graduation, Denial, Counter-expectancy, logico-semantic relations of contrast/comparison) operate as alerts that an attitudinal value is at stake, but without explicitly evaluative lexis being used.

Item (6e) (intra-textual reference/linkage) refers to mechanisms by which an experiential token (i.e. material which, of itself, would not position the reader to apply a positive or negative assessment) activates an attitudinal value via linkage with material elsewhere in the text which is more directly or explicitly attitudinal, what Hood (2006, after Lemke 1998) describes as ‘propagating values’. This mechanism involves ‘encapsulation’ (Sinclair 1993), a strategy which relies on co-associative reference in the same text, where accumulation of intra-textual meanings and references via what Martin (1992) calls ‘identification’ resources are brought together via ‘ideational chaining’ (Don 2007). These strategies which are a product of logogenesis, are referred to as instances of attitudinal ‘propagation’ (Lemke 1998) or ‘semantic prosody’ (Hood 2006). Other means of flagging Attitude may be more straightforward and local–as seen earlier, counter-expectational signals, negation, comparison and parallelism may be variously employed in the co-text to set up attitudinal readings.

Alerting readers to stance

Consider example 4 below which was posted from the sidelines, so to speak, in response to a series of argumentative posts in which one poster (A) claimed the other (B) was being ‘snotty’, to which his addressee (B) responded by calling it ‘bullshit’. This contribution engendered the following comment from another participant (C):

Example 4

<tongue in cheek> I proclaim TRIPLE BULLSHIT!! I challenge you to a snottiness dual, mon ami, name your weapon, time, place.... go ahead, make my day </tongue in cheek> [wvn60.23/TK]

My own reading of attitude here depends on both the appearance of the post in its chronological context, and its indicators of exaggerated ire. For me, the writer implies that the recent behaviour of the poster (s) (you, mon ami) is to be negatively judged, as foolish and as worthy of poking fun at. The negative Judgement: Propriety of the target, B’s (an addressee) behaviour could create further conflict if it were not also couched in humorous terms, and thus the use of strategies of invocation, primarily those of engagement and graduation (item 6) act at the same time to soften the negative Judgement implied. As previously pointed out, such ‘softening’ or use of resources for implying negative assessment, indicates that the target is an in-group member.

The comment in Example 4 is explicitly framed with the words tongue in cheek, a lexical metaphor which also functions as a type of engagement, a matter of indicating dialogic position with respect to the audience: the writer declares that what s/he is saying here is not serious. The intertextual parody activated within this segment is also framed in turn and realised by paratextual elements (Genette 1997: 3) in the use of <angle brackets > which denote html tags being opened and closed. In this way the addressees are positioned as knowledgeable of internet contexts: the use of html tags can be considered to carry (adapted from Maton 2014) an ‘epistemological charge’, by reference to a coding language not everyone might recognise at that time, hence also acting to imply affiliation through knowledge – bearing in mind that this specific electronically-mediated discussion was conducted entirely in ASCII using a text-only delivery. This rhetorical/graphical device deploys strong ‘semantic density’, that is, an example of a high degree of what Maton (2014: 153 ff) calls ‘epistemological condensation’, in the service of construing a relationship of legitimate knower in the field, and thus as someone who is a member of that group of cognoscenti who are knowledgeable about internet protocols. Its use here also acts to co-position addressees as members of this group, which in turn acts to defuse any offence which might have been caused. Here of course, discussion on the nature of evaluative stance and its relationship to invoking attitude shades into the realm of the interpersonal in general. The use of intertextual references in this excerpt support the explicit evaluative lexis used, and point to the fact that engagement devices, rather than specifically attitudinal lexis, operate in these texts to invoke evaluative stances towards not only people and objects, but also propositions (c.f. Hunston &Thompson 2000, Ch1).

There are other indicators in this short text that the writer is playing: exaggeratedly challenging the addressees to a duel, using graduation, both focus and force, and explicit Engagement: Proclaim. Noting the use of these devices is useful in illustrating how not only lexico-grammatical elements are deployed in indicating attitude (or, in this case, stance towards the addressee (s)), but also how paratextual indices (</>) act to flag stance as well.

Within this segment there is also, however, an example of a type of manifest intertextuality (item (4) Fig. 2), with the citing of a famous movie line (boldened above), which acts to underscore the overall negative assessment of the other players (A and B) in this interaction. Their actions are associated in this way with those of the larger-than-life character of Dirty Harry, a character who blasts his way out of trouble. The point here would be that, unless the reader had seen this movie, or at the very least had spent time in twentieth century western society, this reference may not make the same impact on the reader.

