‘Noticeably Better / Visibly Shaken’: The Use of Four Adverbs of High Perceivability with Meanings of Attitude

Abstract

This paper is a study of the use of four adverbs of high perceivability, manifestly, noticeably, patently and visibly (MNP&V), with expressions loaded with evaluative meanings belonging to the Appraisal category of Attitude. The paper starts with a characterization of MNP&V in terms of their meanings of manner and evidentiality and also as devices of Engagement within the Appraisal framework. Next, an analysis is carried out of the occurrences of the adverbs in the British National Corpus, in terms of the Attitude meanings present in the linguistic units of which MNP&V are constituents. The results show that these stretches of language often express Attitude, and that the distribution of the subtypes of Attitude varies across the individual adverbs. All the adverbs except noticeably are strongly associated with negative evaluation, which shows their frequent use for justifying high commitment to the truth of an undesirable proposition on the grounds that this truth is highly perceivable.

Keywords:

• Adverbs
• perceivability
• evidentiality
• manner
• (un)desirability
• Appraisal
• Attitude
• Engagement

1. Introduction and structure of the paper

This paper sets forth an analysis of the use of four adverbs of high perceivability, manifestly, noticeably, patently and visibly (henceforth MNP&V) with stretches of language loaded with meanings belonging to the Appraisal category of Attitude. MNP&V have been selected for analysis on the grounds of semantic similarity, in the sense that they express high perceivability and have a meaning of manner as well as an evidential meaning. The data analyzed are the occurrences of MNP&V in the British National Corpus (BNC). The paper goes a step further in the research carried out in Carretero (2019), where an account was provided of the factors associated with each of these meanings. Drawing on this previous research, the present paper provides an account in terms of the Appraisal system of both the adverbs themselves and the clauses or phrases of which they are constituents.

This work starts from two research hypotheses. The first is that most of the clauses or phrases in which MNP&V occur will be loaded with evaluative meaning. The reason is that these adverbs are devices used by speakers or writers for justifying high commitment to what they are saying, by indicating that this information is highly perceivable: it is then reasonable to predict that this justificatory effort will often be invested when the truth of the information transmitted is favorable or unfavorable to the speaker/writer’s interests (and/or those of the addressee and others). The second hypothesis is that, due to peculiarities of each adverb such as the distribution of the manner and the evidential meanings and the association with perceptual or cognitive evidence (Carretero 2019), the distribution of different evaluative meanings in these clauses and phrases will strongly vary depending on the individual adverb.

The choice of the Appraisal system is due to its current status as a thriving framework for research on evaluative language.¹ Since its emergence in the early 1990s in Australia within the Systemic Functional School of Linguistics, the framework has been developed in the seminal book authored by Martin and White (2005), and since then it has been supported and refined by the work of many scholars of prestige such as Monika Bednarek, Caroline Coffin, Alexanne Don, Matteo Fuoli, Donna R. Miller, Maite Taboada and Geoff Thompson, among many others. For the purposes of this paper, the Appraisal subcategory of Engagement provides an enlightening view of the evaluative meaning of MNP&V (addressed in Section 3.6.), while the subcategory of Attitude and its subtypes capture the different kinds of feelings expressed in the linguistic units of which MNP&V are constituents (addressed in Section 4).

The paper is structured as follows: Section 2 describes the data analyzed, i.e., the occurrences of MNP&V in the BNC. Section 3 characterizes MNP&V in terms of two meanings (manner and evidentiality) which may merge or coexist, and also as devices realizing the Appraisal category of Engagement in the sense that they express strong commitment to the truth of a proposition. Section 4 describes the system of Attitude; the description is illustrated with examples of stretches of discourse containing MNP&V that realize the different subcategories. Section 5 covers the methodology of the quantitative analysis, including two criteria for distinguishing between Attitude subtypes, and a series of inter-annotator experiments that prove the reliability of the categorization in terms of these subtypes. Section 6 presents and discusses the results of this analysis. Finally, Section 7 sums up the main conclusions and provides suggestions for further research.

2. The corpus data

The data that serve as the basis for this study are all the occurrences of MNP&V in the BNC, World Edition, created by Oxford University Press in the 1980s.² The BNC contains approximately 100 million words, divided into written language (90%) and spoken language (10%). Most of the texts were produced from 1975 onwards. The spoken section comprises both context-governed texts (leisure, educational, business and public-institutional) and non-context-governed or ‘demographic’ texts, and the written section includes imaginative texts (drama, prose fiction and poetry) as well as informative texts (leisure, world affairs, commerce and finance, etc.). The occurrences have been retrieved with the SARA software provided with the distribution of the BNC. Table 1 specifies the total number of occurrences of MNP&V in the BNC in spoken and written languages, as well as their frequency per million words.

Table 1. Total number of cases of MNP&V in the BNC.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Written	195	98.48	323	98.48	204	96.68	378	99.47
Spoken	3	1.52	5	1.52	7	3.32	2	0.53
Total	198	100.00	328	100.00	211	100.00	380	100.00
Ratio pmw	1.98		3.28		2.11		3.80

Table 1 shows that the adverbs virtually occur only in the written subcorpus: there are only 17 spoken occurrences of the four adverbs altogether. All the examples included in the paper are cited from the BNC.

3. Description of MNP&V

3.1. Dictionary meanings

MNP&V have a semantic feature of high perceivability, in the sense that the sp/wr has (or claims to have) received sensorial or cognitive stimuli that lead him/her to express full commitment to the information that s/he transmits. Their meaning was searched in the Oxford English Dictionary Online (OED), accessed September 9, 2021. The definitions that the OED provides for the four adverbs are as follows:

• manifestly: 1. Modifying a verb: in a manifest manner. Now rare. 2. Usually modifying an adjective or adjective phrase: as is manifest; evidently, unmistakably, openly.

