3. Using Attitude in textual analysis

3.5 Comparison of attitudinal values favoured by each writer

My comparison of the two texts starts by observing the amount and type of attitudinal values used by each writer, which gives a gross indication of the types of Attitude favoured by each. In order to trace the actual stance of the writer in the message, these attitudes also need to be seen in relation to their sources and the targets, as well as to the 'loading' given to different categories. A writer's argument - as a function of rhetorical staging - can also be traced via such an analysis, by noting what targets are appraised in what order, how these targets fall into semantically-aligned classes, and how they have been appraised in sets of contrasting negative and positive values. Quite often an argument or 'position statement' in these types of (email discussion) texts is 'built up' via linked evaluative positions vis-a-vis their targets, sometimes across several contributions. The targets of appraisal in this case, can be regarded as operating textually as signals of lexical cohesion. Explicitly evaluative conclusions are not always made in these texts, but implied in text units which are then linked via both semantic and syntactic cohesion (repetition and conjunction) to other stretches of text, and over the text as a whole. Dependence on cohesive ties for making evaluations is reflected in the readings of provoked appraisal in both texts, but provoked attitudes appear more evident in text2, where the values of the whole text are 'concentrated' in its last rhetorical unit or closing sequence (see section 4.2.2.1).

Analysis of clusterings of Appraisal or attitudinal values in this study has proved useful in determining boundaries of the rhetorical staging or phases in the sample texts, and so they offer useful indicators for local list norms pertaining to "exchange structure", or what I am calling the rhetorical organisation potential operating in texts of this kind [note1].

3.5.1 Preliminary comparison of the two texts

A preliminary comparison of the texts shows a difference in word-length, with text1 totalling 595, and text2, 885. In Table 3.4 below, the main differences in word and clause counts are displayed. Some of the differences tabulated below are discussed in more detail in Module 1. In the analysis conducted for that module, however, the number of sentences of text2 was limited to 28 to make comparison easier to perform. In this module, the whole texts were used, since values of Appraisal were used to trace the development of the text organisation which depends on the whole body of the text being used as the main unit of analysis.

3.5.3 Ratio of types of Attitude

In text1 (Table 3.5 below) there are 15 instances of Appreciation, 24 instances of Judgement, but only 2 of Affect. So that, it might appear that this writer was concerned to evaluate behaviour more than the objective world. Values of Affect are almost non-existent, with one value of INCLINATION - which is generally low in intensity on the emotional scale, as it usually expresses volition toward some goal, rather than emotional reaction - and one value of INSECURITY whose source was extra-vocalised, i.e. not sourced in the writer (1:20).

Table 3.5

A comparison with text2 shows some differences in orientation. In this text there are 12 instances of Appreciation, 51 of Judgement and 35 of Affect. This means that especially with respect to the category of Affect, there are different relative weightings of attitudinal values expressed in these two texts, with [affect: insecurity] accounting for 12 of the 34 identified attitudes of Affect in text2 (or 35% of its values of Affect), and 12.3% of the total values of Attitude identified in text2 overall . This compares with 2.5% for [affect: insecurity] expressed as a function of all values identified in text1. Other comparisons can be made by reference to Table 3.5 which summarises the main attitudinal values identified in the two texts. It should be noted that analysed values of Affect were further subdivided into negative and positive Affect, which was not done with categories of Judgement and Appreciation.

As noted above, Table 3.5 shows that the greatest overall difference in weighting in the two texts is given to values of Affect, with text1 according only 5% of its evaluative categories to that of Affect, compared to text2's 35%. This may be coincidental to the main topic of the post represented by text2, but it does appear related to aspects of the writer's argument concerning a conviction that thoughts and feelings should be honestly revealed in discussion groups. Somewhat paradoxically, Judgements of Veracity, although present in text2 - as distinct from text1 in which such Attitudes did not appear at all - are still scarce at only 2 instances, or 4% of all its Judgement values, and 2% overall. Within Affect, however, the most prevalent value in text2 is that of Insecurity, and this is related to the overall theme of the text, even if this is not explicitly stated. The next highest weighting in Affect for text2 is given to values of Inclination - 23.5% of all values of Affect, and 8.2% of all the attitudinal values identified in this text overall. In contrast, text1 shows only 2 values of Affect (similarly, Insecurity and Inclination as discussed above) which represents only 5% of all identified values of attitude in text1. For these reasons, these two values of Affect will be discussed in more detail below (3.6), both in terms of the nature of the targets and their sources in these texts, as well as the apparent orientation in text2 towards a theme of 'insecurity'. Meanwhile, a short inspection of the the values of judgement and appreciation identified.