In-group allusions activating assessment

In the case of the following excerpt (Example 5) – which appeared at the end of a post in which the writer had crafted a limerick aimed at another poster, and in which he also alluded to his own identity as a biker which list-members would know (item (1): in-group allusion) – the intertextual reference (item (4), Fig. 2) was originally lost on me, since I was unfamiliar with the limerick which begins, There was a man from Nantucket, depending, for its implications of masculinity, on wording which rhymes with ‘Nantucket’:

Example 5

Biker T-shirt: “I AM the man from Nantucket.” [tvs228.56/stan33]

However, the explicit indication of extra-vocalisation through attribution to a “biker T-shirt”, the inverted commas (item (4), Fig. 2), and the Graduation: Force in the form of capitalised AM (item 6a), all signalled that whoever the man from Nantucket was, he was admired by bikers, and that the writer was associating himself with this highly valorised identity. The invoked attitude towards the ‘real’ target, the writer himself, was therefore interpreted as one of positive Judgement: Capacity, through intra-textual association as well as manifest interdiscursivity (Fairclough 1992). My argument in this case is that had I not been privy to the ongoing conversation which preceded this text, or had not known that the writer had previously identified himself as a bike enthusiast, the intertextual reference would not have been able to invoke the positive attitude towards the ‘real’ target for this analyst.

Attitudinal Tokens

At the end of the invocation spectrum, and excepting the liminal role that lexical metaphors play, attitude may be provoked by other attitudinal values, as has been discussed by Thompson (2014). Commonly, this occurs with values of Affect or Appreciation acting as tokens of Judgement (item (7), Fig. 2). As indicated, of course, lexical metaphor (item (8), Fig. 2) shades into inscribed attitude. In Example 6 below, the lexical metaphor lifeblood acts to positively evaluate the target: new members in any group. These ‘new members in any group’ have not yet done anything, but are posited in this excerpt as a category, with their positive value declared using a relational attributive, are:

Example 6

New members in any group are the lifeblood of the group…they are the new babies of that family . [sally4/8]

Their attribute as an indispensable or vital part of any group can be classed as a positive Appreciation of their Value to the group. This is then linked via parallelism (item (6e), Fig. 2) to their being the new babies of the group (i.e. the online discussion participants) as family. In this way, the lexical metaphor (item (8), Fig. 2) is extended to include new babies linked to new members, with the group likened to a family.

What type of Attitude is being advanced here? New babies in any family, while no doubt positively valued, will not usually be judged under Capacity (unless it be negative), Normality (positive/negative according to the usuality of the behaviour), Tenacity (positive/negative according to ‘appropriateness’ of inclination/psychological disposition), or Propriety (positive/negative according to ethical codes). However, in this instance, a provoked Judgement of positive Capacity of the target new babies associated with new members of any group is suggested due to their ability as lifeblood of the group. While new babies might be essential or vital to a family who desires them, their very innocence provides for non-judgement in terms of their behaviour. Yet, considering the co-text from which this segment was excerpted, this is the very implication the writer was advancing towards the real target of the text: the writer herself, who was at the time a new member of the discussion group. The writer implies that her presence in the group is to be appraised as positive Appreciation: Valuation (an entity having a positive social value), while she personally is not to be judged except perhaps under Capacity: both negative and positive. In this way, she is acting to construct her own textual persona, and at the same time position other group members as potentially overly judgemental – something more obvious in the complete text where these various strategies for provoking Attitude are tightly interwoven throughout the piece, demonstrating that the types of invoking strategies listed as a spectrum in Fig. 2 above are not to be considered as operating on their own, but are means for tracking how attitudinal readings can be activated by logogenesis, as distinct from those inscribed and explicit lexical tokens isolatable in texts. Thus, many of the ‘real’ targets of Attitude of this text do not appear in the text itself.

Evaluative Ambiguity and text organisation

Instances of ambiguous targets and values can mark a shift in attitudinal prosody. When this occurs, it is often signalled by combinations of the strategies identified above in my ‘invocation spectrum’, rather than by any one of these discrete ‘categories’ of invoked Attitude by itself. Below, this type of marking of phase shift is explored in more detail using one of the example texts.

To do this, I shall now turn to one further segment from a longer contribution, in order to illustrate how ambiguous targets or attitudes, especially those marked in context, function to both encapsulate and project an argument being made (Example 7 below). Its argument centres on a discussion of the nature of ‘task’, the association of task with ‘work’, and the valorisation of both these concepts as socially positive via association with other positively evaluated concepts. This topic was not unusual in this discussion group, which had been started by a psychologist inviting participants to discuss group dynamics and the work of Wilfred Bion, whose theories involved the use of ‘tasks’. In effect, the writer creates several clusters of attitudinal stances, creating out of them a type of constellation of stances (c.f. Maton 2014: 168) which act to reinforce their positive axiological charge, i.e. their moral or ethical superiority. A reasonably long extract from the post is offered below (Example 7) in order to enable a discussion of how attitudinal potential develops logogenetically as the text unfolds. In order to support the discussion, an in-text analysis (material in square brackets) of key Attitudinal values is provided.