• noticeably: in a noticeable manner; to a noticeable degree; perceptibly, remarkably.

• patently: In a patent manner; openly, obviously, manifestly, clearly. Also: (in early use) publicly.

• visibly: 1. In a visible manner (†or form); so as to be visible to the eye or sight. [Passing insensibly into next]. 2. So as to be clearly evident, manifest, or perceptible; to an extent which can be (readily) seen or observed; evidently, plainly; manifestly, obviously. 3. By actual sight. Obsolete. Rare.

These definitions hint that MNP&V have a meaning of manner, since they are all paraphrased by ‘in a + base adjective + manner’, and also an evidential meaning, since they are described by means of unmistakably evidential synonyms such as evidently, perceptibly or obviously.

3.2. The meaning of manner

The present characterization of the meaning of manner of MNP&V is based on systemic-functional linguistics. Manner Adjuncts belong to the ideational metafunction in the sense of Halliday and Matthiessen (2014), since they express circumstances associated with the process, thus being part of the meaning of the clause as a representation of ongoing human experience.

Halliday and Matthiessen (2014, 318–320) divide Manner Adjuncts into four subtypes: means, comparison, quality and degree. Among these, the meanings which concern MNP&V are those of Quality and Degree. Quality Adjuncts respond to the interrogative how? or how … ? plus the appropriate adverb, while the corresponding interrogative for degree Adjuncts is how much? The meaning of quality is paraphraseable with ‘in a + base adjective + way’, as is the case of visibly in (1); the meaning of degree is paraphraseable with ‘to + a + base adjective + extent’, as in noticeably in (2). The meaning distinction between quality and degree is subtle: in many cases, both paraphrases are possible, since an action carried out with high intensity may well lead to an easy perceivability, as is the case of manifestly in (3):

1. For a group of young nuns to challenge the full authority of the Chinese state, visibly expressed by the tin-hatted martial-law troops standing guard only a hundred metres away, is an act of calculated defiance.

2. This new method described above is not noticeably any slower than the old one, nor is it noticeably faster. However it does mean that the additional information can be removed from the look-up tree, which in turn means that a compression method can now be applied to the tree, such as utilizing a directed acyclic word graph

3. More than a century ago J. S. Mill argued for universal education on moral grounds, holding that it would manifestly increase the general balance of pleasure over pain, happiness over unhappiness.

3.3. The evidential meaning

Evidentiality is defined here as the epistemic justification for a proposition, based on the kind or source of evidence available (Boye 2012). Earlier work on adverbs of high perceivability, such as Greenbaum (1969), Guimier (1986) or Swan (1988), described the adverbs with terms such as ‘attitude’ or ‘perception’ but not ‘evidentiality’. By contrast, this label is used in more recent work such as Lampert (2014), Grund (2016) or Carretero (2019). The evidential occurrences of MNP&V are distinguished from the manner occurrences by their paraphraseability with ‘it is + corresponding adjective + that’, proposed in Carretero and Zamorano-Mansilla (2013) and Ruskan (2015), or ‘it is + adverb + the case that’, proposed in Nuyts (2009, 142). Two examples of evidential MNP&V are (4) and (5):

(4)	Mr Mayor, councillor says that the labor group is running the council, that's manifestly not the case. The conservative group is running the council.
(5)	By definition, the weapon has not been discarded, as it will remain on the statute book, with the renewal of the order, in terms of the life of the statute. Clearly and patently, if it were to be introduced, it would be introduced without notice.

This evidential meaning pertains to the interpersonal metafunction. More concretely, evidential MNP&V belong to the subcategory of interpersonal Comment Adjuncts called ‘propositional’ in Halliday and Matthiessen (2014, 191).³ The adverbs of this category have propositional scope, but this does not mean that they need to have a syntactic clausal scope. As Boye (2012, 183–184) points out, non-clausal scope may express an implicit proposition, which can be made explicit by means of a paraphrase with clausal scope. For example, in (6) the syntactic scope of manifestly is the adjective bored, but the implicit proposition qualified is ‘the men were manifestly bored’:

(6)

McAllister could not but agree with him, and had to stifle a grin as he led them around, the women staring, and the men, manifestly bored, dragged along to accompany wives and girlfriends who would, that night, describe their visit to the East End in terms that would do justice to a journey up the Amazon.

In its turn, noticeably has a different meaning from the manner and the evidential meaning, when it qualifies the truth of the proposition under its scope in terms of salience, in the sense that the information expressed by this proposition is worth noticing for some reason. An example is (7):

(7)

If you're lucky enough to have a colour printer, you can also fiddle with text and background colours. These, again, are pre-set to a small selection, but, noticeably, they include only those colours that normally print well with a three colour ribbon, which, given the price and target audience of the program, is a sensible move.

3.4. Cases of merger between the manner and the evidential meaning

Certain occurrences of MNP&V are paraphraseable by both the manner paraphrases ‘in a + base adjective + way’ and/or ‘to a + base adjective + extent’, as well as the evidential paraphrases ‘it is + corresponding adjective + that’ and ‘it is + adverb + the case that’. This is the case, for instance, of manifestly in (8):

(8)	If resources are being allocated in order to compensate deprivation, the absolute amount of deprivation must be relevant. Stability has manifestly failed to be achieved in recent years for the allocation of the Rate Support Grant.

