Expressing and Describing Surprise [1 ed.] 9789027265081, 9789027242808

Among emotions, surprise has been extensively studied in psychology. In linguistics, surprise, like other emotions, has

197 64 15MB

English Pages 254 Year 2017

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Expressing and Describing Surprise [1 ed.]
 9789027265081, 9789027242808

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

s.) d e e ( l l e ari C s s n è a Agn Laure L and

B E N J A M I N S C U R R E N T TO P I C S

d g an ise essin urpr Expr ibing S r Desc



Expressing and Describing Surprise

Benjamins Current Topics issn 1874-0081 Special issues of established journals tend to circulate within the orbit of the subscribers of those journals. For the Benjamins Current Topics series a number of special issues of various journals have been selected containing salient topics of research with the aim of finding new audiences for topically interesting material, bringing such material to a wider readership in book format. For an overview of all books published in this series, please see http://benjamins.com/catalog/bct

Volume 92 Expressing and Describing Surprise Edited by Agnès Celle and Laure Lansari These materials were previously published in Review of Cognitive Linguistics 13:2 (2015).

Expressing and Describing Surprise Edited by

Agnès Celle Laure Lansari Université Paris Diderot

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

doi 10.1075/bct.92 Cataloging-in-Publication Data available from Library of Congress: lccn 2017012837 (print) / 2017038071 (e-book) isbn 978 90 272 4280 8 (Hb) isbn 978 90 272 6508 1 (e-book)

© 2017 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Company · https://benjamins.com

Table of contents

Introduction Agnès Celle and Laure Lansari

1

Surprise as a conceptual category Zoltán Kövecses

7

The complex, language-specific semantics of “surprise” Cliff Goddard

27

Grammatical evidentiality and the unprepared mind Tyler Peterson

51

Operationalizing mirativity: A usage-based quantitative study of constructional construal in English Karolina Krawczak and Dylan Glynn The computer-mediated expression of surprise: A corpus analysis of chats by English and Italian native speakers and Italian learners of English Laura Ascone Surprise routines in scientific writing: A study of French social science articles  Agnès Tutin

91

121

153

Surprise in the GRID Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

173

Surprise and human-agent interactions Chloé Clavel

197

Expressing and describing surprise Agnès Celle, Anne Jugnet, Laure Lansari and Emilie L’Hôte

215

Index

245

Introduction Agnès Celle and Laure Lansari

Université Paris Diderot, Sorbonne Paris Cité, CLILLAC-ARP EA 3967

This paper is an introduction to this volume on the description and expression of surprise. In line with the ANR-funded project it is part of,1 this volume aims to bridge the gap between emotion, cognition and linguistics. It stresses the unique status of surprise among emotions, uncovering the link between surprise and epistemicity. Drawing upon the traditional distinction between the description and expression of emotions made by scholars of different persuasions, it argues for a better definition of expressivity as a category in its own right, possibly independent of the expression of spontaneous emotion. This paper also provides a presentation of the nine contributions the volume is made up of. Keywords: surprise, emotion, cognition, expression, description, expressivity, epistemicity, mirativity, evidentiality

As an emotion, surprise has been extensively studied by psychologists. In linguistics, the focus has been on the syntactic classifications of surprise verbs and lexemes (Levin, 1993; Mathieu, 2000; Novakova, Goossens, & Melnikova, 2012), without any notable attempt being made to correlate lexico-syntactic patterns to the emotional dimension of surprise. This volume seeks to bridge the gap between emotion, cognition and linguistics. It focuses on the linguistic dimensions of surprise and relates them both to emotions at large and to underlying cognitive processing. The nine papers of the volume address surprise-related issues from different angles. The papers reflect a variety of theoretical frameworks, methodologies and goals. We believe that this variety of approaches is crucial to the study of surprise, a multi-faceted and complex phenomenon that manifests itself through language but also through prosodic features, gestures, etc., and may be apprehended as a general onomasiological category and/or through linguistic markers. 1. We acknowledge the support of the French National Agency for Research through the Em­philine Project, Emotion(s) Cognition Comportement ANR Programme n°ANR-­11-­EMCO-005. doi 10.1075/bct.92.001int © 2017 John Benjamins Publishing Company

2

Agnès Celle and Laure Lansari

The first three papers seek to enrich our general understanding of surprise from a cognitive viewpoint. In his article on the figurative language of surprise in English, Zoltan Kövecses sketches a general cognitive scenario for surprise that might explain the different cognitive stages of surprise, from cause to effects as manifested in the state of surprise. Adopting a lexical approach to emotion concepts, Zoltan Kövecses argues that surprise involves a loss of control akin to an epistemic failure. Surprise is not construed as a substance, which might explain why, unlike other emotions, it is associated only with a few metonymies and conceptual metaphors. Within the framework of Natural Semantic Metalanguage, Cliff Goddard proposes to study surprise lexemes in English and Malay in order to define a non-Anglocentric semantic typology of “surprise-related” notions in the languages of the world. He puts forward the following configuration to capture the semantic primes of surprise in a cross-linguistic perspective: “this someone thought like this at this time: ‘I didn’t know before that it will / can be like this’”. Using schema theory, Tyler Peterson bridges the gap between the linguistic encoding of surprise and the psychological and cognitive dimensions of surprise. He adopts a typological viewpoint and analyses surprise within the category of “mirativity”, which he redefines as the cognitive process of surprise. He shows that surprise and unexpectedness are implicated by evidentiality. Indeed, the use of an inferential evidential when the speaker has direct knowledge of an event indicates that the speaker is not prepared for that event. In addition, he suggests that non-congruent marking connects to the cross-linguistic category of control. The next three papers are language-specific case-studies of surprise markers, relying on naturally-occurring data extracted from different genres. The first two papers explore the expression of surprise in computer-mediated communication. Dylan Glynn and Karolina Krawczak focus on three surprise “constructions” in English (what a + noun, what the + noun, to my + noun) collected in blogs in order to investigate how they encode mirativity. Examining surprise markers in a similar genre – chats – Laura Ascone shows that surprise is expressed differently by native English speakers, native Italian speakers and Italian learner speakers of English, which reveals complex links between language, cognition and culture. Agnès Tutin’s paper deals with the surprise lexicon in French in a corpus of scientific texts and shows that, in this genre, surprise lexemes may serve the author’s rhetorical strategies. The next two papers relate the expression of surprise to psychology and pave the way for applications in affective computing. Cristina Soriano, Johnny Fontaine and Klaus Scherer analyse surprise in the framework of the GRID project, which aims to define the semantic profile of emotion words across languages and cultures. Surprise appears to be a cognitive operation rather than a pure emotion. Four parameters are shown to be necessary for the description of the emotion

Introduction 3

space: valence, power, arousal and novelty. It is novelty that differentiates surprise from other emotions. As a cross-cultural concept, surprise is argued to be neutral on valence and power, but high on arousal, and very high on novelty. Unlike other emotions, it has an appraisal and expressive profile. Lastly, Chloé Clavel’s contribution takes us beyond the realm of linguistics by showing the role of surprise in human-­agent interaction. In order to improve the naturalness of human-­computer interaction, socio-emotional components need to be integrated. She focuses on the computational models and on the methods used to detect the user’s surprise through language and speech processing. Symmetrically, Embodied Conversational Agents need to display socio-emotional attitudes, including surprise, in an appropriate way. The role of surprise is explored in more detail in a dialogue between museum visitors and a virtual agent. The last paper by Agnès Celle, Anne Jugnet, Laure Lansari and Emilie L’Hôte re-examines the well-known distinction between expression and description of surprise. The authors show that surprise is expressed by mirative utterances and particular syntactic constructions, while it is described using surprise lexemes. Relying on a qualitative corpus study of the noun surprise and the adjective surprised in English, they point out that the use of the surprise lexicon is not linked to truly emotional reactions but rather to argumentative strategies highlighting the epistemic nature of surprise. All nine contributions converge in highlighting: – The unique status of surprise (contra Ekman, 2003) as both a cognitive operation and a component of emotions, and hence the need to put forward an adequate model accounting for the characteristics of surprise, notably the absence of valence – as already noted by Ortony (Ortony, Clore, & Collins, 1988) – which has important implications that have gone unnoticed so far, for example on the use of surprise in text and on the implementation of Embodied Conversational Agents in affective computing. – The essential link between surprise and epistemicity, making surprise a cognitive rather than strictly emotional category: surprise can no longer be viewed as a mere physical startle in reaction to a stimulus, but comes to be redefined as a complex epistemic stance taking into account the speaker’s expectations and as a social phenomenon involving possible negotiations between speaker and hearer (loss of control, denial of surprise, etc.). One major implication is that surprise may be used as a pragmatic device – as part of rhetorical strategies, emotionally-laden narratives where surprise may be reenacted, or ironic statements, confirming the findings of recent works (Celle, in press; Celle & Lansari, 2014, 2015; Novakova & Sorba, 2013).

4

Agnès Celle and Laure Lansari

– The difficulties inherent in the linguistic study of surprise and in the detection of surprise markers. Surprise can indeed be viewed as an onomasiological category with no one-to-one correspondence with specific linguistic forms. Drawing on the now widely-accepted distinction between description of surprise and expression of surprise, established by many authors albeit in slightly different forms (e.g. Marty, 1976; Bally, 1926; Searle, 1969 and more recently, Rett, 2008, 2011), several contributions in this volume suggest that the description of surprise is associated with surprise lexemes and figurative language whereas the expression of surprise is linked to disfluency markers such as exclamations, interjections, etc. This gives credence to a clear-cut division between markers describing surprise and those expressing surprise. Interestingly, it is also suggested that this dichotomy is sometimes insufficient to analyse the aforementioned pragmatic strategies: reenactment and the staging of surprise may well be better captured in terms of expressivity – viewed as a staging strategy aiming at drawing the hearer’s attention (see Legallois & François, 2012). As such, expressivity does not necessarily imply the expression of a spontaneous emotional reaction.

References Bally, C. (1926). Le langage et la vie. Paris: Payot. Celle, A. (in press). Surprise in storytelling. In R. Nita & F. Valetopoulos (Eds.), Expressing emotions in corpora. Peter Lang. Celle, A., & Lansari, L. (2014). ‘Are you surprised?’/‘I’m not surprised’: Surprise as an argumentation tool in verbal interaction. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discours/Emotions in discourse (pp. 267–279). Bern/Frankfurt: Peter Lang. Celle, A., & Lansari, L. (2015). Surprise et exclamation en anglais. In N. Depraz & C. Serban (Eds.), La surprise à l’épreuve des langues (pp. 79–99). Paris: Hermann. Ekman, P. (2003). Emotions revealed: Recognizing faces and feelings to improve communication and emotional life. New York: Times Books. Legallois, D., & François, J. (2012). Définition et illustration de la notion d’expressivité en linguistique. In C. Guimier, N. Le Querler, F. Neveu, & F. Roussel (Eds.), Relations, connexions, dépendances: Hommage Au professeur Claude Guimier (pp. 197–221). Rennes: Presses Universitaires de Rennes. Levin, B. (1993). English verb classes and alternations: A preliminary investigation. Chicago: University of Chicago Press. Marty, A. (1976) [1908]. Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprach­philosophie. Hildesheim/New York: Olms. Mathieu, Y. Y. (2000). Les verbes de sentiment: De l’analyse linguistique au traitement automatique. Paris: CNRS Éditions.