Example 7

1The concept of “task,” has a rich [ Appreciation–inscribe ] history here. 2Not only is there a common sense meaning of task as the job to be done, but it is a technical term in Bion’s group psychology. 3I have been one to see task as analogy -- harking back to its roots in “tax” or an onerous tribute [ Appreciation–inscribe ] to be paid. 4In Bion, it has more positive connotations, [ Appreciation–inscribe ] and being a work group in accomplishment of a task is not only healthy [ Appreciation–inscribe ] but morally good . [ Judgement–inscribe ] 5 It is hard to mesh [Attitude ???] all this . 6I set out to work at the warehouse this morning. 7I will have a task, I suppose, or various ones. 8I must unload some trucks. 9I must aid the company [Judgement: Propriety–Evoke] in any legit [ Judgement–inscribe ] way to help it make a profit [Judgement: Propriety–Evoke] . 10I must fit myself into the sometimes odd social scheme [ Appreciation–inscribe ] there 11My goal, however, for this day is to have as pleasant [Appreciation–inscribe] and as delightful [Appreciation–inscribe] a day as I can -- to tell no lies [ Judgement: Propriety and Veracity–inscribe ], hurt no one on purpose ,[ Judgement: Propriety–Evoke ] and be a good [ Judgement: Propriety–inscribe ] citizen while squeezing the best out of whatever situation I may encounter . [Judgement: Tenacity: positive:–provoked] 12Out of this fluid plan for the day, one that will most likely materialize, which activities constitute ‘tasks’?

It is the function of sentence 5, It is hard to mesh all this, which is the primary focus of the discussion which follows it. Of specific interest is the apparent ambiguity of the evaluative ‘hard to mesh’ in terms of the specific sub-type of attitude being activated, the nature of the attitudinal target and the exact position of the writer in relation to all this. This clause is taken to be a major boundary in the argument structure, marking a transition between one phase - the outlining of background material on the notion of ‘task’ in the psychology of Bion - and the argument which the poster is about to develop. Subsequent to SE5 the types of attitude prevalent in the text switch from Appreciation (teamed with one ambiguously invoked attitude), to a patterned mixture of Appreciation and related Judgement tokens (see the above in-text analysis). In this way, SE5 can be considered to function as type of phase boundary marker in this co-text. In terms of lexical association, something which is ’difficult/hard to do’ is also related to the reference of ‘task’ with its original meaning as an onerous tribute to be paid in SE3, and thus it is also linked via identification (c.f. Martin 1992: 27) to the first phase of the text. In this way, invocations of attitude can sometimes also function as ‘strategies’ through which it becomes more difficult for the analyst to reach conclusions as to the precise attitudinal value projected by the writer, thus creating sites of semantic tension. In turn, the location of these sites of semantic tension appear to occur at strategic junctures in the development of text organisation, as with this example. In support of similar observations (Hunston & Thompson 2000: 11) cite Sinclair (1987) who “argues that evaluation, in writing as in speech, tends to occur at boundary points in a discourse, thereby providing a clue to (‘monitoring’) its organisation.”

The ambiguity inherent in this excerpt (SE5) “It is hard to mesh all this”, centres around whether it is hard for the writer to mesh all this, i.e. all of the thoughts about the nature of ‘task’, which is the theme of his discussion; or whether meshing all this is being presented as objectively hard for everyone. In the first case, the writer negatively appraises his own Capacity, and the positioning with respect to his interlocutors might be one of deference through an expression of lack of expertise in this field (c.f. Martin 1992: 530). In the second, more likely case, he evaluates the nature of all this as ‘complex’ - as difficult to comprehend in its entirety, and the positioning thus calls on some solidarity with the audience. In either case, the appraisal is provoked by the term ‘difficult/hard (to do s.t.)’, which may depend for its negative or positive value on community-held norms regarding ‘difficulty’. In this latter case, the positioning strategy would act to call on equal status (in terms of expertise) with interlocutors. So, in terms of the positioning of audience members, the strategy here construes an equal status with interlocutors in terms of working together on a difficult task, one of which is to decide what the concept of ‘task’ means!