These cases will be considered to have both meanings at the same time, i.e., to be cases of merger in the sense of Coates (1983). It is worth mentioning that Simon-Vandenbergen and Aijmer (2007, 150) acknowledge the coexistence of the manner and evidential meanings of a similar adverb, obviously: “If something is said to be the case in an obvious way, it also means that the speaker considers it obvious that it is the case”. Other references that acknowledge the existence of merger in similar adverbs are Carretero and Zamorano-Mansilla (2013) and Ruskan (2015). This coexistence of meanings is also acknowledged for the Spanish cognate visiblemente in Haβler (2008, 38).

By contrast, Vanderheyden and Dendale’s (2018) paper on the French cognate visiblement and Dendale, Vanderheyden, and Izquierdo Alegría (2020) about both visiblement and visiblemente, state that the manner and evidential meanings of these adverbs do not display cases of merger, for the reason that they express direct perceivability with the meaning of manner and indirect inference with the evidential meaning.

Discussion of this view about the French and Spanish adverbs lies outside the scope of this paper. With regard to MNP&V, this difference does not hold, since for both meanings the mode of access to evidence may be direct or indirect, depending on the case. If we consider some of the examples of MNP&V cited above, we can see that the access to evidence is indirect for manner visibly in (1) and for manner manifestly in (6), since the authority of the Chinese state (in the first case) or the men’s boredom (in the second case) cannot be directly perceived. By contrast, manner noticeably in (2) expresses direct perceivability, since the speed of the method can be directly perceived. As for the evidential examples, access to evidence is indirect for the occurrences of patently and manifestly in examples (4–6), while (9) below is an occurrence of evidential visibly where perceivability is direct:

(9)	She was at supper with her family. All seven of them at table together had shrunk the kitchen, and they were in noisy, exuberant enjoyment of boiled beef and carrots. Visibly, faces were turning red and shiny.

Therefore, due to the arguments set forth above, the position adopted in this paper is that MNP&V do display cases of merger in which the meanings of manner and evidentiality co-occur. It must be noted that the two meanings are also similar in that the very statement of the perceivability indicates that the truth of the proposition is not obvious to the addressee and therefore needs to be highlighted (cf. Haβler 2008, 227): for instance, it would be rare for a speaker to state ‘I am manifestly / noticeably / patently / visibly eating prawn cocktail’ in a face-to-face conversation where the addressee can see what s/he is eating.

3.5. Distribution of the meanings of MNP&V

A quantitative analysis was carried out of the distribution of the meanings of each adverb. The results, presented in Table 2, show that the majority of the occurrences of MNP&V are cases of merger, in which the manner and evidential meanings coexist. This predominance is strongest in manifestly and visibly. As for the rest of the cases, the distribution between meanings varies significantly with each adverb. Noticeably and visibly occur quite frequently with the meaning of manner, while their purely evidential occurrences are very few. Manifestly also displays more manner than evidential occurrences, but the difference is not so great. By contrast, patently displays many more evidential than manner occurrences.

Table 2. Distribution of the meanings of MNP&V.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Manner	35	17.68	123	37.50	10	4.74	81	21.32
Evidential	15	7.58	1	0.30	72	34.12	3	0.79
Merger	148	74.75	189	57.62	129	61.14	296	77.89
Salience	0	0.00	15	4.57	0	0.00	0	0.00
Total	198	100.01	328	99.99	211	100.00	380	100.00

χ² = 339.5833. p = 0.000000.

These differences in the distribution of meanings may well be related to differences in the frequency of domains of evidence in the evidential and merger meanings. Carretero (2019, 302–303) analyzed the evidential and merged occurrences of MNP&V in terms of these domains, which were divided into the following types: perceptual (perceived by the senses), divided into visual and non-visual; cognitive (derived from inferences triggered with the aid of knowledge of the world), and communicative (derived from linguistic messages). The results of the analysis show a strong association of manifestly and patently with cognitive evidence and of visibly with visual evidence, while noticeably displays more balanced percentages of perceptual and cognitive evidence. Therefore, the two adverbs more associated with sensorial perception, noticeably and visibly, display fewer purely evidential occurrences than the other two.

3.6. MNP&V as devices of Engagement

Engagement, one of the three main subcategories that constitute the Appraisal framework, pertains to the speaker/writer’s positioning with respect to the opinions that they are presenting or to possible responses to those opinions. Within Engagement, the most general distinction is that between Monogloss, which presents propositions without recognizing dialogistic alternatives (examples are unmodalized bare assertions and imperative mood), and Heterogloss, which overtly points to existent or possible dialogistic alternatives. Heterogloss allows speakers and writers to engage or disengage with their own words by quoting, reporting, acknowledging other possibilities, denying, countering, affirming, etc. (Martin and White 2005, 35–36). Heterogloss is divided into two larger categories: Expansion, which comprises the dialogic acknowledgement of other positions, and Contraction, which consists in challenging or restricting other positions. Examples of expressions that realize Expansion in most contexts are modal auxiliaries such as may or might, or evidential adverbs that downtone commitment to the claim presented in the proposition such as apparently or seemingly.

