Skip to navigation – Site map

Valence and arousal in WHAT and HOW exclamatives: cognitive simplification versus emotive implication

Julie Neveux


This paper focuses on the cognitive and semantic difference between HOW and WHAT exclamatives in a literary corpus (two collections of short stories written by Katherine Mansfield and published in 1923, Bliss: and Other Stories and The Garden Party: and Other Stories). A bi-dimensional model of emotion is used (T. Colibazzi et al. [2010], J. Posner et al. [2009]), with valence and arousal being systematically studied, among other cognitive, narrative, semantic and syntactic criteria, to analyse 249 occurrences. WHAT exclamatives mainly serve social purposes and express external perception-based processes and typically have marked valence and low arousal, while HOW exclamatives typically signal strong arousal and neutral valence. The “symbolic” (R. Langacker [2009:1]) meaning of specific parts of speech plays a role in such a semantic distribution: nouns in WHAT exclamative phrases reveal a need to categorize and evaluate, two cognitive operations resulting in an overall simplification of the complexity of the world. HOW exclamatives, on the contrary, display the speaker’s effort to identify and qualify, via the adjectival head, a complex and emotionally charged quale (felt as subjectively unique).

Top of page

Full text


1There are many reasons why emotions are difficult to define. They are subjective, cognitive experiences and as such, difficult to translate into language. They are more or less observable, and although they might imply bodily changes (the perception of which corresponds to emotion or to feeling according to A. Damasio [2003: 91-92]), such physiological alterations are not necessarily manifest. They are intimate, introspective processes. External or internal perceptions may trigger them, and even constitute them. Neuroimaging shows that there is a constant interaction between perceptions and emotions, which goes both ways, with more or less impact. L. Ciompi and J. Panksepp [2005: 45] confirm that intense emotions may, for example, prevent informative processing, and vice versa: “Mild to modest emotional arousal generally increases the overall amount of cognitive processing, while intense arousal tends to inhibit it”. We may think of the emotional and cognitive state of a (very sick) patient at the hospital who has just been told some bad news and, facing his or her doctor, cannot listen and process any information anymore.

  • 1 Expressing emotions differs from describing them, as Celle and Lansari [2017: 4] emphasize. This pa (...)

2It is possible that the most intense emotions cannot be studied linguistically. But some linguistic markers and structures do specialize in expressing1 emotions, such as, typically, exclamative constructions. A literary corpus-based study is useful as it provides large enunciative context (diegesis) where linguistic expressions of emotion might be systematically linked to their stimulus/i and cognitive motivation. Katherine Mansfield uses exclamation a lot more than other writers, probably because the expression of emotions (and not their description – which, if she expanded on it, would turn her into what she abhorred, a sentimental writer) is at the core of her work, with most of her protagonists being highly emotional characters frequently resorting to exclamatives. Distinctive linguistic patterns thus emerge for HOW and WHAT exclamatives, in lexicon, syntax, type of emotions and type of perceptual triggers. The systematic comparison of the two constructions favours a deeper understanding of the link between specific linguistic phenomena and specific cognitive motivations, and also of the link between perception and emotion.

1. Method

1.1. Corpus and analytical criteria

3I have gathered all (independent or subordinate) exclamative clauses in two collections of Mansfield’s short stories, Bliss and Other Stories (1923) and The Garden Party: and Other Stories (1923). I then analyzed the 249 exclamatives according to the following 9 criteria, in an effort to comprehend a variety of aspects of the enunciative situation and of the resulting utterance. The criteria are cognitive, lexical, syntactic and narrative:

  • 2 I have exclusively compared HOW and WHAT exclamatives when they worked as predicative complement, f (...)
  • 3 The description of WHAT nominal phrases and HOW adjectival phrases as “exclamative phrases” is take (...)

(1) HOW or WHAT in the exclamative construction2
(2) ellipsis of subject and verb or not
(3) narrative status of the utterance (type of speech, either direct when the character interacts with others, direct thought when the character is alone, etc.)
(4) presence or absence of lexicalized words of emotions in the close co-text
(5) presence or absence of interjections (
Oh! Ah!, etc.)
(6) punctuation
(7) arousal (intensity of the exclamation, high, medium or low) as defined by J. Posner
et al [2009]
(8) valence (in positive or negative value of the exclamation, positive, neutral or negative) as defined by J. Posner
et al [2009]
(9) type of adjectives in the exclamative phrase
3 (evaluative or descriptive)

1.2. A bi-dimensional model of emotion: valence and arousal

1.2.1. Emotions as perception of values

4Classifying emotions has been shown to be problematic, but there is a consensus on the existence of two core characteristics of emotions since J. Russell’s 1980 model, which have been called the two “dimensions” of emotions: valence and arousal. Valence, meaning the evaluative aspect of cognition, is indeed a frequent, and for some authors, necessary component of emotions. Some, like C. Tappolet [2016: xi] even define emotions as “perceptions of values”, i.e. either negative or positive reactions to a stimulus:

emotions are closely related to values or more precisely to evaluative concepts such as fearsome, disgusting, or admirable and the properties that can be taken to correspond to those concepts (C. Tappolet [2016: xiii]).

5Such a definition of emotion does not distinguish significantly between emotions and perceptions (but does not altogether deny it either), and illustrates a “Perceptual Theory of Emotion”.