Introduction 5

Novakova, I., Goossens, V., & Melnikova, E. (2012). Associations sémantiques et syntaxiques spécifiques; Sur l’exemple du lexique émotionnel des champs de surprise et de déception. In F. Neveu, V. Muni Toke, & P. Blumenthal (Eds.), 3e Congrès mondial de linguistique française: Lyon, 4–7 juillet 2012 (pp. 1017–1029). Les Ulis, France: EDP Sciences. Novakova, I., & Sorba, J. (2013). Stupéfier et jalouser dans les séquences textuelles journalistiques: Quel profil discursif pour quelle stratégie argumentative? Le Discours et la Langue: Revue de linguistique française et d’analyse du discours, 4(1), 203–220. Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotions. Cambridge/ New York: Cambridge University Press.  doi: 10.1017/CBO9780511571299 Rett, J. (2008). A degree account of exclamatives. In Proceedings of SALT 18 (pp. 601–618). http://elanguage.net/journals/salt/article/download/18.601/1921. Rett, J. (2011). Exclamatives, degrees and speech Acts. Linguistics and Philosophy, 34(5), 411– 442.  doi: 10.1007/s10988-011-9103-8 Searle, J. (1969). Speech acts: An essay in the philosophy of language. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9781139173438

Surprise as a conceptual category* Zoltán Kövecses

Eötvös Loránd University, Budapest

In this paper, I examine the concept of surprise from a cognitive linguistic perspective. As previous studies indicate, surprise is a not-quite-prototypical emotion category. My focus will be on the structure and content of surprise as an emotion category, as this can be revealed on the basis of the language speakers of English use to talk about it. As regards methodology, I will follow my earlier work and employ a “lexical approach” to emotion concepts (see, e.g., Kövecses, 1986, 1990, 2000) to explore the language-based folk model of surprise in English. I will investigate the conceptual metaphors and metonymies associated with surprise and will propose a cognitive model for the emotion. It is hoped that this methodology enables us to see why surprise is not a prototypical emotion concept on a par with, for example, anger or fear. Keywords: surprise, emotion concepts, metaphor, metonymy, loss of control, lexical approach, corpus linguistics

1. Introduction In a number of publications (e.g., Kövecses, 1986, 1990, 2000, 2008), I proposed a cognitive linguistic framework for the study of emotion language and the conceptual structure of emotions. In the cognitive linguistic view, language is taken to be a good guide to uncovering the content and structure of our conceptual system (see, e.g., Langacker, 1987; Lakoff, 1987), including that portion of it that is related to emotions concepts. As indicated by previous studies in this approach, the language used to talk about the emotions in English and many other languages is highly and mostly figurative, that is, metaphoric and metonymic (see, e.g., * I am grateful to Agnès Celle for inviting me to the conference on surprise and to teach at the University of Paris Diderot for a month in 2014. I want to thank the two anonymous reviewers for their detailed comments on a previous version of this paper. doi 10.1075/bct.92.01kov © 2017 John Benjamins Publishing Company

8

Zoltán Kövecses

Athanasiadou & Tabakowska, 1998; Niemeier & Dirven, 1997; Palmer & Occhi, 1999; Russel, Fernández-­Dols, Manstead, & Wellenkamp, 1995; Soriano, 2005; Wilson, 2012; Yu, 1998, among others). Thus, by studying the metaphors and metonymies, we can identify a certain type of generic conceptual structure for the concept of emotion, as well as more specific conceptual structures for particular emotion concepts. These generic and specific conceptual structures that have been uncovered function as folk theories of emotion and particular emotions. This means that the conceptual structures that emerge from the metaphors and metonymies are taken to be language-based folk models, and not scientific models of emotion. (On the distinction, see, e.g., Fontaine, Scherer, & Soriano, 2013; Holland & Quinn, 1987.) To establish the latter, very different methodologies would be needed, as, for instance, used in neuroscience (Damasio, 1994; Le Doux, 1996), behavioral sciences (e.g, Ekman & Friesen, 1975; Ekman, Levenson, & Friesen, 1983), social sciences (e.g., work in anthropology, such as Lutz, 1988; work in psychology, such as Frijda, 1986; Manstead, Frijda, & Fischer, 2004). Folk theories, as opposed to expert theories, are naïve, nonscientific understandings of the world by lay people for their everyday purposes (see, e.g., Holland & Quinn, 1987). My goal in the paper is to offer a particular language-based folk theory, or model, for the emotion of surprise, as based on the figurative language used to talk about this emotion and to see how this folk model is different from and similar to those of other, more prototypical emotions. The present paper reflects a cognitive linguistic approach to the emotion concept of surprise,1 and in this regard it is different from other approaches that are used in linguistics. My main emphasis is on exploring how people might possibly conceptualize surprise, and not on the syntactic features of surprise as a word or lexeme or on its discourse-­ pragmatic aspects, as can be found in other studies, such as Celle and Lansari (2014). I take these two different linguistic approaches, as well as several others (see the papers in the present volume), as equally legitimate and necessary in the study of surprise. Indeed, I believe that, ideally at least, we would need all of them in a “complete” linguistic-conceptual-pragmatic account of surprise. I will proceed as follows. To begin, I discuss some methodological issues in the study of metaphor. This will be followed by an analysis of the polysemy of the lexeme surprise. Next, I examine the metonymies of surprise. Then, I turn to the metaphors of surprise and their significance in the conceptualization of this 1. I will use small capitals (such as surprise) to indicate conceptual entities (concepts, domains and their elements, models, frames, prototypes) and distinguish conceptual entities from linguistic expressions, which are indicated by italics (such as surprise), and from objects in the world, which are indicated by normal lower-case letters (such as surprise or emotion).



Surprise as a conceptual category

emotion. Finally, I compare the resulting folk theory of surprise with that of prototypical emotions. The paper closes with some concluding remarks. 2. Methodological issues The methodology I am using is based on a distinction between the expression of emotion and the description of emotion (for details on the distinction, see Kövecses, 2000). For example, the emotion of surprise can be expressed in a variety of ways, including interjections, such as “wow!”. This can be done in the immediate presence of a triggering event by the person experiencing the emotion. As a result, the expression of emotion involves the speaker coinciding with the experiencer of the emotion and the “here and now” of the situation in which it is experienced. By contrast, the description of an emotion involves a speaker who can be both different than and the same as the experiencer and the speech act of description (or assertion) can refer to either the past, the future, or the present, as in “She was really surprised.” Sometimes the expression of an emotion can merge with its description, as in “I love you” (For further clarification between the two, see also Celle & Lansari, 2014.) In the paper, attention is paid only to cases that function as descriptions (assertions). In the study of everyday concepts, such as emotion concepts, there are many ways in which language-based folk theories can be arrived at. While these share the idea that they use language about a particular domain as their data, what distinguishes them from one another is how they gather and analyze the language data (see, e.g., Quinn, 2005). The data may be derived from interviews, questionnaires, corpora of various kinds, collections of conventionalized ways of speaking about a topic, and others. In all of them, based on the language data the researcher generates certain hypotheses concerning the conceptual content and structure of the respective topic corresponding to a concept, or conceptual category, in the mind. This is what we can take to be the language-based folk theory of the topic in question. The researcher creates the hypotheses on the assumption that language is a reflection of our conceptual system; that it is, indeed, a “window to the mind” (see, e.g., Handl & Schmid, 2011). Without making this assumption, the language sciences could not be seen as contributing anything to the study of the conceptual system. At the same time, the ideas, or concepts, that is, the folk theories, that they propose based on the assumption can only be regarded as hypotheses. The researchers using the various methodologies to derive and analyze language are not in a position to claim psychological reality for their ideas – that is, the proposed folk theories. They can only offer hypotheses that need to be checked by other scientists who are in the business of studying and validating the psychological

9

10

Zoltán Kövecses

reality of the language-based folk theories by means of conducting experiments of various sorts (e.g., psycholinguists, cognitive psychologists, etc.). In this regard, there is no difference between the researchers who work with language as their main data; they all make a leap from language to thought (i.e., ideas, concepts, folk theories), and therefore their suggestions concerning conceptual structure can only be hypothetical in nature. In the paper, I will employ what can be called the “lexical method,” or the “lexical approach” (for a first description of this, see Kövecses, 1986). The researcher using this method searches for various lexical items or other types of information that are related to the general topic, or concept, under investigation (such as particular emotions indicated by particular lexemes: e.g., anger, fear, surprise). These include synonyms, antonyms, related words, various idioms and phrases, collocations, and, importantly, even the definitions of the lexemes. The most likely sources for these are dictionaries: monolingual and bilingual dictionaries, thesauri, collocation dictionaries, idiom dictionaries of various sorts, and, in general, any collections of words and phrases related to a concept. Very importantly, researchers prefer and tend to use dictionaries that offer example sentences and usage notes (about register, frequency, provenance, etc.) for the linguistic expressions (words, idioms, collocations, etc.) that they contain for the lexeme in question. Clearly, it is unlikely that all of this information can be found in a single dictionary. As a result, often, several different dictionaries of various kinds must be consulted before one obtains all, or at least most or much, of the lexical information that pertains to the lexeme we are interested in. Moreover, dictionaries that are based on large corpora are vastly preferred to ones that are not. As this description of the lexical approach indicates, the various lexical items that belong to a particular concept, or, as it is commonly referred to in CMT, a domain, are dominantly types, not tokens. The linguistic expressions that are identified by the dictionaries represent lexemes. In this method, the actual occurrences of the types (as lexemes), that is, the tokens, are only embedded in discourse when the dictionary provides example sentences (often taken from corpora) for the types. This way, it is not possible to gage the actual frequency of the tokens as based on the dictionary data, which is a drawback of the method. The lexical approach is based on a further assumption. It is that, since it deals with types, that is, with linguistic expressions that have become lexicalized, the types provided by the dictionaries represent the most conventionalized linguistic expressions of a language related to a domain. Since higher degrees of conventionalization can be expected to go together with greater conceptual centrality, given the high degree of conventionalization, or entrenchment, of the types, the researcher can make predictions (hypotheses) concerning the conceptually central cases of concepts, or conceptual categories, at the “supraindividual level.” The



Surprise as a conceptual category

surpraindividual level is the level where we are dealing with decontextualized language (the conventionalized types), as opposed to the “individual level,” where we are dealing with contextualized linguistic usage (of the types) by individuals that results in the tokens of use. (For more on this notion, see Kövecses, 2002, 2010.) At this point, we can note that corpus linguistics, as another method, was proposed in the study of language, including metaphor research, to correct the deficiencies of work at the supraindiviual level and, specifically, to draw attention to contextualized linguistic usage in real discourse (i.e., the individual level) (see, for example, Deignan, 2005; Semino, 2008; Stefanowitsch, 2006). Does this mean that analysis at the supraindividual level have no place in metaphor research? I believe not. Let us examine the reasons for this. In the lexical approach to metaphor, after we have collected all, or most, of the lexical expressions related to a particular general lexeme (such as surprise), we identify the expressions as literal, metaphoric or metonymic. This can be done by means of a variety of methods, of which the best known one is the metaphor identification procedure (MIP) worked out by the Pragglejaz Group (2007). After the identification of the linguistic metaphorical expressions, we group the expressions into thematic clusters. Each particular theme, or topic, around which the metaphorical expressions are grouped constitutes the source domain (typically, a concrete concept, b), whereas the lexeme that serves as our general search term (e.g., surprise in the present case) constitutes the target domain (typically, an abstract concept, a). As noted above, it should be borne in mind that such a and b pairings of concepts (i.e., conceptual metaphors) can only be suggested as hypotheses to be validated. The testing of their psychological reality is a job for psycholinguists and other experimental scientists. Given such hypothetical pairings of conceptual categories, the next step is to see how the two concepts are connected. More precisely, we check which elements of the source are connected to which elements of the target domain. One or several metaphorical expressions may be related to a particular element of the source, and this element has its counterpart element in the target (as in the “heat” element in hot fluid, which can be represented by several metaphorical expressions relating to anger, and which corresponds to the “intensity” element in anger). Such assumed (hypothetical) correspondences between the source and the target are called “mappings.” There are several additional issues concerning metaphorical expression types that can be examined at the supraindividual level (e.g., the metaphorical entailments of the source in relation to the target), but this much will suffice to demonstrate in a nutshell how the lexical approach works. We can see that the approach can uncover the most conventionalized metaphorical linguistic expressions related to a target domain (i.e., the types) on the basis of which researchers