In this sense, MNP&V with both the manner and the evidential meanings are devices of Contraction, since they serve speakers and writers to express high commitment to what they are saying (if they are within the scope of negation, this commitment is denied). Within Contraction, MNP&V belong to the subcategory of Concurrence, defined in Martin and White (2005, 122) as “involv[ing] formulations which overly announce the addresser as agreeing with, or having the same knowledge as, some projected dialogic partner. Typically, this dialogic partner is the text's putative addressee”. With MNP&V, the speaker/writer is not explicitly mentioned as the one who evaluates perceivability as high; rather, this high perceivability is presented as (having the potential to be) shared by the interlocutor and others, i.e., as intersubjective in the sense of Nuyts (2001, 36–39).⁴

There are two types of occurrences of MNP&V where the Concurrence value does not occur. The first consists in the occurrences of noticeably with the meaning of salience, which might be considered as instances of another subcategory of Contraction named Counter, in the sense that they express counterexpectation. The second type comprises a subset of the occurrences with the meaning of manner, where the adverb just indicates that the state or event occurs in a perceivable way, as opposed to a covert way, as in (10):

(10)	A unionist councillor in North Belfast said the feeling among his colleagues was the IRA will target low-profile politicians. ‘The IRA know that members of our party or the DUP who have a high media and public profile will be well-watched by the security forces either visibly or covertly […]’

4. A description of the system of Attitude and its subtypes

Attitude is another of the three main categories of the Appraisal framework, together with the previously seen Engagement as well as Graduation. This last category, which lies outside the concerns of the present paper, consists in emphasizing or downtoning the strength of other expressions. Some expressions of Graduation are intensifiers applied to nouns (real, true, genuine) or to adjectives (very, really), and downtoning hedges (kind of, sort of, or something).

Attitude is defined in Martin and White (2005, 35) as a category that “is concerned with our feelings, including emotional reactions, judgements of behaviour and evaluation of things”. The system of Attitude consists of three subcategories, Affect, Judgement and Appreciation. In the next level of delicacy, each subcategory is divided into subtypes. These types and subtypes will be described and illustrated with examples of MNP&V from the BNC along the subsections of this Section 4.

4.1. Affect

Affect covers the expression of positive or negative feelings by the speaker/writer or someone else, as in I love you, Mary is happy, He left the room sadly. Affect is subdivided into the following three categories:

•	Security / Insecurity, which concerns (lack of) security, confidence, trust, fear, etc.: An example is (11), where the expression visibly distressed communicates the anxiety that Martha has about the probability of having lost her money:
(11)	Martha went out to check her trolley and the rest of us, somewhat half-heartedly, searched the room. ‘Silly idiot for bringing all that money in with her’, another of the girls muttered. ‘I expect she's left it at home’, I said. Martha soon came back, visibly distressed.
•	Satisfaction / dissatisfaction, which covers (lack of) interest, pleasure, favorable surprise, etc. An example is patently bored in (12):
(12)	And the further away, the better, anyway. He had a dismaying habit of pulling off his shirt as soon as he warmed up, and the sight of his bare and muscled torso shining with sweat was not something she dealt with rationally or easily. Sandra moped. She wandered around the lodge, looking patently bored, refusing to take part in activities, and complaining about the service, the primitive facilities, and the lack of time Matthew had for her.
•	Happiness / unhappiness: (lack of) cheer, affection, etc., as in (13):
(13)	But Lisa had less time for her, and was meeting new people, refining a way of living that even a year ago she would have claimed to despise. Phoebe could not, try as she might, despise it for her, because it was so obviously making Lisa happy; while she, clinging to a nostalgic and increasingly sentimental vision of the household, was patently not happy, not useful, and not passionate.

4.2. Judgement

Judgement concerns people’s behavior in terms of morality and adequacy. As Martin and White state (2005, 52), “[w]ith judgement we move into the region of meaning construing our attitudes to people and the way they behave – their character (how they measure up)”. Judgement is divided into Social Sanction and Social Esteem. Social Sanction, which concerns moral issues, is divided into the following two subcategories:

•	Veracity, i.e., truth, falsity and related concepts such as credibility, deception, manipulation, etc.:
(14)	Expressing his view that the case should go on, the judge made clear that he thought Mr. Gilbert was ‘a witness of truth and a very accurate and careful one’, whereas Mr. Everett was ‘patently false’.
•	Propriety, which concerns other moral issues such as (lack of) respect, modesty, fairness, etc.:
(15)	Manners, above all, should be about behaving the same way to everyone you meet; a person who sucks up to the nobs and snubs those he regards as his ‘inferiors’ is manifestly not a gent.

In its turn, Social Esteem concerns issues about the adequacy of people’s behavior, in terms of efficiency or (dis)agreement with expectations. Social Esteem covers the following three categories:

•	Normality, which concerns (ab)normality of behavior. This category has the peculiarity that being normal or abnormal is not intrinsically good or bad; rather, the value depends on the concrete occurrence. Sticking to the usual may be viewed in a positive sense (classic, reliable, tried-and-true, etc.) or in a negative sense (cut-and-dried, ordinary, unremarkable … ). Likewise, departing from the usual may be viewed in a positive sense (original, outstanding …) or in a negative sense (eccentric, odd, dated …) An example of Normality where abnormality is negatively loaded is (16):
(16)	Gooch has stamped his method and selection policy on this squad, and prior to this tour it had worked pretty well. Since assuming the captaincy, his own form has dramatically improved, but in India time has caught up with him. His marital difficulties cannot have helped, while his form and reflexes are patently awry, and he has failed to uplift his team.
•	Capacity, i.e., (lack of) talent, expertise, efficiency …
(17)	A husband and wife duo of lawyers-turned-smallholders may have founded the British Green Party, but there were no lawyers on the list, or surgeons, or stockbrokers; or, for that matter, plumbers or shop stewards. They are not a noticeably intellectual lot, the Greens, and anyone who goes to one of their conferences expecting deeply researched papers on the carbon tax issue or the structure of the future Green confederation of Europe will come away disappointed.
•	Tenacity, i.e., (lack of) will power, courage, reliability, accommodation, etc.:
(18)	The auditorium of the Holloway Odeon hit by a V2 rocket, August 1944. I was a relief manager, sent around the Gaumont circuit. At the Troxy, in Commercial Road, Stepney, they'd have American stars coming over. One big American came out to do his act and I saw him blanch, and then visibly pull himself together before proceeding.