1.2.2. Interaction between perception and emotion

6But some cognitivists insist on the necessity to distinguish between perceptions and emotions, and rather describe them as two types of cognitive (in a large sense) events which need to be studied in their interaction. This is the approach taken in this paper. According to L. Ciompi and J. Pankepp [2005: 23], whose approaches are respectively called “affect-logic” and “affective neuroscience”, “it has become increasingly evident (in neuroscience and in psychology) that emotions and cognition are continually interacting in almost all mental activities”.

7L. Ciompi and J. Panksepp use the word “cognition” for “perception” and emphasize the specific cognitive nature of perceptive processes as informative processes. They show that emotion and cognition interact and that each can dominate the other, which they [2005: 24] characterize as “circular causality”, while attributing the following specific and distinct traits to the two phenomena:

(a) emotional processes are viewed as large scale energetic states of brain and body, that reflect evolutionary adaptive action systems of the internal world
(b) cognitive processes are more informational encapsulated perception-based processes that distinguish differences in the external word. (L. Ciompi and J. Panksepp [2005: 23], my emphasis)

  • 4 “High cortical activity generally tends to inhibit subcortical emotional systems” (L. Ciombi and J. (...)

8Emotions do not need the external world to develop and last, whereas perceptions react to stimuli from the external world. Affective processes, in L. Ciompi and J. Panksepp’s terms [2005: 25], are more “organic”, while cognitive processes have to do with information. The authors underline the effects of emotions on cognition, depending on their level of intensity (arousal), as quoted in the introduction. Intense cognitive processes may inhibit emotions4, while intense emotions reduce cortical processing, as in the example of the patient unable to understand his or her doctor anymore.

1.2.3. Valence as simplification

9A common function of cognition and emotion, which their interaction develops, is, in the terms of L. Ciompi and J. Panksepp [2005: 38, my emphasis], the

experience-based simplification of the infinite complexity of the cognitive world, by reducing it to a few behavior-relevant categories like interesting/indifferent, harmless/dangerous, useful/useless, etc.

10Valence reveals cognitive simplification, and the evaluative dimension is a sign of interaction between the two processes. But only medium or low arousal cognitive states may result in evaluative processes.

1.3. Arousal: how to measure emotions (in my corpus)

  • 5 Not only do literary corpora provide much larger contexts for utterances than “authentic” ones such (...)

11The other “dimension” of emotions is their intensity, their degree of arousal. Again, arousal is difficult to perceive and evaluate, but literary texts make good corpora5 . Criteria 4, 5 and 6 of my above-mentioned list (combining the study of lexical words of emotion, occasional interjections and marks of punctuations in the diegetic context) have helped me measure the intensity of the cognitive state in my 249 exclamatives. The following table gives 4 examples of valence and arousal analysis:

Table 1: Valence and arousal in exclamatives and situation of utterance

Exclamative in context




she had run into the house, through the hall, up the stairs into her bedroom. Down she sat on the side of the bed. How vile, odious, abominable, vulgar, muttered Isabel. She pressed her eyes with her knuckles and rocked to and fro. (“Marriage à la mode”, GP)




How beautiful she looked like that!--simply beautiful--and she was so small in that immense chair. (Reginald’s heart swelled with tenderness, but it was her voice, her soft voice, that made him tremble). (“Mr and Mrs Dove”, GP)




Wreaths of white smoke floated up from somewhere and hung below the roof like misty vines. “How strange it all is,” thought the little governess, “and the middle of the night, too.” She looked out from her safe corner, frightened no longer but proud that she had not given that franc (“The Little Governess”, Bliss)




His eyes grew bright; he quickly, lightly clasped her. “My letters, darling?”

“Perhaps,” she drawled, softly, and she drew her hand over his reddish hair, smiling too, but thinking: “Heavens! What a stupid thing to say!” (“Revelations”, Bliss)



12Arousal corresponds to measure of (high, medium or low) energetic states. Many parameters, in the exclamative construction itself and in the larger situation of utterance, be they cognitive, narrative, lexical and syntactic, have been taken into account, in an attempt to be as exhaustive as possible in our analysis. Descriptions of the characters’ thoughts and movements, for example how Isabel runs up the stairs in (1) and the use of explicit emotive (direct and associated) lexicon such as in swelled with tenderness and tremble in (2), (with tenderness being a direct emotive word and swelled and tremble, in this context, emotion-associated verbs) both indicate high arousal. On the contrary, in (4), the opposition between apparent facial positive feeling (smiling) and inner criticism (adversative but, verb thought and negative valence in adjectival exclamative stupid) all reveal a duality in the character’s state of mind and a cognitive state of control indicating low emotive arousal. High arousal is too strong to be dissimulated and may require privacy as in (1).

2. First results6

2.1. HOW or WHAT

1365% of all exclamatives in the corpus are HOW constructions, and 35% WHAT constructions.

2.2. Narrative mode of speech

14Four modes of discourse have been distinguished, which correspond to specific contexts of utterance and more or less social interactions:
- narration (external focalization);
- direct speech (situation of dialogue, social interaction);
- direct thought (utterance to oneself, close to interior monologue);
- free indirect thought (internal focalisation, no social interaction or the memory / representation of a former one).

15Free indirect thought is the predominant speech mode (53%) of all exclamative constructions. 66% of the 53% are HOW constructions. 7% of all exclamatives are in direct thought, of which 74% are again HOW constructions. Out of the 32 % utterances in direct speech, 56% are WHAT constructions (while WHAT occurrences are much less numerous than HOW occurrences, respectively 35% versus 65%).