11

12

Zoltán Kövecses

can hypothesize the existence of systematic conceptual correspondences between two domains – the conceptual metaphors. The conceptual metaphors can be assumed to be shared by speakers of a language who have acquired the metaphorical expression types relating to a target. Experimental studies indicate that this is indeed the case (see, e.g., Gibbs, 1994, 2006; Gibbs & Colston, 2012). Thus the lexical approach can reveal a considerable portion of the shared metaphorical conceptual system in a linguistic community, though not the metaphorical conceptual systems that particular individual speakers possess. Now, in view of this characterization, corpus linguists suggest that there are certain problems with the lexical approach. In general, their critique amounts to the claim that the lexical approach does not work for the individual level of metaphor analysis. This criticism is valid. But it should be kept in mind that the lexical approach was not designed to capture contextual variation in the use of metaphors and the individually variable metaphorical conceptual systems of individual speakers (on these issues, see Kövecses, 2005, 2010, 2015). More specific criticism includes, first, that the lexical approach does not make it possible to find all the metaphorical expressions related to a target domain (Stefanowitsch, 2006); second, that it misses ambiguous cases of usage between literal, metaphoric or metonymic interpretations (Deignan, 2005; Goossens, 1990); third, that the mappings are less systematic than what they are made out to be by practitioners of the lexical approach (Deignan, 2005); fourth, that there are frequency differences between the uses of the tokens associated with a target domain (Deignan, 2005; Stefanowitsch, 2006); fifth, that metaphors commonly arise from context, and not from predetermined conceptual mappings in our heads (Cameron, 2003; Gibbs, 1999; Semino, 2008); and several others. I believe all of these claims are correct and justifiable. At the same time, they deal with metaphor at another level of analysis than the lexical approach: the level of the particular discourse where individual speakers use contextualized tokens of metaphorical expression types – not the supraindividual level where there are no individual speakers and no tokens, only types of metaphorical expressions without any context. (For an attempt to unify the two kinds of metaphor analysis, see Kövecses, 2005, 2015.) In conclusion to this methodological discussion, I suggest that the linguistic study of metaphor depends, in part, on our goals: whether we want to characterize how individual speakers use metaphors in context (corpus linguistics) or whether we are interested in making hypotheses concerning our shared metaphorical conceptual system (lexical approach). The former can be called the “bottom-up” approach, while the latter the “top-down” one (see Kövecses, 2011). In my view, both are legitimate enterprises. A complete account of metaphor requires both modes of analysis and explanation.



Surprise as a conceptual category

In the rest of the paper, I examine the concept of surprise by means of the lexical approach. 3. Surprise as a polysemous category The lexeme surprise has several distinct but related senses as a noun and verb. The most common senses as a noun provided by dictionaries include: an unexpected or astonishing event, fact, or thing. a feeling of mild astonishment or shock caused by something unexpected. (http://www.oxforddictionaries.com/us/definition/american_english/ surprise)

Of the two, the former indicates the cause of surprise, whereas the latter the emotion of surprise itself. In other words, the word form surprise denotes both the emotion (the latter) and the cause of the emotion (the former). As a verb, it has the meaning: (of something unexpected) cause (someone) to feel mild astonishment or shock. (http://www.oxforddictionaries.com/us/definition/american_english/ surprise)

We can interpret this as the process of causation, that is, as causing. Here we have the sense of cause emotion. Thus the different senses of the lexeme appear to provide for, or establish, a schematic mini frame for surprise: cause (of emotion) – causes – emotion

We can think of this as the schematic prototype of the conceptual category surprise. Given the prototype, we can see that surprise shares a great deal with other emotions: It shares with them a causal structure that is responsible for the onset of various other emotions as well. The cause of emotion is a well-known component of emotions in the literature on emotions (cf., e.g., Fontaine et al., 2013; Manstead et al., 2004), where it is commonly referred to as “antecedent events” or “emotion-eliciting situation.” The interesting question is: What is it that distinguishes surprise from other emotion concepts, such as anger, joy, fear, etc.? I will come back to this issue later in the paper. We can raise another question that concerns the centrality of the word surprise – a question that is different from the issue of what the prototype of the

13

14

Zoltán Kövecses

conceptual category of surprise is. We saw that all three senses of the word are designated by the form surprise. We can ask which sense is the prototypical, or central, one. This question is similar to John Austin’s question in relation to the word healthy that has at least three distinct senses observable in the three expressions a healthy body, a healthy exercise, and a healthy complexion (discussed by Lakoff, 1987). As Lakoff (1987) points out, it is the sense in a healthy body that is central. The other two are derived from it by metonymy: a state for the cause of the state (healthy exercise) and a state for the effect of that state (healthy complexion). In other words, given a number of senses for the word healthy, the one from which the others can be systematically derived is the central one. What is central sense of the word surprise? Is it the cause of emotion sense, the causing emotion sense or the emotion sense? In order to be able to answer the question in a satisfactory way, we need to observe some similarities between the case of healthy and that of surprise. Just like the adjective healthy, in a healthy exercise, the noun surprise can also indicate the quality of being able to cause surprise, as in a surprise attack. In such cases, the noun surprise is used as a modifier. As regards the conceptual structure of this construction, similar to healthy, it is also based on metonymy: the state for the quality of an event or thing that causes that state. In other words, the noun surprise indicating the emotional state of surprise is metonymically used for the quality of an event (attack) that causes or results in that state (i.e., surprise). This is essentially another way of characterizing the cause of emotion: an event or thing that has the quality to cause surprise simply amounts to the cause of surprise. Another parallel between the cases of healthy and surprise is that, just like healthy, the word surprise can also describe the result of the emotion. This is also based on metonymy: a state for the result / effect of that state. A healthy complexion indicates the result or effect of a healthy body. In the same way, the word surprise can also indicate the effect of the surprise as an emotion. When this happens, the adjective (from the past participial form of the verb surprise) surprised, is used, as in a surprised look. Taking into account the result or effect of surprise as an emotion in the previous paragraph gives us a more complete characterization of the conceptual structure of surprise: cause (of emotion) – causes – emotion | effect of emotion

This schematic frame shares a large portion of the frames associated with prototypical emotions (see Kövecses, 1990, 2000): An event or thing that has the



Surprise as a conceptual category

quality of causing surprise (given the appropriate situation) causes the emotion of surprise, and the emotion causes the emotional self (who is now in the state of surprise) to produce certain effects or responses. 4. Metonymies of surprise What precisely are the effects or responses produced by surprise? We can begin to investigate this issue if we look at the words and idioms that are provided by online dictionaries as synonyms of surprise in its various senses (such as www. learnersdictionary.com, http://www.merriam-webster.com, http://www.thefreedictionary.com, http://www.collinsdictionary.com, etc.). Some of the responses appear to be physical ones, such as eye-opener. Opening the eyes wide is one expressive response surprised people produce. This is probably what the previous example a surprised look might mean. Many, or maybe most, of them, however, are some mental or cognitive response, such as being speechless. Here is a list of the effects of surprise mentioned by dictionaries and thesauri above, as well as additional ones, such as http://www.oxforddictionaries.com: (1) PHYSICAL EFFECTS: eye-opener; leave open-mouthed; take someone’s breath away; breathtaking; take aback; shake up; bowl over; floor; startled; staggered

These are all physical responses of various kinds: physiological, expressive, and behavioral (see Kövecses, 2000). They can all metonymically indicate surprise. The general metonymy for such cases is: the physical effects of an emotion for the emotion (see Lakoff & Kövecses, 1987). Each of them could be further specified by a more specific case of metonymy, such as eyes opening wide for surprise or the mouth opening for surprise. Many of these are likely to be universal, but some of them may be culture-specific. In addition to such highly conventional cases of physical responses, there are also less entrenched cases – less entrenched both linguistically and conceptually. One of these is offered by one of the dictionaries: (2) I was so surprised that I dropped it (http://www.oxforddictionaries.com/definition/american_english-thesaurus/ surprise)

It is linguistically less entrenched because the word drop is not one of the conventionalized linguistic expression types in dictionaries, and it is less entrenched

15

16

Zoltán Kövecses

conceptually because dropping something, unlike many of the other cases mentioned previously, is not specifically associated with surprise as a response. The number of such involuntary occurrences is likely to be infinite and culturally extremely variable. It is unpredictable which culture uses which event as a response to the emotion; only empirical research could decide. However, the occurrences (responses) can be understood metonymically to denote surprise or a high degree of surprise. But many, or maybe even most, of the effects (or responses) produced by surprise are mental, or cognitive: (3) MENTAL EFFECTS: speechless; stupefy; mind-blowing; dumbfound; daze; astound; shock; astonish; amaze; stun; flabbergast; dumbstruck; thunderstruck; confounded; flummoxed; turn up / one for the books (http://www.oxforddictionaries.com)

The dictionary definitions of all of the words above contain, one way or another, the element of surprise. They denote a particular mental response to the experience of surprise. Here again, a generic-level metonymy characterizes these mental responses: mental responses of surprise for surprise. In this case as well, the particular responses belong to more specific metonymies, such as inabiltiy to speak for surprise. Perhaps the most common specific metonymy involves a sudden unpleasant and upsetting feeling, as in shock or stun, where the specific-level metonymy could be something like an upsetting feeling for surprise, and in others it can be inability to think clearly for surprise, as in stupefied, dumbstruck, confounded, dazed, and so on. Some of the responses above can be considered to be both physical and mental effects of surprise. For example, the response of being speechless is probably a mixture of the two. However, the general point remains that the responses function as conceptual metonymies in our conceptualization of surprise. The analysis above in terms of some generic and specific conceptual metonymies should not be taken to imply that the individual words characterized by particular conceptual metonymies have the same meaning. Clearly, each of them has a distinct meaning. The Natural Semantic Metalanguage (NSM) as discussed by Cliff Goddard (Goddard, this volume) is a good way of making finer distinctions in the meanings of these items.



Surprise as a conceptual category

5. The surprise process metaphorically conceived We just saw that the responses that surprise produces are in part physical and in part mental. When we hear statements that something is an eye-opener or that someone is stunned by some news or that a person becomes speechless when he or she notices something or that people are shocked by an event, we can metonymically infer that the person is greatly surprised. However, the statements actually involve both the cause of surprise and the particular effect that is caused, as in the sentences below (sentences taken from various dictionaries): (4) Her biography is a real eye-opener. (http://www.merriam-webster.com/dictionary/eye-opener) (5) The community was stunned by the tragedy. (http://www.oxforddictionaries.com/definition/english/stun) (6) When he walked into his surprise birthday party, he was completely speechless. (http://www.yourdictionary.com/speechless) (7) She was shocked at the state of his injuries. (http://www.oxforddictionaries.com/definition/english/shock#shock-2)

In such examples, there is the cause of surprise that produces the response. In other words, it is the entire chain of the surprise process (as described above) going back to the cause that produces the physical or mental response indicated by the metonymies discussed above. At the same time, it should be noticed that the metonymies are embedded in a metaphorical interpretation of the surprise process. The metaphor involved can be given as: surprising someone is unexpectedly impacting someone. The impact can be either physical or mental. My claim here is that lexical items like eye-­opener, stun, speechless, shock, and so on, function as both metonymies and metaphors. As metonyms, they indicate a particular response, and as metaphors, they indicate how the response comes about all the way from the cause of surprise. There are, however, several lexical items that indicate how the response comes about metaphorically without at the same time indicating what the response will actually be. Here are some of these (with examples taken from various dictionaries): (8) bolt from / out of the blue: He seemed to be very happy in his job, so his resignation came as a bolt out of the blue. (http://idioms.thefreedictionary.com/a+bolt+from+the+blue) (9) bombshell: The news came as a bombshell. (http://www.oxforddictionaries.com/definition/english/bombshell)

17

18

Zoltán Kövecses

(10) rude awakening: We had a rude awakening when we saw the amount of our phone bill. (http://idioms.thefreedictionary.com/a+rude+awakening) (11) wake-up call: The bombing was a wake-up call to strengthen domestic security. (http://www.thefreedictionary.com/wake-up+calls) (12) turn up / one for the books: There have been a lot of scandals in local politics over the years, but this is one for the books. (http://www.learnersdictionary.com/definition/book)

We can see how the surprising someone is unexpectedly impacting someone metaphor works if we set out the mappings of the metaphor: the person affected by the physical/psychological force → the person experiencing the emotion/surprise impacting the person physically or psychologically → causing the emotion/ surprise the force responsible for the physical or psychological impact → the cause of the emotion/surprise the physical or psychological sensations/feelings produced → the emotional feeling/surprise caused the responses to the impact associated with what the person feels → the physical or psychological responses produced by the cause and the emotion of surprise the unexpectedness of the physical/psychological impact → the unexpectedness of the cause of the emotion/surprise

In other words, in the source domain there is a physical or psychological force that impacts a person suddenly and unexpectedly and it results in certain physical or psychological responses in that person. This is a very simple and fairly specific scenario or frame that is conceptually mapped onto what happens when a person experiences surprise. A major conceptual consequence of this set of metaphorical mappings is that the person who experiences surprise has no control over his or her own self; he/ she undergoes a process in relation to which he/she is helpless. This is because when we are impacted by a sudden force, we are not capable of taking control of what happens. Losing control quickly is a chief feature of surprise, as is revealed by the metaphors and metonymies we have seen up to this point.