4.3. Appreciation

Appreciation concerns aesthetic and functional evaluations; in contrast to Judgement, it applies mostly to non-human physical objects and to works of art or literature, as in This machine is useless or The novel is fascinating. Appreciation is divided into three subtypes:

•	Reaction, which describes the ways how people perceive the entity concerned:
(19)	The car occupants did not move at first, a rising condensation choked the dying ventilation, the inside of the windscreen visibly clouded to reveal nothing more than a general silhouette of the two figures in the front seat deliberating their actions.
•	Composition, which concerns consistency and internal complexity:
(20)	Just as the Spirit loses its way the harder you drive, so the Carlton is elevated to a higher level. Through a fast sweeping corner, taken, say, in third gear at 80mph, the Vauxhall is noticeably quicker by as much as 10mph.
•	Valuation, which concerns the extent of the worth of something:
(21)	The transaction would have been difficult to refuse. After prolonged persecution, the prospect not only of a respite, but also of power, manifestly appeared worth the concessions.

4.4. Criticism on the subcategories of Attitude as criteria for analysis

The Attitude categories have sometimes been criticized in the literature (Mauranen and Bondi 2003; Bednarek 2009; Fuoli 2018; Ruskan 2020, 32–34 inter alia) for several reasons, which may be summarized as follows:

• The categories are elusive and not always suitable for analysis in terms of a straightforward correspondence between form and meaning. The main feature for assigning categories might be the lexical meaning of the evaluative span or else the kind of entity evaluated.

• The analysis heavily depends on contextual factors and knowledge of the world on the part of the analyst.

• The size of the evaluative spans is indeterminate, and some spans may well contain other smaller spans.

In view of these factors, a number of criteria were designed for the quantitative analysis as part of the methodology. These criteria, together with a number of experiments aimed at checking inter-rater reliability, are described in Section 5.

4.5. Positive and negative polarity

In order to analyze Attitude in the linguistic spans of which MNP&V are linguistic constituents, it is also relevant to distinguish favorable and unfavorable evaluation or, in other words, positive and negative polarity of the evaluation. This distinction is based on meanings or connotations of lexical items or their combination, as in (22), where intelligible and effectual lead to interpret the evaluation as positive, or on the speaker/writer’s perspective, as in (23), which was classified as a case of negative polarity on the grounds that the stretch of discourse is written from the perspective of the referent of she, not from the perspective or her husband or the other woman mentioned:

(22)

even, as Bruckner used to show his classes, in a fifth or octave struck at a piano. The essence of music's higher flights is transcendental in the highest degree imaginable, in ways that are manifestly intelligible and effectual, and quite insusceptible to verbal accounting.

(23)

Her eyes were blurred by sudden tears as she locked the door behind her. As she drove away her heart speeded up into an erratic, agitated rhythm. Calm down, she told herself, through gritted teeth. And his present involvement with another woman was patently obvious by that exquisite négligé, the array on the dressing-table. But, with a man of Luke's charisma, that was only to be expected. So why the sudden blaze?

5. Methodology of the analysis

5.1. Criteria used for subcategorizing Attitude in the present analysis

All the occurrences of MNP&V were analyzed in terms of the meaning of Attitude communicated by the stretch of language corresponding to the syntactic unit of which they are constituents, which is a clause in certain cases and a phrase in others. The Attitude meaning was thus assigned in a holistic way, without taking into account the internal structure of the evaluative span. The subtype of Affect, which covers the explicit expression of feelings, is comparably clearcut, while the distinction between Judgement and Appreciation poses more problems, since both categories concern implicit expression of feelings (or ‘institutionalized Affect’ in terms of Martin and White 2005, 45). Two criteria, proposed earlier in Taboada and Carretero (2012, 288), were applied:

• Ethics versus aesthetics. This was the prioritary criterion. Ethical evaluations were classified as Judgement, and aesthetic evaluations as Appreciation.

• The kind of entity evaluated. When the evaluation was neither ethical nor aesthetic, it was classified as Judgement if it concerned persons or their psychological features such as skills, character or willpower, and as Appreciation if it concerned inanimate entities.

5.2. Experiments on the inter-rater reliability of the categories

In order to test the inter-rater reliability of the Attitude categories, a number of experiments were carried out, which involved three annotators. Annotator A was the author of this paper, and Annotators B and C were two 4th-year BA undergraduates of the Degree in English Studies at the Complutense University of Madrid. In all the cases, Annotator A annotated sets of occurrences, and each of Annotators B and C annotated half the occurrences of each set. The degree of inter-annotator agreement produced by the pairs A-B and by A-C was quantified by means of two measures:

1. the percentage of examples that showed agreement;

2. Cohen’s Kappa coefficient, a statistical test designed to measure inter-annotator agreement (Cohen 1960). This coefficient was calculated using the GraphPad calculator.⁵ The results have been interpreted in the way proposed by Landis and Koch (1977), a reference cited in the webpage:

   • - Kappa <  0: No agreement

   • - Kappa between 0.00 and 0.20: Slight agreement

   • - Kappa between 0.21 and 0.40: Fair agreement

   • - Kappa between 0.41 and 0.60: Moderate agreement

   • - Kappa between 0.61 and 0.80: Substantial agreement

   • - Kappa between 0.81 and 1.00: Almost perfect agreement

After calculating the degree of agreement, each experiment was followed by a discussion devoted to the analysis of the occurrences of disagreement.