2.3. Syntax (potential ellipsis of subject and verb)

16Elliptical constructions represent 38% of all exclamatives, 65% of these are WHAT constructions.

2.4. Potential extraposition

17Extraposed subject constructions (with subsequent fronting of exclamative phrases, WHAT appearing during the extraposition process if one follows a generative analysis of such structures) are not frequent in my corpus, for example “What an esstrordinary (sic) thing to come all this way here to tell me!” (“The Daughters of The Late Colonel”, GP) or, with mention of matrix verb, “What a pity it is that you did not arrive by daylight” (“Je ne parle pas français”, Bliss). They represent only 6% of all exclamatives and 65% of the extraposed constructions are in WHAT.

2.5. Potential subordination

18My corpus includes very few subordinate exclamative clauses (9%), and 95% of the exclamative embedded constructions are HOW exclamatives.

2.6. Valence

1939% of all occurrences show negative valence, 32% positive valence and 28% neutral valence (neither positive nor negative, e.g. intriguing, strange, extraordinary (when synonymous with strange), as Figure 1 shows.

Figure 1: Valence in all exclamatives

Figure 1: Valence in all exclamatives

20- 54% of negatively-valenced exclamations include WHAT (versus 46 % with HOW)
- 88% of neutrally-valenced exclamations are HOW constructions,
versus only 9% of WHAT constructions
- 61% of positively-valenced exclamations are HOW constructions (
versus 39% in WHAT).

21WHAT exclamatives tend to express negative valence (or positive valence) but almost never neutral valence, while HOW exclamatives, when showing marked valence, are more positive than negative and are often characterized by neutral valence. This is the most visible difference in valence statistics, as illustrated in figure 2.

Figure 2: Neutral valence: HOW

Figure 2: Neutral valence: HOW

2.7. Arousal

22Very few low arousal exclamatives (7%) have been found, which is not surprising if one remembers the nature of our corpus: a corpus consisting of exclamatives, exclamation being the typical illocutionary force of exclamatives, logically rather shows medium or high (rather than low) arousal rather than low. Yet, and this is all the more remarkable, 83% of the low arousal exclamatives are constructed with WHAT, while 91% of medium arousal occurrences (themselves representing 65% of all occurrences) and 79% of the high arousal occurrences (representing 29% of all occurrences) are in HOW.

23The most visible differences in arousal statistics is the compatibility of HOW exclamatives with medium and high arousal, and that of WHAT exclamatives with low arousal, as illustrated in Figures 3 and 4.

Figure 3: High arousal: HOW

Figure 3: High arousal: HOW

Figure 4: Low arousal: HOW

Figure 4: Low arousal: HOW

2.8. Type of adjectives (evaluative or descriptive)

24The adjectives studied function either as modifying the noun in WHAT constructions or as heads of HOW constructions, and are distributed this way (see also Figures 5 and 6):
- WHAT occurrences include 25% descriptive adjectives versus 75% evaluative adjectives
- Only 34% of HOW adjectives are clearly evaluative, while 56% are descriptive (following Cotte’s classification [1996, 134-35]), and the rest not easily identifiable.

Figure 5: Adjectives in WHAT exclamatives

Figure 5: Adjectives in WHAT exclamatives

Figure 6: Adjectives in HOW exclamatives

Figure 6: Adjectives in HOW exclamatives

25The results can be interpreted as showing a clear difference in cognitive motivation, signaled in the syntax and semantics of the two WH-exclamative constructions.

3. WHAT exclamatives: marked valence and low arousal

3.1. WHAT exclamatives and social interaction/perception

26Lansari and Celle [2015: 92-93] describe some exclamatives as expressing social, “proactive” interaction, serving social purposes such as self-staging, and representing cases of “ethical expressivity” in the terms of François and Legallois [2012], cases of exclamation motivated by representation rather than emotion. Such occurrences are co-speaker oriented.

27In my corpus, exclamatives that correspond to this psycho-linguistic, social motivation are almost systematically WHAT exclamatives. Typically, in my corpus, WHAT exclamatives combine the following semantic, syntactic and narrative characteristics:
- they occur in direct speech
- they are not embedded (which is to be expected as they typically occur in
immediate social interaction);
- their subject and verb can be omitted (since subject and verb are implicit traits of the situation and shared by the co-speaker);
- they are motivated by the evaluation of a social situation (rather than by intense emotion);
- they display medium or low arousal and marked valence (mostly negative but also positive).

3.1.1. Appraisal of social norms

28Let us have a look at some examples in context:

(1) “Egg and--” Mrs. Sheridan held the envelope away from her. “It looks like mice. It can’t be mice, can it?”
“Olive, pet,” said Laura, looking over her shoulder.
Yes, of course, olive.
What a horrible combination it sounds. Egg and olive.” (“Garden Party”, GP)
(2) “No ice! No ice! No ice!” she shouted gaily.
And Dennis chimed in from under his hat. “Only to be had from the fishmonger’s.”
And Bill Hunt, emerging, added, “With whole fish in it.”
what a bore!” wailed Isabel. And she explained to William how they had been chasing round the town for ice while she waited for him. “Simply everything is running down the steep cliffs into the sea, beginning with the butter”. (“Marriage à la Mode”, GP)
(3) Mrs. Kember rose, yawned, unsnapped her belt buckle, and tugged at the tape of her blouse. And Beryl stepped out of her skirt and shed her jersey, and stood up in her short white petticoat, and her camisole with ribbon bows on the shoulders.
“Mercy on us,” said Mrs. Harry Kember, “
what a little beauty you are!”
“Don’t!” said Beryl softly; but, drawing off one stocking and then the other, she felt a little beauty. (“At the Bay”,
(4) Silence.
“But what will you do? You’ll go back? You’ll see him?”
That made her sit right up and stare across at me.
What an extraordinary idea!” she said, more coldly than ever. “Of course I shall not dream of seeing him. As for going back--that is quite out of the question. I can’t go back.”
It’s impossible. For one thing all my friends think I am married.” (“Je ne parle pas français”, Bliss)