Surprise as a conceptual category

6. The notion of “loss of control” in surprise Interestingly, the strong element of loss of control in surprise is explicitly revealed by the origin of the word surprise. This is what one dictionary says about it: late Middle English (in the sense ‘unexpected seizure of a place, or attack on troops’): from Old French, feminine past participle of surprendre, from medieval Latin superprehendere ‘seize’. (retrieved from http://www.oxforddictionaries.com/definition/english/ surprise)

As can be seen, the concept of surprise derives metaphorically from the mini scenario of an attack – on either a place or a person. The conceptual metaphor we can postulate to account for this would be surprising (someone) is an unexpected seizure / attack. We can lay out the structure of the conceptual metaphor with the help of the mappings that follow:

place or person seized or attacked → the person surprised the seizing / attacking → the causing of surprise the seizure / occupation → the surprise the unexpectedness of the seizure / attack → the unexpectedness of surprise

And, most importantly, since the people (who possess the place) or the soldiers occupied or attacked lose control over the place and/or themselves: loss of territory or loss of ability to act → loss of self-control

The surprising (someone) is an unexpected seizure / attack metaphor can be considered a special case of the more generic metaphor controlling is possessing. Unexpectedly seizing something or attacking someone results in losing things (including, metaphorically, abilities as things). This is how a loss of things (possessed objects) can metaphorically indicate loss of control (over action), and this is why, in particular, we have the metaphorical conventional expressions losing control or loss of control. Many additional metaphorical expressions of surprise are based on the specific surprising (someone) is an unexpected seizure / attack metaphor and on the generic controlling is possessing metaphor, including the following (taken from dictionaries): (13) take someone by surprise: The question took David by surprise. (http://www.oxforddictionaries.com/definition/english/surprise)

19

20 Zoltán Kövecses

(14) catch someone unawares: Burchill’s shot caught the goalkeeper completely unawares. (http://www.macmillandictionary.com/dictionary/british/unawares) (15) catch someone off guard: Tom caught Ann off guard and frightened her. (http://idioms.thefreedictionary.com/catch+off+guard)

This is possible because acts of taking, catching, as well as seizing, gripping, and capturing, are acts as a result of which someone loses something, and thus metaphorically loses control. We can find a number of these words in the description of several other emotions, such as fear and anger (see also below). To the degree that surprise and the other emotions share this kind of terminology, they also share the conceptualization of these psychological states as states of temporary loss of control, as shown, for example, by Fontaine, Scherer, and Soriano, eds., (2013). Such a conceptualization is based on the two metaphors (more precisely, source domains) discussed above. Finally, mention must be made of another conceptual metaphor of surprise, as this is revealed by a set of collocates of surprise, including the verbs express, register, show, feign, hide (http://oxforddictionary.so8848.com/search?=surprise). They all portray surprise as an emotional state, where the responses or symptoms associated with the emotion are metonymically viewed as standing for the emotion and are collectively construed as an entity or object, yielding the conceptual metaphor surprise is an object. Thus, surprise can be shown, expressed, registered, and hidden, meaning that its characteristic symptoms can be shown, expressed, registered, and hidden. (It is difficult to tell whether feign is an instance of the surprise is an object conceptual metaphor.) Overall, the result that the main distinguishing feature of surprise is a sudden loss of self-control is consistent with and confirms the GRID analysis of surprise (Soriano, Fontaine, and Scherer, this volume). What the present analysis may add to the GRID result is that the “novelty” feature of surprise, that is, its sudden and unexpected nature, is closely related to the loss of self-control in a person who constantly attends to and monitors the events going on around him/ her in order to maintain cognitive control. In other words, the kind of conceptual metaphor analysis presented here views suddenness and unexpectedness in surprise not as one of the features among the many that characterizes surprise and several other emotions but as an inherent part of a cultural-cognitive model of this emotion – that is, as a part of the flow of particular experiences we interpret as surprise (and call it surprise).



Surprise as a conceptual category

7. What makes surprise different? Judging by the language used about surprise and the conceptual metaphor analysis based on this language, it can be suggested that surprise as a conceptual category is similar to other prototypical emotions in several respects: 1. It shares with other emotions much of their schematic conceptual structure. In it, there is an event or thing that causes a person to be surprised. This process leads to the production of a variety of physical and mental responses. 2. Given the study of the conventional language of surprise in dictionaries, many of the generic expressive and behavioral responses associated with surprise are shared by other, more prototypical emotions. 3. The language used to talk about surprise is highly metonymic and metaphoric, as attested by the dictionary data. The conceptual metonymies and metaphors of other emotions are shared by surprise on a generic level. Thus, surprise shares with other emotions the generic conceptual metonymy the effects of an emotion for the emotion. It also shares the generic conceptual metaphor controlling is possessing, as shown by examples such as seized by fear or gripped by anger. 4. Surprise seems to share with other emotions the idea that entering an emotional state goes together with a loss of self-control. This means that the cognitive model of surprise overlaps with that of other emotions, such as fear and anger, as also suggested, among others, by Fontaine et al. (2013). 5. Similar to other emotions, surprise is viewed as a state that can be shown, feigned, and hidden. However, there are also remarkable differences between surprise and other emotions, as can be seen in light of the analysis above. These include the following: 1. Most of the prototypical emotions are characterized by a control component in their schematic conceptual structure. The person tries to make an attempt to control his socially prohibited or discouraged responses and actions. As the present study indicated, this is not the case for surprise, where there are no such social considerations to control the emotion. No metaphorical expressions, like struggling with, wrestling with, fighting off, or holding back surprise, were found in the dictionaries that, with other emotions, signal people’s attempt to control emotions (see Kövecses, 1990, 2000). 2. While surprise shares generic expressive and behavioral responses with other emotions, it seems to elaborate less on physiological responses as displayed by conventionalized linguistic expressions. This may only be the result of the kind of linguistic analysis performed here. Other kinds of research (such as

21

22

Zoltán Kövecses

physiological studies), however, may reveal that physiology is involved in surprise, and, consequently, possibly in its conceptualization as well. But in the latter case, the physiological responses would not be linguistically coded as highly conventionalized linguistic expressions. 3. While prototypical emotion concepts (such as anger, fear, sadness, joy) are typically characterized by a large number of different conceptual metonymies and metaphors (see Kövecses, 2000), surprise appears to utilize only a few. As regards the metonymies, at the specific level most of them seem to be unique to surprise, such as eyes opening wide for surprise and the mouth opening for surprise, though some of the others are more general, such as inability to speak (for surprise) and inability to think clearly (for surprise), and apply to other emotions. As regards the metaphors, given the methodology, at the specific level I found only two conceptual metaphors that appear to play a major role in the conceptualization of surprise: surprising someone is unexpectedly impacting someone and surprising (someone) is an unexpected seizure / attack. The former might be more limited in scope and be fairly specific to surprise. The latter appears to be more widely shared at a general level, where the metaphorical use of words, such as take, grip, catch, seize, and so on, does apply to other emotions as well. What is most remarkable, though, about the surprise metaphors is the absence of a number of conceptual metaphors in surprise that characterize other emotions. Thus, surprise has no such common emotion source domains as opponent, captive animal, social superior, fire, fluid in a container, and so on (Kövecses, 1990, 2000). 4. Despite the significant overlap in cognitive models between surprise and other emotions, surprise lacks an important part of the model that other emotions have: the stage of attempting to control surprise. This is made apparent and explained by the absence of some of the conceptual metaphors in surprise mentioned above: opponent, captive animal, and social superior. These, and some others like fluid in a container, serve the purpose of capturing this aspect of emotions, that is, they profile the component of attempted control and social regulation in emotions (see Kövecses, 1990, 2000). The control-related metaphors of surprise (surprising someone is unexpectedly impacting someone and surprising (someone) is an unexpected seizure / attack) are metaphors that unequivocally indicate a sudden and complete loss of control, without any attempt to control what happens in surprise (though there are metaphors that indicate the “hiding” of surprise – i.e., the hiding of loss of control). This is why attempt at control is so conspicuously absent from the model of the conceptual structure of surprise, unlike from those of other, more prototypical emotions.



Surprise as a conceptual category

5. Surprise is metaphorically conceptualized as an object (that can be shown and hidden), but, unlike other emotions, it is not construed metaphorically as a substance inside the body container. Thus, one can be full of joy but not of surprise (though one can be full of surprises, which, however, indicates the person who is often the source of surprise and not the emotion itself, as in the case of joy). 6. Unlike more prototypical emotions, there is no distinctive action, or “action tendency” (Fontaine et al., 2013; Manstead et al., 2004), associated with surprise, such as retaliation in anger or flight in fear. The surprise scenario simply stops with producing certain responses as a result of a causal process. This may explain why the “attempt at controlling” aspect of prototypical emotions is missing from surprise. If there is no action that is characteristically associated with surprise, there is no need to control it. 8. Conclusions Based on the language used to talk about surprise, as provided by dictionaries, I outlined a tentative cognitive model for surprise. The model builds heavily on (1) the lexical structure of the category, (2) the conceptual metonymies as found in surprise-related language, and (3) the conceptual metaphors as found in the same. As regards the lexical structure of surprise, it consists of a number of related senses, such as cause of emotion, causing emotion, and emotion. Together, they reveal a large part of the conceptual structure of surprise. As regards the conceptual metonymies, they appear to define the major (i.e., conventionalized) behavioral and expressive responses associated with surprise. That is, they constitute the effect component of its conceptual structure. As regards the metaphors, they characterize the most fundamental quality of surprise; namely, that it is a process in the course of which the self loses control. The loss of control may be thought of as an epistemic failure on the part of the self. It seems that surprise is an emotion that serves to reveal our incapacity to handle the mentally unmonitored or unattended-to aspects of the world. The examination of the various senses of surprise and the metaphors and metonymies related to the concept of surprise revealed a particular conceptual structure, a highly schematic folk theory, represented below: cause (of surprise) – causes – surprise | effects of surprise

23

24

Zoltán Kövecses

This schematic conceptual structure partially coincides with that of several prototypical emotions (anger, fear, joy, sadness). But prototypical emotions have a more extended conceptual structure that includes an elaborate “control” aspect and an “action” component (indicated in bold in the diagram below), as suggested by a number of researchers in the cognitive linguistic study of emotions (Kövecses, 2000; Lakoff & Kövecses, 1987; Soriano, 2005; Yu, 1998): cause (of emotion) – causes – emotion – control – action | effect of emotion

The control aspect of prototypical emotions typically consists of several stages: attempt at control, loss of control, and lack of control. As we saw above, in surprise there is an immediate loss of control and there is no attempt to resist the emotion. In prototypical emotions, there is also a loss of emotional control, but it is preceded by an attempt to control the emerging emotion. This is one reason why the schematic folk theory of surprise is different from that of prototypical emotions. Another reason is that, although surprise manifests itself in a number of physical and mental effects like other emotions, it does not have a clear action tendency (the action aspect) that prototypical emotions are characterized by. Together, these features of surprise make it a nonprototypical emotion. This conclusion receives support from the GRID project that was conducted by means of a completely different methodology (see Fontaine et al., 2013). As far as the methodology used in the paper is concerned, I briefly outlined two potentially useful ways of doing linguistic metaphor analysis: the lexical approach and the corpus linguistic one. I suggested that the two methods complement each other. In the paper, I chose to employ the lexical approach that, in my view at least, has yielded important results concerning the conceptualization of emotions in general and, in the present paper, of surprise in particular. It has made it possible to see in specific ways what surprise shares conceptually with other more prototypical emotions, and, more importantly, what it is that it does not, and hence, why it is not a prototypical emotion concept – at least in folk theoretical terms. As has been pointed out in the paper, the lexical approach and the entirely different methodology of the GRID project have produced remarkable convergences in our understanding surprise. In a way, it would be surprising if corpus linguistics did not further corroborate these findings. At the same time, we can expect corpus linguistics to reveal many subtler details in the use of the concept both in terms of the specific contexts in which it is employed within the same culture and in terms of the global linguistic/cultural systems in a more general cross-cultural perspective.