The experiments were organized in two series.⁶ The first consists in the annotation of 160 occurrences of each of MNP&V in the BNC, selected with the randomizer included in the SARA software, in terms of the expression of Attitude in the stretch of language of which they are constituents. Annotator A annotated all the occurrences, and Annotators B and C annotated half of them (80 occurrences each). The annotators had the option to annotate each occurrence as non-evaluative or else as marked for Attitude. In the latter case, the occurrences were to be annotated in terms of one of the subcategories of Attitude (i.e., Affect, Judgement and Appreciation) and also as having positive or negative polarity (in other words, whether the evaluation is favorable or unfavorable). The results of the inter-annotator agreement of subtypes of Attitude are specified in Table 3. Cases agreed to be unclear in the discussion of the results were discarded from the calculation of agreement.

Table 3. Inter-annotator agreement in the annotation of Attitude for MNP&V (160 occurrences of each adverb).

ADVERB	manifestly	noticeably	patently	visibly
Inter-annotator agreement (A and B): 80 occurrences
Occurrences of agreement (No. / %)	62 / 80.52	64 / 83.12	56 / 70.89	45 / 56.25
Kappa	0.618	0.747	0.468	0.394
Unclear cases	3	3	1	0
Inter-annotator agreement (A and C): 80 occurrences
Occurrences of agreement (No. / %)	65 / 81.25	64 / 82.05	38 / 48.10	46 / 58.97
Kappa	0.723	0.747	0.268	0.414
Unclear cases	0	2	1	2

The degree of agreement was substantial for manifestly and noticeably, and fair or moderate for patently and visibly. Inter-annotator discussion unveiled that the lower agreement on patently was mainly due to the tendency of Annotators B and C to prioritize the kind of entity evaluated to the detriment of the ethics vs. aesthetics criterion, with the result that many cases of Veracity and Propriety were mistakenly labeled as Appreciation (e.g., the collocations patently false or patently offensive). With regard to visibly, the reason for the disagreement was mainly that Annotators B or C failed to identify many cases of Affect, a category of which this adverb displays many occurrences in comparison to the other three (see Section 6), a fact which took these two annotators by surprise. Therefore, since the main reasons for the cases of disagreement about patently and visibly were clearly identifiable, the categories of Affect, Judgement and Appreciation as defined above may well be considered as sufficiently reliable to be used by Annotator A (the author of this paper) for annotating the remainder of the cases of MNP&V and for presenting and discussing the results of this analysis.

Concerning the annotation of polarity (positive, negative or neutral), the results of the degree of agreement are presented in Table 4. Unclear cases were also discarded in the calculations of the agreement.

Table 4. Inter-annotator agreement in the annotation of MNP&V in terms of polarity of Attitude (160 occurrences of each adverb).

ADVERB	manifestly	noticeably	patently	visibly
Inter-annotator agreement (A and B): 80 occurrences
Occurrences of agreement (No. / %)	65 / 84.42	60 / 77.92	67 / 84.81	74 / 92.50
Kappa	0.786	0.646	0.585	0.859
No. of unclear cases	3	3	1	0
Inter-annotator agreement (A and C): 80 occurrences
Occurrences of agreement (No. / %)	66 / 82.50	67 / 85.90	66 / 83.54	72 / 92.31
Kappa	0.662	0.786	0.680	0.859
No. of unclear cases	0	2	1	2

According to the results of Table 4, the agreement between annotators A and B and between annotators A and C in the annotation of polarity may be considered as reliable, since all the cases fall into the categories of substantial or almost perfect agreement, except for the moderate agreement between A and C for patently. The results point then to the high reliability of polarity as defined above.

The second series of experiments concerned the next scale of delicacy in the system of Attitude, i.e., the subtypes of Affect, Judgement and Appreciation. Due to the time limitations of Annotators B and C, the analysis was restricted to 60 occurrences annotated in the first series. Again, Annotator A annotated the 60 occurrences while Annotators B and C annotated 30 occurrences each. The samples were randomly selected by block sampling, which is sufficiently representative in this case, given that these occurrences already underwent randomizing in the first series. Taking as point of departure the Attitude value of each occurrence agreed in the discussion that followed the first series of experiments, Annotators A, B and C had to select a given subtype of each of the main categories of Attitude:

• For Affect, the selection was between Security, Satisfaction and Happiness;

• For Judgement, the subcategories of Social Sanction and Social Esteem were not considered, and the options were directly their subtypes: Veracity, Propriety, Normality, Capacity and Tenacity;

• For Appreciation, the selection was between Reaction, Composition and Valuation.

The results of the degree of agreement are presented in Table 5. Non-evaluative cases included in the selected block, as well as unclear cases after discussion, were discarded from the agreement calculations.

Table 5. Inter-annotator agreement in the annotation MNP&V in terms of subtypes of Attitude (60 occurrences of each adverb).

ADVERB	manifestly	noticeably	patently	visibly
Inter-annotator agreement (A and B): 30 occurrences
Occurrences of agreement (No. / %)	18 / 62.07	12 / 50.00	19 / 65.52	22 / 88.00
Kappa	0.558	0.432	0.601	0.862
No. of non-evaluative and unclear cases	1	6	1	5
Inter-annotator agreement (A and C): 30 occurrences
Occurrences of agreement (No. / %)	22 / 81.48	15 / 65.22	13 / 50.00	24 / 92.30
Kappa	0.784	0.587	0.421	0.888
No. of non-evaluative and unclear cases	3	7	4	4

The results show that the percentages of inter-annotator agreement amount to 50% or more in all the cases. According to the Kappa coefficient, the degree of agreement is moderate in 5 cases, substantial in 1 case and almost perfect in 2 cases, both of which correspond to visibly. The subcategory that proved to be most problematic was Propriety: 16 occurrences were annotated with this category by Annotator A and with other categories by Annotators B and C. In spite of this issue, it may be considered that the degree of agreement is sufficiently satisfactory for Annotator A (the author of this paper) to consider the analysis of the Attitude subtypes as reliable and to proceed with the rest of the occurrences, in order to fulfill the aims of the paper.