29The first two examples occur in quintessentially social contexts: preparing a party in upper class society (a frequent butt of Mansfield’s satire) and exclaiming about “superficial”, food-related matters of (im)propriety, which must be understood as meeting (or not) with conventional expectations of what is proper (required standards). All exclamatives presented above show either low or medium arousal from the speaker. Negative valence in (1) combines with low arousal. “What a horrible combination” is merely a cognitive response to a sandwich name for a flag, and elicits everything but horror, in a conventionalized use of the hyperbole. Such is the paradox of social exclamatives, which we will go back to later, and which we may call a phenomenon of surface intensity and in-depth semantic weakening (such cases mimic arousal but actually express valence).

30WHAT exclamatives signal the speaker’s reaction to and evaluation of a social situation according to some expectations which the speaker masters and with which he or she may play. It is the game of social appearance. Horrible is not horrible for real. Isabel’s wailing in example (2) for no reason and her friend is shouting gaily about the same issue in the left co-text. “Mercy on us”, exclaims Mrs Kember in example (3), who uses the evocation of “pity” the better to emphasize the compliment about the young girl’s appearance. Sentiments are evoked but merely for social purposes. They are used and displayed (wailed) but represented as not being felt. Statements 1 and 3 are appraisals of perceptions of external world (it sounds) in (1) and Beryl’s appearance in her bathing suit. In example (4), it is yet another type, more private, of social interaction. Only two characters face each other after the departure of a common friend (for the male narrator) or lover (for the female character exclaiming). The judgmental component of the exclamation is even more prominent in this case. The intra-diegetic narrator suggests an idea which is deemed highly inappropriate by the co-speaker. The adverbial adjunct more coldly than ever emphasizes the lack of emotion, the cognitive/evaluative process at stake and re-instates matters of moral appropriateness in a context where feelings have until then prevailed. Ethics are back in the game of love affairs, and this is marked by a WHAT exclamative. Solemn modal shall and utter deontic (ethically-based) impossibility modal can’t, and expressions of modality, that is quite out of the question and it is impossible, all emphasize the return to a more objective, less subjective (cold) reality and signal complex cognitive processing. It is a case of appraisal of situation defeating emotion.

31Exclamatives using WHAT typically express social perception, i.e. perception of an external situation and ensuing informational cognitive processing (corresponding to definition (b) of L. Ciompi and J. Panksepp [2005: 23] quoted earlier). Such perceptions are cognitively complex, they combine an assessment of an idea / a situation / an appearance according to behavioural/social norms, as well as expected reactions to those (micro) events which they reproduce. Mrs Kember in (3) knows that Beryl needs to be looked at, Mrs Sheridan in (1) knows that her sandwiches will be perfect and that expressing horror manifests her (upper class, expected) flippancy. Such exclamatives are the products of pure social interaction, in its etymological sense. Social originally comes from the Latin noun socius, probably from the verb sequor, follow. Socius used to be the associate, the partner, the one who followed. Social interaction consists in agreeing and convening, to expected, pre-set patterns of “good” behaviours, clothes, food. Mansfield, who was a New Zealander, never felt at ease with the normative British (and European) spirit, sometimes interpreting it as that superiority complex of the colonizer toward the colonized (as shown by I. McDaniels [2011: 74]). In a miniature way, her recurrent use of WHAT exclamatives is a linguistic tool of hers to point to this social phenomenon, which consists of mimicking feelings while in reality appraising and judging.

3.1.2. Nouns and categorization

32One reason why WHAT exclamatives are compatible with expression of judgment is the nature of their head. Nouns, as E. Mignot [2014], following R. Langacker [2009] explains, typically categorize. Choosing a nominal category for a referent inscribes it within a class of referents. Far from being fuzzy, the frontiers delineated by the nominal category are clear-cut. They are exclusive. The existence of prototypes within categories does not preclude clear categorization.

33The way WHAT exclamatives work in my corpus confirms this cognitive motivation: WHAT exclamatives express the speaker’s need to categorize situations, ideas, events or people into large categories and to evaluate them (with the help of simple evaluative criteria, such as good or bad). Their valence is simple to the point of caricature. As L. Ciompi and J. Panksepp put it, valence in cognition produces “behavior-relevant categories” [2005: 38]. WHAT exclamatives also illustrate the overall motivation of cognition (when influencing emotion): simplification. They signal a cognitive reduction of the complexity of the external world.

3.2. WHAT exclamatives and cognitive simplification: hypernymic heads and evaluative adjectives

34Simplification results from two parallel cognitive processes, both present in WHAT exclamatives, categorization (CA) and evaluation (EV). Two syntactic and semantic patterns can be found:
(a) WHAT + det + (CA+EV) NOUN (e.g. “What a fool”)
(b) WHAT + det + (EV) ADJ + (CA) NOUN (e.g. “what a stupid thing” (to say))

35Evaluation is conveyed either by an adjectival modifier (in (b)) or by the head noun itself in the absence of adjectives (in (a)). Evaluative and categorizing nouns are for example mistake, fool, idiot, rot, nonsense etc. Either the head noun in (a) conveys the two semantic and cognitive traits evaluation + categorization, or the traits are distributed, in the case of hypernymic heads (in b), where the adjective evaluates and the head (largely) categorizes.