Surprise as a conceptual category

References Athanasiadou, A., & Tabakowska, E. (Eds.). (1998). Speaking of emotions: Conceptualization and expression. Berlin: Mouton de Gruyter.   doi: 10.1515/9783110806007 Cameron, L. (2003). Metaphor in educational discourse. London: Continuum. Celle, A., & Lansari, L. (2014). ‘I’m surprised’/’Are you surprised?’: Surprise as an argumentation tool in verbal interaction. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discours/Emotions in discourse (pp. 267–279). Bern/Frankfurt: Peter Lang. Damasio, A. (1994). Descartes’ error: Emotion, reason, and the human brain. New York: Penguin. Deignan, A. (2005). Metaphor and corpus linguistics. Amsterdam: John Benjamins. doi: 10.1075/celcr.6 Ekman, P., & Friesen, W. V. (1975). Unmasking the face. Englewood Cliffs, NJ: Prentice Hall. Ekman, P., Levenson, R. W., & Friesen, W. V. (1983). Autonomic nervous system activity distinguishes among emotions. Science, 221, 1208–1210.   doi: 10.1126/science.6612338 Fontaine, J. J. R., Scherer, K. R., & Soriano, C. (Eds.). (2013). Components of emotional meaning: A sourcebook. Oxford: Oxford University Press. doi: 10.1093/acprof:oso/9780199592746.001.0001 Frijda, N. H. (1986). Emotion. Cambridge: Cambridge University Press. Gibbs, R. W. (1994). The poetics of mind: Figurative thought, language, and understanding. New York: Cambridge University Press. Gibbs, R. W. (1999). Taking metaphor out of our heads and putting it into the cultural world In R. W. Gibbs & G. J. Steen (Eds.), Metaphor in cognitive linguistics: Selected papers from the 5th International Cognitive Linguistics Conference (pp. 145–166). Amsterdam: John Benjamins.   doi: 10.1075/cilt.175.09gib Gibbs, R. W. (2006). Embodiment and cognitive science. Cambridge/New York: Cambridge University Press. Gibbs, R. W., & Colston, H. (2012). Interpreting figurative meaning. Cambridge/New York: Cambridge University Press.   doi: 10.1017/CBO9781139168779 Goddard, C. (this volume). The complex, language-specific semantics of “surprise”. Goossens, L. (1990). Metaphtonymy: The interaction of metaphor and metonymy in the expression of linguistic action. Cognitive Linguistics, 1, 323–340.   doi: 10.1515/cogl.1990.1.3.323 Handl, S., & Schmid, H.-J. (Eds.). (2011). Windows to the mind: Metaphor, metonymy and conceptual blending. Berlin/New York: Mouton de Gruyter. Holland, D., & Quinn, N. (Eds.). (1987). Cultural models in language and thought. Cambridge/ New York: Cambridge University Press.   doi: 10.1017/CBO9780511607660 Kövecses, Z. (1986). Metaphors of anger, pride, and love. Amsterdam: John Benjamins. doi: 10.1075/pb.vii.8 Kövecses, Z. (1990). Emotion concepts. Berlin/New York: Springer Verlag. doi: 10.1007/978-1-4612-3312-1 Kövecses, Z. (2000). Metaphor and emotion. Cambridge/New York: Cambridge University Press. Kövecses, Z. (2002). Metaphor: A practical introduction. Oxford/New York: Oxford University Press. Kövecses, Z. (2005). Metaphor in culture: Universality and variation. Cambridge/New York: Cambridge University Press.   doi: 10.1017/CBO9780511614408

25

26 Zoltán Kövecses

Kövecses, Z. (2008). Metaphor and emotion. In R. W. Gibbs (Ed.), The Cambridge handbook of metaphor and thought (pp. 380–396). Cambridge/New York: Cambridge University Press. doi: 10.1017/CBO9780511816802.023 Kövecses, Z. (2010). Metaphor: A practical introduction, 2nd edition. Oxford/New York: Oxford University Press. Kövecses, Z. (2011). Methodological issues in conceptual metaphor theory. In S. Handl & H.-J. Schmid (Eds.), Windows to the mind: Metaphor, metonymy and conceptual blending (pp. 23–39). Berlin/New York: Mouton de Gruyter. Kövecses, Z. (2015). Where metaphors come from: Reconsidering context in metaphor. Oxford/ New York: Oxford University Press.   doi: 10.1093/acprof:oso/9780190224868.001.0001 Lakoff, G. (1987). Women, fire, and dangerous things. Chicago: The University of Chicago Press. doi: 10.7208/chicago/9780226471013.001.0001 Lakoff, G., & Kövecses, Z. (1987). The cognitive model of anger inherent in American English. In D. Holland & N. Quinn (Eds.), Cultural models in language and thought (pp. 195–221). Cambridge/ New York: Cambridge University Press. doi: 10.1017/CBO9780511607660.009 Langacker, R. (1987). Cognitive grammar. Vol. 1. Stanford: Stanford University Press. LeDoux, J. (1996). The emotional brain. New York: Simon and Schuster. Lutz, C. A. (1988). Unnatural emotions: Everyday sentiments on a Micronesian atoll and their challenge to western theory. Chicago: The University of Chicago Press. Manstead, A. S. R., Frijda, N., & Fischer, A. (Eds.). (2004). Feelings and emotions: The Amsterdam Symposium. Cambridge: Cambridge University Press.   doi: 10.1017/CBO9780511806582 Niemeier, S., & Dirven, R. (Eds.). (1997). The language of emotions. Amsterdam: John Benjamins.   doi: 10.1075/z.85 Palmer, G., & Occhi, D. (Eds.). (1999). Languages of sentiment: Cultural constructions of emotional substrates. Amsterdam: John Benjamins.   doi: 10.1075/aicr.18 Pragglejaz Group. (2007). MIP: A method for identifying metaphorically used words in discourse. Metaphor and Symbol, 22(1), 1–39.   doi: 10.1080/10926480709336752 Quinn, N. (Ed.). (2005). Finding culture in talk: A collection of methods. New York: Palgrave Macmillan. Russel, J. A., Fernández-Dols, J. M., Manstead, A. S. R., & Wellenkamp, J. C. (Eds.). (1995). Everyday conceptions of emotion. Dordrecht: Kluwer.   doi: 10.1007/978-94-015-8484-5 Semino, E. (2008). Metaphor in discourse. Cambridge: Cambridge University Press. Soriano, C. (2005). The conceptualization of anger in English and Spanish: A cognitive approach. Unpublished doctoral dissertation. University of Murcia, Murcia, Spain. Soriano, C., Fontaine, J., & Scherer, K. (this volume). Surprise in the GRID. Stefanowitsch, A. (2006). Words and their metaphors. In A. Stefanowitsch & S. T. Gries (Eds.), Corpus-based approaches to metaphor and metonymy (pp. 64–105). Berlin: Mouton de Gruyter.   doi: 10.1515/9783110199895 Wilson, P. (Ed.). (2012). Dynamicity in emotion concepts. Frankfurt: Peter Lang. Yu, N. (1998). The contemporary theory of metaphor. A perspective from Chinese. Amsterdam: John Benjamins.   doi: 10.1075/hcp.1

The complex, language-specific semantics of “surprise”* Cliff Goddard

Griffith University, Australia

This study is conducted using the NSM (Natural Semantic Metalanguage) methodology, which seeks to explicate complex language-specific concepts into configurations of simple universal concepts (Goddard, 2011; Goddard & Wierzbicka, 2014a; cf. Ye, 2013). The study has three main dimensions. It begins by turning the lens of NSM semantic analysis onto a set of words that are central to the “discourse of the unexpected” in English: surprised, amazed, astonished and shocked. By elucidating their precise meanings, we can gain an improved picture of the English folk model in this domain. A comparison with Malay (Bahasa Melayu) shows that the “surprise words” of English lack precise equivalents in other languages (cf. Goddard, 1997). The second dimension involves grammatical semantics, seeking to identify the semantic relationships between agnate word-sets such as: surprised, surprising, to surprise; amazed, amazing, to amaze. The third dimension is a theoretical one, concerned with the goal of developing a typology of “surprise-like” concepts. It is argued that adopting English-­specific words, such as surprise or unexpected, as descriptive categories inevitably leads to conceptual Anglocentrism (Wierzbicka, 2014). The alternative, non-Anglocentric strategy relies on components phrased in terms of universal semantic primes, such as ‘something happened’ and ‘this someone didn’t know that it will happen’, and the like. Keywords: lexical semantics, lexical typology, surprise, NSM, Anglocentrism

* The explications in this study have been discussed extensively with Anna Wierzbicka, who provided many helpful suggestions. I am also grateful to Bert Peeters for a number of helpful suggestions. A version of this study was presented at the International Symposium on Describing and Expressing Surprise, U. Paris-Diderot, 10 October 2014. I thank the participants for useful feedback. Two anonymous reviewers for RCL provided useful comments. doi 10.1075/bct.92.02god © 2017 John Benjamins Publishing Company

28

Cliff Goddard

1. Introductory comments What can languages tell us about how people around the world think and feel differently about the “unexpected”? Is it possible to build a semantic typology of how “surprise-related” concepts are expressed in the lexicons and grammars of the world’s languages? The conviction behind the present study is that before we can take up these questions we would do well to look carefully into a language close to home, namely, English. My main focus is on English “surprise words”, such as surprised, amazed, astonished and shocked, with some contrastive reference to Malay. English warrants special attention on account of its de facto role as a global lingua franca and the consequent tendency for English language categories to be taken for granted (“naturalised”) in theory formulation and scholarly discussion generally, with a danger of implicit conceptual Anglocentrism. I will employ the Natural Semantic Metalanguage or NSM approach to semantic analysis (Goddard, 2011; Goddard & Wierzbicka, 2014a, 2014b; Levisen, 2012; Peeters, 2006; Ye, 2013). There is a long tradition of NSM studies into emotional meaning and mental states generally. The most significant single work is Anna Wierzbicka’s (1999) book Emotions across languages and cultures (Wierzbicka, 1999) – itself the culmination of a decade’s previous work. It seems important also to mention Wierzbicka’s (2014) latest book Imprisoned in English, which is pertinent because of its focus on the need to avoid Anglocentrism in the human sciences. In addition to Wierzbicka’s work on emotion semantics, which has mainly been on English, Russian, and Polish, a variety of other NSM researchers have undertaken similar studies into a diversity of languages, including Chinese, Japanese, Malay, Lao, French, German, Portuguese, Danish, Ewe, Arrernte, Mbula (PNG) and Koromu (PNG). Many of these studies are included in edited collections such as Harkins and Wierzbicka (2001), Enfield and Wierzbicka (2002) and Goddard and Ye (2014).1 NSM bibliographies and other resources are available at the NSM homepage hosted by Griffith University [short URL: bit.ly/Lz6QbN]. 1.1

The NSM methodology

The NSM methodology (Goddard, 2011; Goddard & Wierzbicka, 2014a) depends on findings that have emerged from a decades-long program of cross-­linguistic research. These findings indicate that there is a smallish set of indefinable, elementary meanings that constitute the basic building blocks of linguistic meaning in 1. NSM researchers have also spent a good deal of time on the semantics of interjections, including “surprise-related” interjections (Goddard, 2014a; Gladkova, Vanhatalo & Goddard, 2015; Wierzbicka, 2003, Ch. 8), in English and in other languages.