For both texts, values of Judgement, and therefore evaluation of human behaviour, seem of more interest than values of Appreciation or Affect, but I suggest that these relative weightings would be found in almost any text taken from this context of interaction [note2]. As was pointed out earlier, for the writer of text1, the relative weighting of values of Judgement is higher than that found in text2, at 59% and 53% respectively. In terms of [judgement: propriety] on the other hand, text2 evidences a higher proportion than text1 does. Further research might reveal whether such patternings of evaluative orientation are typical and thus 'unmarked' for these contexts. Turning to appreciation, less than half the number of values of appreciation were indentified in text2 than in text1 (12.3% against 36.5%) when calculated as a function of all values used in each text. If this is normalised to values per 1,000 words however (c.f. Chart 3.1), it appears that text2 uses a higher proportion of values Judgement: Capacity and Propriety, whereas text1 uses a higher proportion of values of Judgement: Tenacity. So that, the writer of text1 appears more concerned to evaluate using values of Appreciation: Valuation, and Appreciation: Reaction than does text2.

Within Judgement, the most obvious difference in the texts occurs with values of Capacity and Tenacity. In Text1, Tenacity shows the highest percentage of the attitudinal values identified (26.8% of all values in text1, or 18.5 per 1,000 words), and text2 accords a similarly high ratio to values of [judgement: capacity] (18.5% of all values, or 20.3 per 1,000 words). This tends to indicate that the writer of text2 is concerned to evaluate human capability, whether negative or positive, while text1 is concerned to evaluate the nature of human reliability and dependability. A similar comparison of Judgement weightings given to values of Propriety identified in the texts reveals that the writer of text2 seems much more concerned with correctness and morality than the writer of text1. Text2 makes judgements of Propriety in 23.7% of all its value statements (or 26 per 1,000 words), whereas Propriety represents 19.5% (or 13.4 per 1,000 words: cf. Chart 3.1) of text1's overall identified values. It needs to be stressed once more, that these weightings take into account those attitudinal values which were evoked, or provoked by co-textual signals, and so some of these evaluations are not inscribed or made overtly in the texts. This was felt to represent a clearer demonstration of the possible attitudinal values interpretable in the texts, and it also highlights one of the areas in which differences in textual persona can be examined: text2 is rich in attitudinal positioning, but these are not generally explicitly inscribed. The writer uses a series of moves in order to effect her evaluative positions, and these rhetorical strategies which depend on ambiguity of source, target, or attitudinal value seem to be characteristic of this writer's 'style'.

3.6 Sources and Targets of Attitude in the texts

While the weightings of attitudinal values favoured by each of the writers might be somewhat indicative of their respective interpersonal positionings, a closer look at the identified values of Judgement: Capacity, Tenacity, and Propriety, along with the targets and sources of the evaluative positions shows the orientations of the writers more clearly. This will be undertaken below, along with further discussion of values of affect.

3.6.1 Comparison of favoured sources of Attitude

One interesting area in the findings is that relating to the attributed source of the appraisal values. 28% of Text2's sources of attitudinal values can be traced to the writer herself, when this is confined to those sources which are overtly identified in the text i.e. as the "I" of the Addresser. In addition, there are several other instances of hidden or implied attributions to the self, as well as a fair number of 'averred: non-sourced' statements which leave the nature of the source of the attitudes in the text unnamed, as general knowledge or accepted states. If these are taken into account as instances of self as source, then the proportion of appraisers linked to the writer is closer to 86%. In Text1 there is a similar large proportion of "I" as appraiser, with 23% of all overt or 'emphasised' sources located there. If other instances of hidden or averred non-sourced appraising is counted as well, then this proportion climbs to 49% - still a great deal less than in Text2, however. The remainder of sources in Text1 is taken up by those indexed as outside the text - 4 attributions to authorities, 3 to general knowledge and common sense, 4 to the context of the email list itself including a reference to inclusive-we, and 4 to the writer's company management. Such extra-vocalisations tend to make for a less self-centred text, or, in other words, the heteroglossic positioning in the text seems to be more open.

The tracing of sources of Appraisal, especially in terms of the sources of attitude as either intra- or extra-vocalised, addresses the nature of authorial persona obliquely, as a by-product of the strategies for the evaluative co-positioning of interlocutors, and the construction of a dialogic space in the text. As Martin & White observe:

With these texts, sources of appraisal are felt to be one of the means for characterising the actual 'voice' of the Addressers, especially when they take up Speaker roles . This is related to what Goffman (1974, 1981) differentiates as principal and animator, or in Genette's (1980) terms, intradiegetic narrators. However, the terms are not exactly the same: the principal identifies the self as the source of the proposition, the creator or responsible one, whereas the animator is only the vehicle for the statement, and may distance the self from responsibility for it. Principals and animators may be one and the same Addresser. In this sense, the Addresser is always an animator, but may not always be a principal. On the other hand, if an Addresser takes up a Speaker role, then they are animator only, with the principal explicitly located elsewhere. The Speaker role occurs most obviously, when for example, writers relinquish the Addresser role, and introduce quotations or the ideas of another writer, or when they change from the usual registers common in this context to write poetry or engage in performances of interdiscursivity.