36Nouns in WHAT exclamatives are often hypernyms, such as thing, idea and thought (nouns characterized by H.J. Schmid [2000: 4] as “shell nouns”). They are superordinate terms which erase any specificity entailed in the situation.

  • 7 The referent of the subject may be the speaker herself/himself, but it remains in these cases a cri (...)

37It appears that the adjectives conveying evaluation in my corpus have few semantic traits and display a somewhat Manichean interpretation of the world. Adjectives such as good and high degree good, such as enchanting, bad and high degree bad, such as loathsome, tend to qualify things. Neutral valence tends to be incompatible with WHAT exclamatives. On the contrary, it is marked valence (more negative than positive) that is typical of the semantics of WHAT-phrases. In (a) patterns, the nouns categorize as a consequence of cognitive appraisal: the subject7 is declared either a fool or a genius (“what a fool”, “what a genius he was”), and it seems that neutral valence) does not favour WHAT exclamatives. But it does give rise to HOW constructions.

4. HOW exclamatives: high arousal and neutral valence

4.1. General characteristics

38HOW exclamatives, as the statistics have shown, display different semantic, syntactic, narrative and cognitive characteristics:
- they are motivated by intense arousal or medium arousal
- they occur in emotive context
- they can have neutral valence
- they show fewer ellipses of subject and verb (and do not result from interaction)
- they can be subordinate
- they typically occur in free indirect thought (and almost never in direct speech, except when they are self-addressed).

4.2. Introspective evaluation: no simplification

39The adjectival heads of HOW exclamatives bear more semantic traits than WHAT adjectives and do not convey such cognitive simplification. Rather, they reveal a difficult identification of a complex quale (specificity of a feeling or internal perception), a complexity sometimes rendered by a juxtaposition of adjectives or by repetitions of the exclamatory phrase itself:

(5) “How vile, odious, abominable, vulgar,” muttered Isabel. (“Marriage à la Mode”, Bliss)
(6) How hideous life is--revolting, simply revolting… (“Bliss”,
(7) “How wrong, how wicked, feeling as I do”. (“Mr and Mrs Dove”,

40The succession of synonymous adjectives in all three examples conveys the speaker’s strong emotive implication and a lack of cognitive control. The speaker does not react to any immediate, social situation but expands on some actual feeling (in 7), past behaviors (in 5), existential condition (6). HOW exclamatives are often cases, in L. Ciompi and J. Pankepp’s model of interaction, of emotion weakening cognitive processes.

41All three examples entail some judgmental / appraisal trait, but this is not the case of the majority of HOW exclamatives, and such evaluation does not go hand in hand with categorization but with tentative identification. The evaluation entailed by HOW exclamatives has experiential, introspective basis. “Life” and “civilization” are possible generic subjects/topoi of HOW exclamatives but the referents “ideas” or “things”, which have particular, context-dependent meaning and do not have any existential reference, are typically to be found in WHAT exclamatives. At the syntax and semantics interface, two patterns emerge:
(a) WHAT + det + (EV.) ADJ + “idea” / “thing” => particular reference and specific exclamation, external perception
(b) HOW+ (EV.) ADJ + “life” / “civilization” (be) => generic exclamation, internal representation

42Appraisal comes from interaction in WHAT exclamatives and typically applies to something external (perception of external world), whereas appraisal in HOW exclamatives typically applies to internal emotive processes (thereby illustrating A. Damasio’s [2003: 91] definition of feeling as perception of internal state). Bertha Young (in “Bliss”), who is overwhelmed by bliss and a need to run and dance in the street, deplores the social constraints preventing her from doing so, and exclaims “How idiotic civilization is!” Negative valence is strong, but it emanates from personal and even anti-social motivation (not from social interaction). And the exclamation does not come as confirming or informing suitability of a social micro-event, but as a denunciation of the conventional standards and macro-structures which result in self-censorship:

(8) What can you do if you are thirty and, turning the corner of your own street, you are overcome, suddenly by a feeling of bliss--absolute bliss!--as though you’d suddenly swallowed a bright piece of that late afternoon sun and it burned in your bosom, sending out a little shower of sparks into every particle, into every finger and toe?
Oh, is there no way you can express it without being “drunk and disorderly”? How idiotic civilisation is! Why be given a body if you have to keep it shut up in a case like a rare, rare fiddle? (“Bliss”,

43Description of bodily movements and explicit lexicon of emotion are present in the contexts of HOW exclamatives, and the interferences of physicality on language is manifest in frequent interjections, repetitions of exclamative phrases.

4.3. HOW exclamatives and descriptive adjectives

44HOW exclamatives often include descriptive adjectives, as in the following examples: how tired, how frightened, how dusty, how tender, how submissive, how thoughtful, how white, how un-French, how peaceful, how successful, etc. Some adjectives like intriguing and strange, recurrent in HOW constructions, are more difficult to classify. In the semantic adjectival continuum proposed by P. Cotte [1996: 134], such adjectives stand between the enunciative and the referential pole of the nominal phrase, at the frontier between evaluative and descriptive adjectives. They indeed both transcribe a subjective reaction towards the referent rather than one of its objective properties. But this evaluation is neutral, it has no positive or negative value, and essentially involves expressing the failure of evaluation and description.