The complex, language-specific semantics of “surprise”

Table 1.  Semantic primes (English exponents), grouped into related categories i, you, someone, something~thing, people, body

substantives

kinds, parts

relational substantives

this, the same, other~else

determiners

one, two, some, all, much~many, little~few

quantifiers

good, bad, big, small

evaluators and descriptors

know, think, want, don’t want, feel, see, hear

mental predicates

say, words, true

speech

do, happen, move

actions, events, movement

be (somewhere), there is, be (someone/something)

location, existence, specification

(is) mine

possession

live, die

life and death

when~time, now, before, after, a long time, a short time, for some time, moment

time

where~place, here, above, below, far, near, side, inside, touch

place

not, maybe, can, because, if, like very, more, like~as logical concepts • Exponents of primes can be polysemous, i.e., they can have other, additional meanings. • Exponents of primes may be words, bound morphemes, or phrasemes. • They can be formally complex. • They can have combinatorial variants or “allolexes” (indicated with ~). • Each prime has well-specified syntactic (combinatorial) properties.

all or most languages. These elementary meanings are known as semantic primes. They are listed in Table 1 using English exponents. Comparable tables have been drawn up for a diverse range of languages (see the studies in the edited volumes: Goddard, 2008; Goddard & Wierzbicka, 2002; Peeters, 2006). Semantic primes have a grammar of combination, which allow them to combine into phrases, sentences and texts. Evidence suggests that this combinatorial grammar is essentially language-independent, in the sense that the same combinations can be realised in all or most languages. Together, semantic primes and their associated grammar make up the Natural Semantic Metalanguage. As well as simple combinatorial possibilities (e.g. ‘this someone’,2 ‘at many times’, etc.), the grammar of semantic primes includes extended valencies and

2. Although it is somewhat unidiomatic, NSM researchers use the expression ‘this someone’ in English-based NSM, in preference to ‘this person’, because there are reasons to believe that the English word person carries significant Anglo cultural baggage.

29

30

Cliff Goddard

Table 2.  Selected valency frames for semantic primes happen, know, think and feel something HAPPENS something HAPPENS to someone/something something HAPPENS somewhere (in a place)

[minimal frame] [undergoer frame] [locus frame]

someone KNOWS something about something/someone someone KNOWS that … someone KNOWS who/what/where/how …

[topic of knowledge frame] [propositional complement] [wh-complement]

someone THINKS something about someone/something someone THINKS like this: “…”

[topic of cognition frame] [quasi-quotational frame]


someone FEELS something (good/bad)

[minimal frame]

complement options. It is not possible, or necessary, here to give a sketch of the full NSM grammar, but for illustrative purposes Table 2 displays selected valency and complement frames for the key semantic primes that are relevant to the present study, namely happen, know, think and feel, using English exponents. 1.2

The importance of metalanguage

Using the NSM metalanguage allows us to decompose complex language-specific concepts into configurations of simple concepts that are shared across languages. This allows a very high degree of semantic resolution and enables us to access language-­specific meanings using rigorous, evidence-based procedures of semantic analysis. It also helps to safeguard scientific discourse from the conceptual Anglocentrism that can arise from taking English-specific words as a starting point for theory construction (Wierzbicka, 2009a, 2009b, 2014). In relation to “surprise”, for example, it enables us to test the extent to which certain meaning components of English words such as surprise, shocked, and the like, are specific to English; and which components may be shared across all or most languages. Ultimately, NSM can enable us to get clear about what we are talking about, and to carry out scientific and hermeneutic discourses with improved clarity and precision.3 3. There has been some debate about whether surprise is or isn’t an emotion. To my mind this is a fairly pointless exercise unless we have a clear and agreed understanding of what we mean by an “emotion”. In principle, the meaning of the English word emotion, like any ordinary English word, can be explicated using NSM methods (cf. Wierzbicka, 1999, 2009a, 2009b), but the situation is complicated by the fact that the term emotion has been adopted into psychology, where it is used with a range of partly theory-dependent, and often obscure, meanings.



1.3

The complex, language-specific semantics of “surprise”

Outline

The remainder of the paper has three sections. Section 2 turns the lens of NSM semantic analysis onto a set of words that are central to the “discourse of the unexpected” in English: surprised, amazed, astonished and shocked. Grammatically, they are past participial adjectives. By elucidating the precise meanings of these words, and the differences between them, we gain an improved picture of the English folk model in this domain. Comparison with the nearest comparable words in Malay shows that these English words lack precise equivalents in other languages. Section 3 turns to grammatical semantics; specifically, to the relationship between agnate words of different grammatical status. We consider present participial adjectives, e.g. surprising, amazing, shocking, then the corresponding verbs, e.g. to surprise; to amaze, to shock. The general finding is that the differences in grammatical status are systematically correlated with differences in meaning, though certain core components of meaning are shared. The concluding discussion in Section 4 draws out the broader methodological implications of the present study and proposes a way forward for constructing a non-Anglocentric semantic typology of “surprise-related” notions in the languages of the world. 2. The lexical semantics of “surprise words” in English and Malay This section proposes a series of NSM semantic explications. It is not possible for reasons of space to provide thorough exemplification or justification of every detail. 2.1

English surprised, amazed, astonished and shocked

Like other NSM explications, the explications below start with a Lexicosyntactic Frame which is intended to capture the very generic, top-level, structure shared by words of the relevant subclass (Goddard, 2011, 2015). The explications for surprised, amazed, astonished and shocked, in contexts like This someone X is surprised (amazed, etc.), all have the following Lexicosyntactic Frame: Furthermore, it is well established that the meaning of the word emotion is significantly culture-­ specific and historically contingent (Dixon, 2003). For both reasons, it seems advisable that interdisciplinary dialogue should try to avoid the term emotion altogether, or at least use so-called “scare quotes” to draw attention to the dangers of using this term uncritically.

31

32

Cliff Goddard

this someone X thought something about something at that time at the same time this someone felt something because of it

This portrays this subclass of words as depicting a “blend”, so to speak, of a thought and a feeling. Note that linguistic evidence indicates clearly that the concepts behind surprised, amazed, astonished and shocked involve a “feeling” component – because it is possible to speak about feeling surprised, feeling amazed, feeling astonished, or feeling shocked. After this, the template has three sections labelled Trigger Situation, Thought, and Feeling, respectively. The Trigger Situation depicts a recent event which has an effect on the knowledge-state of the experiencer. As noted by Wierzbicka (1995, p. 225), the past participial form of the adjective “points to an external cause”. a short time before it was like this: – something happened – because of this, this someone knew something about something

Next comes the Thought component, and it is the content of this component that largely distinguishes between surprised, amazed, astonished, and shocked. The experiencer’s thought is not about the content of the knowledge as such, but rather about its unexpectedness. For example, the Thought section for surprised reads as follows. Meanings like these can rightly be seen as “meta-cognitive”.4 after this, this someone thought about it like this: “I didn’t know before that it will be like this  I know it now”

Finally comes the Feeling section, which expands on the feeling that has been foreshadowed in the Lexicosyntactic Frame. when this someone thought like this, this someone felt something like people feel at many times when they think like this

The second line presents the experiencer’s feeling as “typical”. For some of the explications, the feeling quality is indicated as specifically ‘good’ or ‘bad’. For example, with shocked the feeling is ‘very bad’.

4. In early work, Wierzbicka (1992, p. 549) employed a different prototypical thought for surprise, namely: ‘I didn’t think before now: this will happen; if I thought about this I would have said, this will not happen’. Current NSM theory does not allow counterfactual constructions in the metalanguage grammar; and in any case, the proposed counterfactual component was counter-intuitively complex. Phrasing the prototypical thought in terms of the semantic prime know, rather than think, gets around both these difficulties.



The complex, language-specific semantics of “surprise”

I now run through the four explications with brief comments. For visual emphasis, the components that differ between the explications are underlined. 2.1.1  When first thinking about surprise (especially with reference to the noun), one is inclined to think of fairly dramatic events, surprise parties, and the like. In ordinary usage, however, people often speak of being surprised by events or facts that are only mildly unexpected, e.g. I was surprised he remembered my name; I’m surprised to see you here; She was surprised by Wendy’s offer of help. The main line in the Thought component has therefore been worded as: ‘I didn’t know before that it will be like this’. It might be objected that every hour of the day we encounter situations that we didn’t know would be the case and yet this does not mean that we are surprised all the time. But this objection would miss the point that (according to this explication) one is only liable to be surprised when one mentally “registers” that a situation is unexpected, takes note of it (so to speak), and, furthermore, feels something because of it. [A] Someone X was surprised (at that time). this someone X thought something about something at that time at the same time this someone felt something because of it

lexico­syntactic frame

a short time before it was like this: – something happened – because of this, this someone knew something about something

trigger situation

after this, this someone thought about it like this: “I didn’t know before that it will be like this  I know it now”

thought

when this someone thought like this, this someone felt something like people feel at many times when they think like this

feeling

2.1.2  Amazed is often described in dictionaries as “extreme surprise” or as being “very surprised”. Accordingly, the first two lines of the Thought section of explication [B] are exactly the same as for surprised, while the additional final line (‘I don’t know how it can be like this’) accounts for the impression of extra “strength”. This line also links amazed with “wondering”.

33

34

Cliff Goddard

[B] Someone X was amazed (at that time). this someone X thought something about something at this time at the same time this someone felt something because of it

lexicosyntactic frame

a short time before it was like this: – something happened – because of this, this someone knew something about something

trigger situation

after this, this someone thought about it like this: “I didn’t know before that it will be like this I know it now at the same time I don’t know how it can be like this”

thought

when this someone thought like this, this someone felt something good like people feel at many times when they think like this

feeling

Amazed generally seems to have a “positive” feel to it (a “slight feel of happiness or goodness”, as one commentator in an online forum put it), so the associated Feeling is described in terms of ‘feeling something good’. Admittedly, it is possible to find examples such as I was amazed at how expensive everything is in Sweden, where the new information content is presumably unwelcome; but even in sentences like these, the speaker does not seem terribly troubled by the unwelcome fact. Sentences which combine amazed with a complement which is unequivocally tragic or serious tend to sound insensitive, e.g. ?I’m amazed at the rapid spread of the ebola virus or ?She was amazed to discover that so many journalists had been killed in Russia. Attributing a positive feeling to amazed accounts for this incongruity effect. 2.1.3  The main idea behind explication [C] for astonished is that the subject experiences something like “disbelief ” (i.e. thinks: ‘it can’t be like this’), while at the same time facing the reality that it is like this (‘I know now that it is like this’). [C] Someone X was astonished (at that time). this someone X thought something about something at this time at the same time this someone felt something because of it

lexicosyntactic frame

a short time before it was like this: – something happened – because of this, this someone knew something about something

trigger situation

after this, this someone thought about it like this: “it can’t be like this at the same time I know now that it is like this”

thought

when this someone thought like this, this someone felt something like people feel at many times when they think like this

feeling



The complex, language-specific semantics of “surprise”