5.3. The database

In order to carry out the quantitative analysis, the occurrences annotated in the experiments described above and the remaining occurrences were registered in an Excel database, consisting of four worksheets, one for each adverb, containing the following fields:

1. Example: each occurrence of the adverb in its immediate linguistic context.

2. Attitude type of the stretch of discourse in which the adverb occurs.

3. Attitude subtype of the stretch of discourse in which the adverb occurs.

4. Polarity of the Attitude, positive or negative.

6. Results and discussion

To start with, the results of the quantitative analysis of the frequency of evaluative and non-evaluative cases of MNP&V, specified in Table 6, show that the four adverbs most often occur with stretches of language displaying an evaluative meaning of some kind; the percentage is above 85% for manifestly, patently and visibly, while noticeably displays a comparably higher number of occurrences with non-evaluative spans.

Table 6. Evaluative vs. non-evaluative occurrences of MNP&V.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Evaluative	171	86.36	232	70.73	182	86.26	345	90.79
Non-evaluative	21	10.61	76	23.17	20	9.48	28	7.37
Unclear	6	3.03	20	6.10	9	4.27	7	1.84
TOTAL	198	100.00	328	100.00	211	100.01	380	100.00

χ² = 56.4458 / p = 0.0000.

Table 7 registers the occurrences with positive and negative polarity of the evaluation, i.e., favorable or unfavorable evaluation. The results show that the evaluation is most often unfavorable. This tendency of the four adverbs to occur with spans that display unfavorable evaluation hints that, since the transmission of undesirable information is inherently face-threatening, it tends to have a stronger need to be supported by evidence than the transmission of desirable or ‘neutral’ information. The adverb that most often occurs with unfavorable information is patently, followed by manifestly, which suggests the association of negative evaluation with the drawing of evidence from the cognitive domain.

Table 7. Positive and negative polarity of the evaluation in the evaluative occurrences of MNP&V.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Positive	54	31.58	103	44.40	33	18.13	136	39.42
Negative	117	68.42	129	55.60	149	81.87	209	60.58
TOTAL	171	100.00	232	100.00	182	100.00	345	100.00

χ² = 35.5833 / p = 0.0000.

The association of these adverbs with negative polarity agrees with the frequent collocational patterns of the adverbs most strongly associated with negative evaluation, manifestly and patently, with adjectives starting with the negative prefixes in- (including allomorphs as in illegal or irrational) or un-. These adjectives may be considered to be associated with negative evaluation on the whole, even considering exceptions due to the lexical meaning of the adjectives (for example, inoffensive or unassuming) or negative utterances in which the evaluative polarity of the whole evaluative span is the opposite of the polarity of the adjective (as in not unfair). By contrast, these adjectives are not common with noticeably, which displays the lowest percentage of negative polarity. The total number of collocations of MNP&V with adjectives with the two prefixes is specified in Table 8.

Table 8. Collocations of MNP&V with adjectives starting with the negative prefixes in- and un-.

	manifestly	noticeably	patently	visibly
in- adjectives	11	1	8	3
un-adjectives	23	2	17	6

With regard to the distribution of the main subtypes of Attitude in MNP&V, displayed in Table 9, the adverbs show significant differences. Affect is infrequent with all the adverbs except for visibly, for which this category encompasses almost 50% of the cases. These results agree with the observation by Izquierdo Alegría (2016, 115) about its Spanish cognate visiblemente, that it usually occurs with adjectives denoting psychological or physical states. On the whole, Judgement is remarkably more common than Appreciation. All the adverbs display a high number of occurrences of Judgement, but the percentages are higher for manifestly and patently. As for Appreciation, the number of occurrences is highest for noticeably, while manifestly, patently and visibly display a sizeable number of cases.

Table 9. Main subtypes of Attitude in the evaluative occurrences of MNP&V.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Affect	7	4.09	22	9.48	12	6.59	166	48.12
Judgement	122	71.35	101	43.53	123	67.58	117	33.91
Appreciation	42	24.56	109	46.98	47	25.82	62	17.97
TOTAL	171	100.00	232	99.99	182	99.99	345	100.00

χ² = 256.2317 / p = 0.0000

Concerning the distribution of the meanings in the next degree of delicacy, specified in Table 10, the subcategories of Affect are scarce except for visibly, which displays a high number of occurrences of Security, followed by Satisfaction. Happiness is uncommon with the three adverbs. As for the subtypes of Judgement within the subtype of Social Sanction, Veracity is only realized by the adverbs most associated with cognitive inference, especially patently; Propriety is more evenly distributed across adverbs, the lowest percentage corresponding to visibly. Concerning the Judgement categories within Social Esteem, Normality is not frequent with any of MNP&V, while Capacity is evenly distributed across adverbs and Tenacity is most common with the adverbs associated to perceptual evidence, i.e., noticeably and visibly. In their turn, the subcategories of Appreciation display a more or less balanced number across occurrences, the most remarkable facts being that noticeably and visibly occur quite frequently with Composition, noticeably is also frequent with Valuation, and manifestly is rare with Reaction.