45The other adjectives in the list also demonstrate cognitive efforts made by the speaker, in context of high or medium emotive arousal, to identify some qualitative state, as the frequent use of adjectival suffix “-ful” clearly shows. Adjectives in “-ful” are more descriptive, but they still represent some of the cognitive/emotive effect (arousal) motivating the exclamation and indeed describe and characterize the specific essence of a quale (a thing subjectively perceived/felt) as objectively having what is expressed in the first element. By definition, central adjectives denote qualities referring to a substance/an “object”, but they also inform about the speaker.

  • 8 A corpus search on the COCA does show some instances of WHAT exclamatives using (followed within tw (...)

46The motivation of HOW exclamatives is the qualitative identification and expression of a specificity. This is very different from the need to judge and categorize expressed in WHAT exclamatives. On the one hand, the speaker names a specific and salient quality whereas in the other, he or she identifies common features allowing classification and evaluation. Strange and its synonyms intriguing, curious and fascinating are only used in HOW exclamatives in my corpus8. This is consistent with the typical cognitive state of HOW exclamatives: neutral valence and perplexed/confused cognitive state induced by high emotive arousal.

47In such a cognitive state, the speaker is not able to categorize and barely able to identify; she or he can merely acknowledge the specific un-conventional and un-ascribable quality of what she or he feels/thinks/imagines/perceives. Rather than sanctioning according to pre-established social behavior-relevant categories, he or she emphasizes the impossibility of categorisation. Strange simply does this: acknowledging difference.

48HOW exclamatives typically show high arousal and neutral valence and occur in emotive contexts.

49There is yet another lexical sign of the difference in cognitive motivation for HOW and WHAT exclamatives: the same adjective can have a different meaning in HOW and WHAT exclamative phrases, polysemy resulting from the variation in arousal and valence (all measurable in context).

5. Semantic variation in WH-exclamative phrases

5.1. Irony: the (extraordinary) case of “extraordinary” in HOW and WHAT exclamatives

50There is indeed another aspect, which is a consequence of their social inscription, in which WHAT exclamatives differ from HOW exclamatives: they often display irony. As such, they signal low emotive arousal and, more remarkably, may result in semantic variation (of the adjective or even of head noun), their meaning being more (socially) contextually dependent. “Extraordinary” is used in both (8) and (9):

(9) Silence.
“But what will you do? You’ll go back? You’ll see him?”
That made her sit right up and stare across at me.
What an extraordinary idea!” she said, more coldly than ever. (“Je ne parle pas français”,
(10) In a corner of the garden there were two brimming tubs of water. Three little girls, having thoughtfully taken off their drawers and hung them on a bush, their skirts clasped to their waists, were standing in the tubs and tramping up and down. They screamed, their hair fell over their faces, they splashed one another. But suddenly, the smallest, who had a tub to herself, glanced up and saw who was looking. For a moment she seemed overcome with terror, then clumsily she struggled and strained out of her tub, and still holding her clothes above her waist, “The Englishman! The Englishman!” she shrieked and fled away to hide. Shrieking and screaming the other two followed her. In a moment they were gone; in a moment there was nothing but the two brimming tubs and their little drawers on the bush.
“How--very--extraordinary!” said she. “What made them so frightened? Surely they were much too young to... (“The Man without a Temperament”,

  • 9 “Extraordinary” in (a) and “extraordinary” in (b) would probably be analyzed by C. Paradis [2001: 5 (...)

51“Extraordinary” does not have the same meaning in the two examples. In (9), it means inappropriate, and the extra-ordinariness, the un-conventionality is understood as negative valence. In (10), “extraordinary” means exactly what it means, there is no derivative, contextually-dependent meaning, and it means unusual and strange9. “Very” in (11) intensifies the literal meaning, where the same intensifier intensifies the pejorative meaning in yet another WHAT exclamative:

(11) “The porter?” snapped Josephine. “Why ever the porter? What a very extraordinary idea!” (“The daughters of the Late Colonel”, GP).

52There is a cognitive distance in exclamations (9) and (11) which transforms a usually neutrally or positively-valenced word into a negative one. This results from an enunciative posture typical of social (and upper class) interaction, that of irony. The speaker shows no sincerity and no personal implication, but an appropriate command of social codes. Irony implies a “normative bias”, as D. Wilson and D. Sperber [2012: 131] state:

There is a widely noted normative bias in the uses of irony. The most common use of irony is to point out that situations, events or performances do not live up to some norm-based expectation. Its main use is to criticise or to complain.

53In D Wilson and D. Sperber’s [2012: 134] “echoic account” of verbal irony, the ironic statement “necessarily involves the expression of a certain type of attitude to the attributed thought” (my emphasis).

54HOW exclamatives, motivated by intense arousal and subjective endorsement, do not show a discrepancy between thought and attitude, typically react to some emotive content rather than to some abstract and propositional one. As such, they are less compatible with irony.

5.2. WHAT-exclamative nouns: a context-dependent meaning

55Social behaviors are so deep-rooted and cultural references so implicit that the meaning of words in WHAT exclamatives is sometimes mainly conveyed by the context, as in the apparently non-evaluative head nouns “position” and “picture” in the following examples:

(12) What a picture!
(13) What a position!