2.1.4 Being shocked is triggered by thinking that something one has newly come to know is both ‘very bad for someone’ and very unexpected. The new information is often about something that has happened, but not necessarily so. One can easily be shocked to discover that someone has cancer (and one is not shocked by the discovery, but by the fact). There is an associated very bad feeling. [D] Someone X was shocked (at that time). this someone X thought something about something at this time at the same time this someone felt something because of it

lexicosyntactic frame

a short time before it was like this: – something happened – because of this, this someone knew something about something

trigger situation

after this, this someone thought about it like this: “this is very bad for someone I didn’t know before that it can be like this”

thought

when this someone thought like this, this someone felt something very bad like people feel at many times when they think like this

feeling

It is important to distinguish between be shocked, as explicated in [D], and get a shock. The latter expression belongs to a specialised “get an X” construction which is not about how someone is thinking and feeling at a given time, but rather about something that happens to someone as a result of suddenly having a certain kind of “bad thought”, cf. get a fright, get a scare, get a nasty surprise. The construction has a causative counterpart with “give someone an X”, e.g. give someone a shock, give someone a fright, give someone a scare, give someone a nasty surprise. Both constructions are extremely interesting but cannot be pursued here. 2.2

Comparison with Malay (Bahasa Melayu)

Malay (Bahasa Melayu) is the national language of Malaysia. The explications below are updated from Goddard (1997), which contains additional exemplification and discussion. There are three main “surprise words” in this language: terkejut, terperanjat, and hairan. Another word, kagum, is closer to “(being) astonished”, but I will not touch on it here. Terkejut and terperanjat both involve prefix ter-, which is (semantically) roughly analogous to a passive participle. It implies suddenness and lack of control. On linguistic evidence terkejut and terperanjat are not “feelings”. (One cannot use the word rasa ‘feel’ with them, i.e. expressions like *rasa terkejut and *rasa

35

36

Cliff Goddard

terperanjat are unacceptable, whereas, rasa hairan, by contrast, is perfectly fine). They depict cognitive reactions to an unexpected event, something like the English expressions “be startled, get a surprise” and “get a shock”, respectively. In both cases, the reaction involves a “disruptive” cognitive effect (Goddard, 2003, pp. 212–213). Deciding about the Lexicosyntactic Frame for these verbs means taking into account other similar verbs, including terpegun ‘be stunned, speechless for a moment’, tergamam ‘be flustered’, and possibly others. It seems that the following Lexicosyntactic Frame is applicable in all cases: ‘this someone X thought something at that time; something bad happened to this someone because of it; it happened in one moment’. Hairan, by contrast, can be considered an ordinary “emotion” word. We will now treat each of the words in turn, illustrating usage with examples drawn from contemporary Malaysian literature, including popular novels, cartoon novellas, short stories and magazine articles. The sources are given in the Appendix. Given the semantic orientation of the present study, interlinear glosses have not been provided but the following comments may assist. Malay lacks inflectional morphology, having no obligatory grammatical marking for tense, case, definiteness, or number. The preferred order of phrases is the same as for English, that is, Subject–Verb–(Object), and, like English, the language has prepositions. The main deviation from English word-order is that most nominal modifiers, such as demonstratives and adjectives, follow the noun, though quantifiers precede. Aspectual particles, like sudah ‘already’ and the negative tidak (or tak), immediately precede the verb. 2.2.1  Terkejut is often used about someone being “startled” out of a daydream or reverie or from deep thought, as in examples (1)–(2) below. In example (2), the person is absorbed in memories of her dead mother; it seems to her that she can see her mother’s face reflected in the waters. Equally, however, the experiencer is often not depicted as absorbed in thought. For example, in (3) Siti Munah is at work in the kitchen when a sound from outside makes her terkejut. The terkejut reaction is very common in response to an unexpected loud noise, as in this example, but it can be used about visual and verbal stimuli, and about situations where someone “starts” without any external stimulus.5 Example (4) shows that the triggering stimulus can be seeing something, in this case the unexpected appearance of the ghost of a dead friend. (In the story, it is established that the 5. The English language does not really have any common expression for when a person spontaneously comes out with a ‘startle reaction’ in the absence of an external trigger. Expressions like He gave a start or She started suddenly are possible but relatively rare.



The complex, language-specific semantics of “surprise”

protagonist has seen the ghost before and is not afraid of him, so it is just the suddenness of the ghost’s appearance which triggers the terkejut.) (1) Tiba-tiba Zarof terkejut dari lamunannya bila bahunya ditepuk dan orang itu terus berdiri di sisinya. (TDL, p. 24) ‘Suddenly Zarof was startled from his reveries when a hand landed on his shoulder. Someone was standing beside him.’ (2) Serentak dengan itu, seekor katak melompat mengocak permukaan sungai. Dia terkejut seketika.  (MLB, p. 10) ‘At the same time as that, a frog jumped disturbing the surface of the river. She was startled for a moment.’ (3) ‘Pengecut!’ tiba-tiba Ludin menempik entah kepada siapa. … Siti Munah yang sedang memasak di dapur terkejut.  (S, p. 45) ‘‘Cowards!’ Ludin suddenly shouted at no one in particular. … Siti Munah who was cooking in the kitchen was startled.’ (4) Wah! Kenapa kau menjelma dengan tidak semena-mena? Buat orang terkejut!’  (MK, p. 57) ‘Hey! Why do you appear without warning? You startled me!’

The range of use can be explained by attributing the terkejut experience to a sudden thought along these lines: ‘something happened at this moment, I don’t know what’, as shown in explication [F]. The terkejut experience also causes a brief period of “cognitive paralysis”, as set out in the Cognitive Effect section. In Malay cultural belief, the terkejut experience is considered potentially harmful (i.e. ‘it can be bad for this someone’), as stated in the final, Evaluation section. [F] Someone X terkejut (pada masa ini). this someone X thought something (at this time) something bad happened to this someone because of it it happened in one moment

lexicosyntactic frame

it happened because it was like this at this time: something happened in the place where this someone was when it happened, this someone thought like this: “something happened at this moment, I don’t know what”

scenario

because of this, for a very short time after this, this someone couldn’t think about anything

cognitive effect

when this happens to someone, it can be bad for this someone

evaluation

37

38

Cliff Goddard

2.2.2  Terperanjat is much more unpleasant the terkejut. It is quite like English get a shock or get a fright. Other translation equivalents include ‘stunned’ and ‘taken aback’, as well as ‘surprised’ and ‘startled’. In some cases, a physical description is provided, emphasising the debilitating effect. The stimulus for terperanjat is not normally a mere noise. Terperanjat is typically a reaction to unexpectedly bad news, though the new information does not necessarily have to be bad. It is essential, however, that the degree of unexpectedness, so to speak, is very high. The terperanjat experience is thought of as being harmful – potentially, seriously so. A person who terperanjat will often put a hand to his or her heart. It is even possible to speak of a person dying of terperanjat. For example: Dia mati terperanjat sebab anaknya masuk penjara ‘He died of shock because his son went to prison’ (terkejut would be impossible here). Naturally then, people are supposed to be mindful of the possibility of making others terperanjat and to avoid it, or at least forewarn an addressee of possible bad news. This is illustrated in example (5). Example (6) is a context where someone terperjanat from unexpectedly good news. Kudin hears that he has been promoted, after having resigned himself to the idea that a rival was sure to get the position over him. Examples (7) is interesting in that the stimulus does not involve human action. The experiencer gets a fright when her cat unexpectedly appears with large mouse in its mouth. (5)

Man: ‘Baiklah mak… tapi satu je saya nak cakap…’ Mak Man: ‘Apa dia?’ Man: ‘Mak jangan terperanjat ye?’  ‘Man: ‘OK mum… but there’s just one thing I want to say…’ Mak Man: ‘What’s that?’ Man: ‘You won’t be shocked, will you?’’

(AM, p. 65)

(6) Sebaik sahaja orang yang dilayan bercakap itu meletakkan telefonnya, gagang telefon di tangan Kudin dengan sendiri terlepas dari genggamannya. Kudin kelihatan pucat dan tidak bermaya. Dia terperanjat dengan berita yang baharu sahaja diterima dari orang-orang Bahagiannya.  (TMM, p. 25) ‘As soon as the person calling had finished speaking, the receiver dropped from Kudin’s hand. He looked pale and lifeless. He was stunned by the central executive’s final recommendation’. (7) ‘Opp…’ Syamsinar terperanjat sewaktu kucing hitam yang diberi nama Meow itu lalu di hadapannya, menggonggong seekor tikus mati. Syamsinar terpaku seketika. Dia mengurut-urut dadanya.  (MLB, p. 38) ‘‘Opp…’ Syamsinar got a fright when the black cat called Meow appeared in front of her, carrying a dead mouse in its mouth. She was transfixed for a minute. She rubbed her chest.’



The complex, language-specific semantics of “surprise”

These examples are consistent with explication [G]. The event is depicted as something that happens to a person in response to a sudden, triggered thought along these lines: ‘I didn’t know before that it can be like this’. In response, the experiencer is mentally thrown off balance for some time. Notice that the open-ended time-period attributed to this disruptive effect (‘for some time’) contrasts with that attributed to the very brief disruption induced by terkejut (‘for a very short time’). The Malay belief in the harmfulness of the terperanjat experience is reflected in the final component (‘it can be very bad for this someone’). [G] Someone X terperanjat (pada masa ini). this someone X thought something (at this time) something bad happened to this someone because of it it happened in one moment

lexicosyntactic frame

it happened because it was like this at that time: something happened in the place where this someone was when it happened, this someone thought like this: “I didn’t know before that it can be like this”

scenario

because of this, for some time after this, this someone couldn’t think about anything else

cognitive effect

when this happens to someone, it can be very bad for this someone

evaluation

2.2.3  Hairan is very different to terkejut and terperanjat, both grammatically and semantically. As mentioned, it is or can be considered a “feeling” (kerasaan). The expression rasa hairan ‘feel hairan’ is perfectly acceptable. It is not unpleasant. The hairan experience is extended in time, rather than momentary. All this means that in terms of content, hairan is a bit like “wondering at” something, or being puzzled and amazed, with an associated feeling (cf. Heider, 1991, p. 65). Consistent with the fact that hairan takes a while, the expressions bertambah hairan ‘grow more surprised’ and semakin hairan ‘(his/her) surprise mounted’ refer to hairan building up over time. Often hairan has a visual dimension: you are seeing something which you can’t believe, as in (8). The stimulus isn’t always visual, however. One can hairan upon hearing something, or upon learning of something in some other way, as in (9). (8) Di balik rumpun pokok pisang yang patah tumbang itu, terlentang seekor binatang yang tidak pernah diduga oleh mereka. ‘Seladang?’ serentak mereka berkata hairan.  (S, p. 35) ‘Behind the broken banana trees lay an animal, the like of which they had never expected to see. ‘A wild ox?’ they shouted in surprise.’

39

40 Cliff Goddard

(9) ‘Pelabuhan kita sekarang ini, dulunya ialah sebuah pulau’, terang Zarof. ‘Pulau?’ ‘Yah, pulau. Mengapa hairan?’ ‘Pulau kan di tengah laut?’ (TDL, p. 20) ‘This harbour we have now, it used to be an island’, he replied. ‘An island?’ ‘Yes. Why are you surprised?’ ‘An island should be in the middle of the sea, shouldn’t it?’

In the light of these examples, we can propose the explication in [H]. Notice that, in contrast to terkejut and terperanjat, this does not depict hairan as something that happens to a person, but as a feeling which springs from a sustained thought process. The explication uses the standard NSM template for “emotion” meanings. [H] Someone X hairan (pada masa ini). this someone X thought something about something for some time (at this time) at the same time this someone felt something because of it

lexicosyntactic frame

this someone thought like this: “I don’t know how it can be like this I want to know”

thought

when this someone thought like this, this someone felt something like people feel at many times when they think something like this

feeling

2.2.4  Why all this attention to the micro-semantics of lexical items?, it may be asked. What does it buy us? Simply put, it dramatises the culture-specificity of English. At the full lexeme level, there is very little correspondence between the two languages. Not only is there no Malay word for “surprise”, not a single word in either language has an exact counterpart in the other.6 Furthermore, and this is perhaps an even more dramatic difference, there is no Malay word that even associates “suddenness” with feeling.