Table 10. Delicate subtypes of Attitude in the evaluative occurrences of MNP&V.

	manifestly		noticeably		patently		visibly
	No.	%	No.	%	No.	%	No.	%
Af-Security	2	1.17	10	4.31	7	3.85	93	26.96
Af-Satisfaction	5	2.92	7	3.02	4	2.20	68	19.71
Af-Happiness	0	0.00	5	2.16	1	0.55	5	1.45
Jud-Veracity	28	16.37	0	0.00	48	26.37	0	0.00
Jud-Propriety	56	32.75	50	21.55	47	25.82	40	11.59
Jud-Normality	9	5.26	6	2.59	9	4.95	12	3.48
Jud-Capacity	21	12.28	15	6.47	16	8.79	29	8.41
Jud-Tenacity	8	4.68	30	12.93	3	1.65	36	10.43
Ap-Reaction	3	1.75	24	10.34	13	7.14	13	3.77
Ap-Composition	22	12.87	48	20.69	15	8.24	39	11.30
Ap-Valuation	17	9.94	37	15.95	19	10.44	10	2.90
TOTAL	171	99.99	232	100.01	182	100.00	345	100.00

χ² = 424.0441 / p = 0.0000.

This distribution of meanings agrees with some collocations occurring three or more times, uncovered by sorting out left or right collocates in alphabetical order. Manifestly often collocates with adjectives related to Social Sanction (Veracity and Propriety), such as unfair (6 occurrences), perverse (3) or untrue (3). So does patently, which occurs 11 times with false and also displays occurrences with offensive (5), unfair (3) or untrue (3). In its turn, noticeably has, as left or right collocates, verbs that may trigger Judgement or Appreciation depending on the entity to which they apply, such as improve (9 occurrences) and increase (8). As for visibly, it often collocates with adjectives that express Affect, such as shaken (19 occurrences), shocked (8), moved (5) or affected (4), while the verb relax in all its forms totals 15 occurrences as a left or right collocate.

7. Conclusions and suggestions for further research

This paper has set forth an analysis of the use of four adverbs of high perceivability, manifestly, noticeably, patently and visibly (MNP&V), with linguistic expressions loaded with evaluative meanings pertaining to the Appraisal category of Attitude. The corpus under analysis was the BNC. The adverbs have been described as having a meaning of manner and an evidential meaning, which, from the systemic-functional point of view, pertain to the ideational and interpersonal metafunctions respectively. Both meanings were shown to coexist in most of the occurrences of the four adverbs. With a few exceptions, MNP&V were shown to be realizations of the Appraisal category of Engagement, subtype Concurrence, in the sense that the high perceivability is presented as suitable for being shared with the interlocutor (and others).

Next, MNP&V were subjected to a quantitative analysis in terms of Attitude meanings displayed in the linguistic units (clauses or phrases) of which they are constituents. This analysis proved to be reliable by two criteria set for assigning subtypes of Attitude and two series of inter-annotator experiments. The resulting quantitative analysis confirmed the first hypothesis, by uncovering that MNP&V are most often constituents of linguistic units loaded with Attitude meanings. A related finding is that this evaluative meaning is often unfavorable: it may be then interpreted that MNP&V are often used as face-saving strategies, in the sense that they present the unfavorable information communicated as strongly justified by perceivability. The second hypothesis was also confirmed, since the analysis of the subtypes of Attitude unveiled remarkable distributional differences between the adverbs. Manifestly and patently occurred mainly with Judgement and visibly with Affect, while noticeably displayed a balanced number of cases of Judgement and Appreciation. The distributional differences persisted at the next level of delicacy for the subtypes of Affect, Judgement and Appreciation. These differences agreed with a number of collocations of each adverb which occurred three or more times in the data.

On the whole, the analysis carried out in this paper shows that, in the data analyzed, each of the adverbs studied displays peculiar associations with Attitude meanings, with the exception that manifestly and patently display similar behavior. In future, research of this kind might be extended to other adverbials of high perceivability, including frequent adverbs such as clearly, obviously and evidently, and also others such as distinctly, overtly or transparently, as well as cognate adverbs in other languages. The hypotheses might well set forth that these adverbs often occur in linguistic spans loaded with Attitude meanings, and that cognates in different languages might be dissimilar regarding the frequency of occurrence with each of these meanings. This hypothetical dissimilarity would agree with Simon-Vandenbergen and Aijmer’s (2007) finding that adverbs of certainty in English displayed few narrow one-to-one correspondences with their translation equivalents in a number of European languages (Swedish, Dutch, German and French). It is also plausible that adverbs etymologically related to vision, such as Spanish visiblemente, French visiblement or German sichtbar, might display a strong tendency to occur with linguistic spans expressing Affect.

Acknowledgments

This research has been carried out as part of the Research Project with reference number PID2021–125327NB-I00, funded by the Spanish Ministry of Science and Innovation (MCIN) and the European Regional Development Fund (ERDF). I gratefully acknowledge the support provided by the funding entities.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 The label “Appraisal Theory” is widespread, but I will be use the term “framework”, since I believe that it better described as a method of analysis of evaluative language within Systemic Functional theory rather than a theory by itself. The term “framework” is the one which appears in the Appraisal website (accessed February 13, 2022) https://www.grammatics.com/appraisal/

2 The BNC is also available online, in the version created by Brigham Young University (BYU), at http://corpus.byu.edu/bnc/.

3 Halliday and Matthiessen (2014) do not explicitly include any of MNP&V as interpersonal Comment Adjuncts, but they include similar adverbs of high perceivability such as evidently, obviously and plainly.

4 Carretero (2020, 54) shows that this characterization is also valid for the similar adverb clearly and its Spanish cognate claramente.

5 https://www.graphpad.com/quickcalcs/kappa1/?K=4

6 For a more detailed account of the experiments and their results, see Carretero (2022).