56Both head nouns seem devoid of evaluative semantic traits. But the context shows signs of evaluation, either positive or negative.

(14) The drawing-room was full of sweet smelling, silky, rustling ladies and men in black with funny tails on their coats--like beetles. Father was among them, talking very loud, and rattling something in his pocket. “What a picture!” cried the ladies. “Oh, the ducks! Oh, the lambs! Oh, the sweets! Oh, the pets! All the people who couldn’t get at Moon kissed Sun, and a skinny old lady with teeth that clicked said: “Such a serious little poppet,” and rapped him on the head with something hard. (“Sun and Moon”, Bliss)
(15) How did one prove things, how could one? Suppose Kate had stood in front of her and deliberately made a face. Mightn’t she very well have been in pain? Wasn’t it impossible, at any rate, to ask Kate if she was making a face at her? If Kate answered “No” --and, of course, she would say “No” --
what a position! How undignified! (“The Daughters of the Late Colonel”, GP)

57The contexts elucidate the meaning of the nouns: “picture” has positive valence, and “position” negative valence. Both emanate from social interaction entailing social norms. In example (14), the children of a very mundane mother, Sun and Moon, who were noisy and not lovable in the morning, have been “costumed” and turned into perfect inanimate ideal “pictures” for the big evening party. In example (15), Josephine imagines a dialogue with her maid where the maid would prove disrespectful, a position which would be intolerable for this shy, fragile and very formal character.

58Evaluation is a typical cognitive trait of WHAT exclamatives and the “normative bias” described by D. Sperber and D. Wilson gives its specific, context-dependent meaning to the word. If valence is not an explicit semantic component of the noun, it is implicitly present in the social norms described in the context. Apparent neutral valence in WHAT exclamatives can be a “false” neutral. Similarly, apparent positive valence as in “extraordinary” may imply real negative value. Cognitive distance and mastery of social codes and verbal interaction confirm the typical motivation of WHAT exclamatives: low emotive arousal and marked valence. Such a distance is never at work in HOW exclamatives, where words are used for what they really mean. In example (2), a HOW exclamation follows a WHAT phrase: out of the imagined social interaction (internal perception) develops a feeling of humiliation (“How undignified”).


59This paper has presented a psycho-linguistic difference between HOW and WHAT exclamatives in a corpus of Mansfield’s texts. HOW and WHAT work in a micro-system, at the interface of semantics and syntax. HOW exclamatives render intense arousal and neutral valence while WHAT exclamatives, as a linguistic expression of social and normative interaction, typically show medium or low arousal and marked valence. Emotion that motivates HOW exclamatives weakens cognitive processes whereas clear perception of external situations informs appraisal and allows critical distance (such as irony) in WHAT exclamatives.

60Further research must be done in order to prove our intuition that such is the typical meaning and difference between the two exclamatives in the English language in general. The syntax of exclamatives is partly responsible for their specialization: constructed around a head noun, WHAT exclamatives tend to judge and categorize, while the adjective in HOW phrases focuses on qualitative description. The meaning of WH-markers has been studied elsewhere (J. Neveux [2018]), but much more needs to be said about the difference between the words HOW and WHAT.

Top of page



Mansfield Katherine, 1923, Bliss: and Other Stories, London: Constable.

Mansfield Katherine, 1923, The Garden Party: and Other Stories, London: Constable.


Celle Agnès & Lansari Laure (eds.), 2017, Expressing and Describing Surprise, Amsterdam: John Benjamins.

Celle Agnès & Lansari Laure, 2015, « Surprise et exclamation en anglais », in Depraz Nathalie & Serban Claudia (eds.), La surprise à l’épreuve des langues, Paris : Hermann, 79-99.

Ciompi Luc & Panksepp Jaak, 2005, “Energetic effects of emotions on cognition: complementary psychobiological and psychosocial findings”, in Ellis Ralph D. & Newton Natika (eds.), Consciousness and Emotion, Amsterdam-Philadelphia: J. Benjamins Publishing Company, 23-55.

Colibazzi Tiziano et al., 2010, “Neural systems subserving valence and arousal during the experience of induced emotions”, Emotion 10 3, 377-389,

Cotte Pierre, 1996, L’Explication grammaticale de textes anglais, Paris : PUF.

Damasio Antonio, R., 2003, Looking for Spinoza: Joy, Sorrow and the Feeling Brain, Orlando, Austin, New York: Harcourt.

Huddleston Rodney & Pullum Geoffrey, 2002. The Cambridge Grammar of the English Language, Cambridge: Cambridge University Press, 918-923.

Konderak Piotr, 2016, “Between language and consciousness: linguistic qualia, awareness, and cognitive models”, Studies in Logic, Grammar and Rhetoric, The Journal Of University of Bialystok 48.1, 285-302.

Langacker Ronald W., 2009, Investigations in Cognitive Grammar, Berlin: Mouton de Gruyter.

Legallois Dominique & François Jacques, 2012, « Définition et illustration de la notion d’expressivité en linguistique », in Le Querler Nicole, Neveu Franck & Roussel Emmanuelle (eds.), Relations, Connexions, Dépendances : hommage au professeur Claude Guimier, Rennes : Presses universitaires de Rennes, 197-221.

McDaniels Ivy, 2011, “Young Writer, ‘Young Country’: Katherine Mansfield’s Ambivalence towards Colonial New Zealand”, in Atkinson Adam, Hopton Tricia, Mitchell Peta and Stadler Jane (eds.), Pockets of Change: Adaptation and Cultural Transition, New York: Lexington Books, 71-80.