Conclusion

In her Corpus Approaches to Evaluation: Phraseology and Evaluative Language, Hunston states:

[There is agreement] that where words or phrases are typically associated with affective meanings [i.e. positive/negative assessments or responses], an affective interpretation may be warranted even when there is no apparently affective lexis involved. (Hunston 2010: 62)

To which she adds,

…a word or phrase may have evaluative associations [i.e. convey or activate a positive or negative assessment] only in certain types of texts or even in more restricted contexts. (2010: 64)

With this in mind, the purpose of this paper has been to identify some of the specific analytical challenges associated with such language (i.e. language which is ‘affective’ even while ‘there is no apparently affective lexis involved’) and to assist such analyses by providing a more extended account than has previously been available, of the inter-, extra- and intra-textual resources typically associated with this indirect ‘invocation’ of attitudinal stance. Underlying this purpose is a call for this type of ‘analytic accountancy’ to be more rigorously carried out in using Appraisal analysis.

This paper has offered a slight adjustment of the taxonomy of attitudinal invocation sub-types proposed by Martin and White (2005), proposing that classificatory weighting be given to the distinction between, on the one hand, resources which rely entirely on the reader referencing and applying attitudinally-charged cultural/communal frames (Evoke) and, on the other hand, resources which involve some overt signalling in the text that an attitudinal value is at stake (Provoke). The proposed array also notes that these resources should not be seen as a typology, but rather considered as overlapping sets of mechanisms or text-strategies in context. My reasoning arises from the observation that there is a notable rhetorical difference between utterances where the writer relies on the reader to make an attitudinal inference (evoking) and utterances where the reader is given signals that an attitudinal value is being offered (provoking). With respect to my concern regarding the enactment of textual persona, this distinction is critical given that the more a writer relies on evocation, the more they construe a relationship of affiliation and axiological alignment with the intended/putative reader. For the same reason, engagement resources and the co-positioning of addressees have been shown to operate in activating or invoking attitude as well. Furthermore, as indicated earlier, the use of either invoked negative or positive assessments can be used to gather whether audience members are construed as in-group members or not. In this way, particular textual personae will be construed according to the extent that the writer does assume such ‘solidarity’ with readers, and according to the particular value positions (via particular attitudinal assessments) of specific targets, and for which such writer-reader alignment is assumed. This two-way Evoke versus Provoke taxonomy is thus also designed to support the analysis of textual persona in that it facilitates tracking of the degree to which and the contexts in which the writer projects this ‘like-minded’ reader.

The paper has also offered an extended repertoire of mechanisms and resources by which attitudinal values may be invoked. Various forms of intertextuality were noted to contribute to the invocation of attitude in certain social contexts, with attitudinal evocation (what Martin and White term ‘afforded’ attitude) observed to involve intertextual reference, in the broadest, Bakhtinian sense of the term. In addition, the spectrum presented includes the potential for attitudinal invocation of logic-semantic relations such as those of contrast and parallelism, as well as the potential of one type of attitudinal inscription to invoke a different type of attitude – something already noted widely in the literature but not yet formalised to any degree, except perhaps in Thompson (2014).

In this sense, it was noted that some invocations involve indicating a stance towards multiple attitudinal targets–typically both what might be termed an ‘immediate’ co-textual target, but also making reference to an ‘ultimate’ actual target. This possibility was addressed briefly by Martin and White (2005: 67) when they noted that an inscribed Appreciation of an artefact might invoke a Judgement of the capacity of the creator of the artefact (thus immediate target = artefact; ultimate target = creator of the artefact). However, this paper has argued that this functionality operates more widely across a range of different inscriptions (for example, inscribed Affect functioning to invoke Judgement) as well as the role these rhetorical ‘strategies’ may play in the construal of textual persona. Extending this further, the invocation of targets via associative reference – as distinct from the invocation of attitude itself – was also highlighted, and I argue that the invoking of targets by association is also an important means of enacting evaluation and positioning audiences. One strategy raised was the use of ‘ideational chaining’ or ‘chain interaction’ (Martin 1992: 371, Hasan 1985) where the targets of attitude were implied by co-association within the same text, sometimes referred to as intra-textual reference, whereas in other cases, the associative mechanisms for implying the ‘real’ target depended on audience awareness of the cultural frames in which the concurrent argument operated. In this way, the discussion also drew attention to the possibility that invocations may be ambiguous or under-specified with respect to the what/whom is being targeted for assessment, a potential which, again, may have consequences for the textual persona being performed.

By addressing issues across these multiple fronts, it is hoped that the paper has clarified somewhat our understanding of the linguistic mechanisms by which attitudinal values may be indirectly invoked and hence assist those whose research agenda requires that they tackle this methodologically challenging domain of meaning making.