At times less obviously, each evaluative proposition can be more or less traced to an internal or external voice. As described above (2.3), under Engagement, three voicings are possible for statements: monologistic, intra-vocalised, and extra-vocalised. [note3] While monologistic statements are those which do not indicate any source for the statement, intra-vocalised statements do indicate some trace of the writer-voice, especially when these are framed by, for example, interpersonal adjuncts, or projecting clauses. This has some rhetorical significance if, for example, the purpose of analysis is to differentiate between the voicing as reflected in the sources of appraisal in the following two statements:

Ex. 3.1:

New members in any group are the lifeblood of the groupÉ(2:8)

Ex. 3.2:

I know there are people here who fear meÉ (2:33)

In contrast to that in example 3.1, the statement in 3.2 can be seen as two separate propositions:

I know [s.t.]
and
There are people here who fear me

Each of these propositions derived from example 3.2 could be classed as "averred: non-sourced", or as bare assertions attributed to no-one. On the other hand, the original clause complex of 3.2 contains a projected clause containing the evaluation and framed by I know, and can be classed as sourced to the Addresser 'I'.

In contrast, in the case of example 3.1, the statement is unsourced and therefore monologistic: the Addresser does not appear as the principal, but only as the animator of ideas or propositions which do not need to be argued, and which are represented as not contingent on any specific subjectivity. These types of monologistic statement thus have a high degree of rhetorical significance in terms of the evaluative positioning of interlocutors, who are positioned as in agreement, or as sharing the evaluation. Whereas, in the case of example 3.2, rhetorical significance of another kind operates on this statement: here the writer acknowledges the contestability of claiming to know the condition of others' affective states, by, in effect, indexing her own subjectivity.

3.6.2 Comparison of Attitudinal targets, and textual cohesion

The targets of Appraisal, those objects or causes of attitudinal values, function as a tracking device for the unfolding of the discourse and its arguments. Targets appear to be grouped in 'clumps' of semantically-related collocates, and it is possible to trace the development of the argument by following the targets through the text. It also helps in tracking the strategies the writer uses for making the larger points of evaluative positioning in terms of contrasts and parallels, and is related to the interpersonal metafunctional values in texts said to be prosodic or field-like in nature (as contrasted with the ideational: particulate, and the textual: wavelike - c.f. Young et al, 1970). In tandem with other types of analysis, such groupings of targets of Appraisal are useful in characterising register, the boundaries of 'interactional moves' or move complexes (strategies), and the larger rhetorical units within texts. Such overlapping and embedded 'units' could be grouped under Gregory's (op cit) collective heading of 'phase', as discussed above (section 1.3.1), or Cloran's (1993) notion of rhetorical unit (see for example Hasan 1996: 117) which suggests that texts are composed of one or more rhetorical units and these in turn are composed of one or more messages. This thesis suggests that rhetorical units within and between larger text units can be identified using Appraisal analysis as one means of highlighting such rhetorical staging. In the next section, I examine the two example texts by discussing some of the ways in which this might be achieved. In Module 3, Chapters 4 and 5, targetting will be again discussed in more detail with particular reference to its use in construing both textual identity, and texture via 'identity chaining'.

notes:

1. This also relates to what Quirk et al (1972: 265) distinguish as dynamic and stative adjectives: "For example, a stative adjective such as tall cannot be used with the progressive aspect or with the imperative: *He's being tall ; *Be tall. On the other hand, we can use careful as a dynamic adjective: He's being careful, Be careful."

2. For a different genre, that of teacher anecdotes, McCabe found that 30% of the total Attitude tokens belonged to Affect, 32% to Appreciation, and 38% to Judgement in a corpus of 12 texts (personal communication).

3. As discussed previously, monologistic is used instead of monologic in order to differentiate monologistic statements from what might be confused with the common understanding of the term monologue. While these texts are in fact monologues in one sense, one of their values for research into textual interaction is that projection into dialogue (Hoey, 2001) is indicated in these texts by the mode-bleeding and other strategies for signalling involvement, as outlined in Part I of this module.

Next: Section 4: 4. Appraisal and Discourse Organisation

Back: Section 2: 3. Attitude as part of Appraisal

References

Index