Michaelis Laura, 2001, “Exclamative Constructions”, in Haspelmath Martin, König Ekkehard, Österreicher Wulf & Raible Wolfgang (eds.), Language Typology and Language Universals: An International Handbook, Berlin, New York: Mouton de Gruyter, 1038–1050.

Mignot Elise, 2014, « Le sens des lexèmes adjectivaux dans les noms construits de l’anglais : une réflexion sur le nom en tant que partie du discours », in Daval René, Frath Pierre, Hilgert Emilia & Palma Silvia (eds.), Res per Nomen IV, Les théories du sens et de la référence – Hommage à Georges Kleiber, Reims : Epure, Éditions et Presses Universitaires de Reims, 237-252.

Neveux Julie, 2018, “Grammar and feelings: a study of WH- exclamatives in Katherine Mansfield’s short stories”, in Sorlin Sandrine (ed.), ESA n°12, La Société de Stylistique Anglaise (1978-2018): 40 ans de style, 192-222.

Nouwen Rick & Chernilovskaya Anna, 2015, “Two types of exclamatives”, Linguistic Variation 15.2, 201-224.

Paradis Carita, 2001, “Adjectives and boundedness”, Cognitive Linguistics, 12, 47-64.

Posner Jonathan E. et al., 2009, “The neurophysiological bases of emotion: An fMRI study of the affective circumplex using emotion-denoting words”, Human brain mapping 30.3, 883-95,

Russell James A., “A Circumplex Model of Affect”, in Journal of Personality and Social Psychology, 1980, Vol.39, No. 6, 1161-1178.

Schmid Hans-Jörg, 2000, English Abstract Nouns as Conceptual Shells: From Corpus to Cognition, Topics in English Linguistics 34, Berlin, New York: Mouton de Gruyter.

Sperber Dan & Wilson Deirdre, 2012, “Explaining Irony”, Meaning and Relevance, Chapter 6, Cambridge University Press, 123-145.

Tappolet Christine, 2016, Emotions, Values and Agency, Oxford: Oxford University Press.

Zanutti Raffaella & Portner Paul, 2003, “Exclamative clauses: At the syntax-semantics interface”, Language 79.1, 39-81.

Top of page


1 Expressing emotions differs from describing them, as Celle and Lansari [2017: 4] emphasize. This paper focuses only on the linguistic expression of emotions and perceptions.

2 I have exclusively compared HOW and WHAT exclamatives when they worked as predicative complement, for the systematic (semantic and cognitive) comparison of the two structures to be more rigorous. I have thus not taken into account (for this paper) occurrences where HOW functions as an adverbial exclamative, as in “how beautifully he sang”, or “how he sang” (the change in function impeding systematic comparison). But cases of adverbial HOW typically show a qualitative and emotive implication of the speaker.

3 The description of WHAT nominal phrases and HOW adjectival phrases as “exclamative phrases” is taken from R. Huddleston & G.  Pullum [2002: 919].

4 “High cortical activity generally tends to inhibit subcortical emotional systems” (L. Ciombi and J. Panksepp [2005: 29]).

5 Not only do literary corpora provide much larger contexts for utterances than “authentic” ones such as BNC and COCA, but a cognitive approach might all the more be undertaken as Mansfield’s aesthetic project was specifically to address emotions (and their linguistic expression). In this approach, characters will be treated as subjects of a cognitive experiment (one comparison with “real” corpora will however be provided).

6 The examples corresponding to this part are given and commented upon in part 3.

7 The referent of the subject may be the speaker herself/himself, but it remains in these cases a critical judgment about one’s own social behaviour, and hardly addresses emotions. Extraposed WHAT constructions confirm this, the semantic subject developed in the subordinate clause always denoting a recent past behavior (“What an extraordinary thing to come all this way to tell me” (“The Daughters of the Late Colonel”, GP,) “What a mistake to have come!” “Revelations”, Bliss) “What a good thing you’ve got my cape” (“The Man without a Temperament”, Bliss).

8 A corpus search on the COCA does show some instances of WHAT exclamatives using (followed within two words by) “strange” but they are less numerous than HOW (196 WHAT exclamatives versus 541 HOW exclamatives).

9 “Extraordinary” in (a) and “extraordinary” in (b) would probably be analyzed by C. Paradis [2001: 5-6] as respectively “extreme” adjective (modified by totality modifiers) and “scalar” adjective (modified by scalar modifier like “very” in our example).

Top of page

List of illustrations

Title Figure 1: Valence in all exclamatives
File image/jpeg, 28k
Title Figure 2: Neutral valence: HOW
File image/jpeg, 28k
Title Figure 3: High arousal: HOW
File image/jpeg, 28k
Title Figure 4: Low arousal: HOW
File image/jpeg, 28k
Title Figure 5: Adjectives in WHAT exclamatives
File image/jpeg, 28k
Title Figure 6: Adjectives in HOW exclamatives
File image/jpeg, 26k
Top of page


Electronic reference

Julie Neveux, « Valence and arousal in WHAT and HOW exclamatives: cognitive simplification versus emotive implication », Lexis [Online], 13 | 2019, Online since 14 March 2019, connection on 25 March 2019. URL : ; DOI : 10.4000/lexis.3335

Top of page
  • Logo Centre d’études linguistiques
  • Logo Université Jean Moulin – Lyon 3
  • OpenEdition Journals