6. Not surprisingly, psychologists working on presumed “basic emotions” have had problems finding Malay equivalents for ‘surprise’. Boucher (1983, p. 409) chose hairan, but explained: ‘it connotes more of the “astonishment” meaning of surprise than the English term does’. McAndrew (1986) chose terkejut. Working with Indonesian, Heider (1991, pp. 56–57) also opts for terkejut though he recognises that it (and terperanjat, which he regards as synonymous) has a consistently ‘negative tone’, compared with English surprise.



The complex, language-specific semantics of “surprise”

In terms of closeness of fit between the two languages, the closest match is probably between terperanjat and the English expression get a shock. They are both construed as more or less sudden “happenings” with a disruptive cognitive effect, and both are thought of as potentially harmful (terperanjat more seriously so). The semantics of words like terperanjat and terkejut forge a conceptual link between encountering the unexpected and possible harm to the self. From an Anglo cultural platform, it would be easy to see this in terms of a culturally-based link between surprise and fear, or to say that for cultural reasons Malay people “fear” being “surprised” (cf. Heider, 1991). This kind of interpretation would be Anglocentric, however, because it takes the English-specific categories of “surprise” and “fear” for granted as ontological categories and projects them onto Malay. This inevitably produces a distorted and conceptually inauthentic picture of Malay – just as would happen in reverse if we took the Malay-specific categories for granted and projected them onto English. We return to this point in Section 4. 3.  English grammatical semantics of the “unexpected”: Present participial adjectives (surprising, shocking, etc.) and verbs (to surprise, to shock, etc.) In this section, the question is: How can we model the semantic relationships in agnate word-sets such as: surprised, surprising, and to surprise; i.e. between past participial adjective, present participial adjective, and verb? The treatment must necessarily be brief but the topic is an important one, because the availability of different part-of-speech options allows “surprise-related” meanings to be integrated into discourse with greater ease. In saying this, I do not want to give the impression that the meanings expressed by different part-of-speech counterparts are exactly the same. On the contrary, the corresponding words belonging to different part-of-speech categories express different meanings (Wierzbicka, 1990a, 1995), while at the same time sharing significant components of meaning. 3.1

Present participial adjectives: surprising, amazing, astonishing, shocking

Frequency figures in Collins Wordbanks Online indicate that surprising is a much more frequent word than surprised – about four times as frequent. It is most often used predicatively, e.g. That’s (not) surprising. In addition, there is the adverb surprisingly. The word amazing is also very common. Biber, Johansson, Leech, Conrad, and Finegan (1999) placed it in a list of the top twenty -ing adjectives.

41

42

Cliff Goddard

Going back to Jespersen (1933), linguists have noted the strong tendency for present participial adjectives to express some kind of “simultaneity” with respect to a contextually given time period (De Smet, n.d.). Present participial adjectives are also sometimes termed “active-subjective”. Explications [I] and [J] below show the meaning structures for surprising and shocking, used as predicate complements. Two explications are sufficient for illustrative purposes, but it will be clear enough how the corresponding explications for amazing and astonishing are constructed. Essentially, to say that something is surprising, amazing, shocking, etc. is to make a claim about the kind of cognitive-­ affective reaction that ‘someone’ (an unspecified, hypothetical someone) can have when thinking about it. This reaction has two aspects: a certain prototypical thought and a feeling that typically goes with this thought. As one might expect, the content of the thought and the nature of the feeling are the same as with their past participial counterparts. [I] It is surprising. it is like this:

lexicosyntactic frame

when someone thinks about this, this someone can think like this: “I didn’t know before that it will be like this I know it now”

potential thought

when someone thinks like this, this someone can feel something like people feel at many times when they think like this

feeling

[J] It is shocking. it is like this:

lexicosyntactic frame

when someone thinks about this, this someone can think like this: “this is very bad for someone I didn’t know before that it can be like this”

potential thought

when someone thinks like this, this someone can feel something very bad like people feel at many times when they think like this

feeling

Although to describe something as surprising, shocking, etc., is – obviously – to make a subjective assessment, the “sourcing” of this assessment is occluded. The adjective conveys the impression that the assessment is somehow obvious or understandable, an effect which is shared by other “emotionally loaded” descriptive adjectives, such as disgusting, annoying, and frightening. The explication shows



The complex, language-specific semantics of “surprise”

how this effect is achieved; i.e. the thought is not presented as belonging to the speaker or to any specific person, but is attributed to a hypothetical someone in a typical situation. 3.2

English experiencer-object “surprise verbs”: to surprise, to amaze, to shock, to astonish

The verbs to surprise (someone), to amaze (someone), to shock (someone), to astonish (someone), etc. belong to a largish class of “experiencer-object” emotive verbs in English, along with to frighten, to annoy, to anger, to upset, to distress, to depress, and others. They are morphologically closely related to the adjectival form (cf. Levin, 1993). It is worth noting at this point (we will return to it later) that English also has a set of “experiencer-subject” emotive verbs, such as enjoy, worry, grieve, rejoice, pine, sulk, and brood, but of these only the first three (enjoy, worry, grieve) are particularly common. For speakers of English, it seems only natural that “surprise verbs” should be of the experiencer-object variety, but some languages, such as Russian and Polish, commonly use experiencer-subject “surprise verbs”. English verbs like to surprise, to amaze, to shock, etc. can appear in various sentence patterns, including: (a) with a discourse anaphoric it or that as subject, denoting a fact, situation, or the like, e.g. It really surprised me (implying that the experiencer came to know about something, and this triggered the reaction);7 (b) with a person as subject, e.g. He surprised them by attending the service (implying that this person did something unexpected, and this triggered the reaction); or (c) a noun designating a speech-act, e.g. The question surprised me (implying that someone said something unexpected, which triggered the reaction). The verbs to surprise and to amaze also occur in some common speech formulas and semi-fixed expressions, e.g. It wouldn’t surprise me if …, You might be surprised at …, and It never fails to amaze me that … I will only address the frame listed in (a) above, i.e. sentences like That really surprised (shocked, amazed, etc.) me, which have a discourse anaphoric that or it as subject. The first matter that needs attention is the Lexicosyntactic Frame and this can only be decided with reference to the larger subclass of experiencer-object verbs, including to frighten and to annoy (in uses such as That frightened me and It

7. There is a second meaning to surprise2: e.g. Nancy had surprised him in the foyer; They surprised the enemy.

43

44 Cliff Goddard

annoyed me). I propose the following. The first line depicts an unspecified event, which is the causal trigger for the experiencer’s thought and associated feeling. something happened at that time because of this, this someone X thought something about something at that time at the same time this someone felt something because of it

This Lexicosyntactic Frame would appear to be applicable to English experiencer-­ object verbs generally. Unlike the “surprise verbs”, other experiencer-­object verbs, such as to frighten and to annoy, do not depend on any change in the experiencer’s knowledge-state. What all the experiencer-object verbs share, however, is the idea of a thought (and an associated feeling) being triggered by an event of some kind. Since the experiencer’s thought is presented as being “caused”, the implication is that the experiencer had little or no control over it. The content of the thought is specific to each individual verb and is depicted in the Thought section of the template. For example, with to frighten the thought is: ‘something very bad can happen to me now, I can’t do anything’ (Wierzbicka, 1990b); and for to surprise it is (unsurprisingly, at this point): ‘I didn’t know before that it will be like this, I know it now’. Only one explication is presented below, but it is easy enough to see how comparable explications for to amaze, to shock, and to astonish can be constructed. [K] It surprised someone X (at that time) to learn/see, etc. something happened at that time because of this, this someone X thought something about something at that time at the same time this someone felt something because of it

lexicosyntactic frame

this someone thought like this about something: “I didn’t know before that it will be like this I know it now”

thought

when this someone thought like this, this someone felt something like people feel at many times when they think like this

feeling

3.3

Active “surprise verbs”: A note on Slavic languages

In English the predominant mode of description for “surprise-like” meanings is the use of adjectives (participial in form). The language does have some “surprise verbs”, as we have just seen, but in terms of frequency they clearly play second fiddle to the adjectives, and, furthermore, even the verbs are distinctly “happen-­



The complex, language-specific semantics of “surprise”

oriented”. In Malay, as we saw in Section 2.2, verbs are the predominant mode of description for “surprise-like” meanings, and the Malay verbs are even more “happen-oriented” than the English ones. In both languages, “surprise-like” experiences (except for Malay hairan) are seen as brief and short-lived. Neither language envisages, so to speak, that a person may have an active cognitive involvement in generating a “surprise-like” experience or that such experience may be protracted in time. It is important to note, therefore, that some languages, in particular Slavic languages such as Russian and Polish, do allow such possibilities. It is well-known that Slavic languages are extremely rich in intransitive active emotion verbs, which, in some grammatical frames (with nominative subject, and often with the verb in reflexive form) depict a person bringing about, or at least sustaining, a certain emotion by willingly turning his or her thoughts in a certain direction (Wierzbicka, 1990a, 1995). For example, a sentence like On grustil o nje [he.nom feel.sad-past.masc about her] expresses a meaning something like “He was saddening himself about her” or “He was making himself sad by thinking about her”. The surprising thing, from an English language perspective, is that the same grammatical mechanism is available for “surprise-like” meanings, using the active verb udivljat’sja (imperfective),8 which is something like English ‘to marvel’. For example: Ja udivljajus’ ego smelosti “I surprise-myself at his bravery” i.e. ‘I marvel at his bravery’. Moreover, the most common Russian counterparts for English surprised and amazing – namely, udivlennyj (adjective) and udivitel’nyj (participle) – are grammatical derivatives of the perfective form udivit’sja of the same verb. Obviously a good deal of research would be needed to tease out the precise semantics of the Russian and Polish systems, which is a matter for further research, but it is clear that the semantics of the Slavic languages reflect significantly different perspectives to both English and Malay. 4. Implications for semantic typology Emotion words do not translate well across languages. Not only surprise, but also the other premier English emotion words, such as happiness, fear, sadness, anger, and disgust – lack precise semantic equivalents in languages such as Danish, French, German, Spanish, Russian and Chinese (Gladkova, 2010a, 2010b; Goddard, 2010, 2014b; Harkins & Wierzbicka, 2001; Levisen, 2012; Russell, 8. Something similar is possible in French with the s’étonner de (related to être étonné and étonnant), but only very marginally so with the verb se surprendre de (Peeters, p.c.).

45

46 Cliff Goddard

1991; Wierzbicka, 1998, 1999; Ye, 2006), let alone in more far-flung languages of the world. The implications are profound. In order to think and speak clearly in this arena, and in order to avoid conceptual Anglocentrism, we need to set aside complex English words, such as surprise, and try instead to think and speak in terms that are simpler and transposable across languages. As the founder of cultural psychology, Richard Shweder (Shweder, 2004) has put it, we need to “deconstruct emotions for the sake of comparative research”. Rather than formulating research questions in terms of language-specific words such as surprise, we would do better to work in terms of possible sub-­ lexical semantic components – expressing these in simple translatable words, i.e. semantic primes. On the basis of the present study, I would propose that the following configuration of semantic primes can act as a cross-translatable semantic touchstone for “surprise-related” notions. this someone thought like this at this time: ‘I didn’t know before that it will/can be like this’

We can then formulate research questions such as the following: – What role does this configuration play in the lexicons of the world’s languages? Is it the case that all or most languages have one or more common words that include this configuration as a key component? – How often do such words also include reference to an associated “feeling”? If there is a feeling involved in a given language, is this feeling characterised as good or as bad, or is it left unspecified? – What other semantic components tend to occur with it? For example, how common is the evaluation ‘something bad can happen to this someone because of this’? How common are components like: ‘for a short time after this, this someone can’t think well about anything’? What variants of these components are found? – How do such semantic components tend to be distributed across different word-­classes, e.g. adjectives, verbs, interjections, and grammatical constructions, e.g. impersonal constructions, reflexive constructions, exclamatory constructions? By using the simple cross-translatable terms of the NSM metalanguage as a theoretical vocabulary for formulating questions and answers about meaning, we can hope to develop a clear and precise semantic typology of “surprise-related”



The complex, language-specific semantics of “surprise”

notions, without the danger of Anglocentric conceptual bias entering into the terms of description.9

Sources of the Malay textual examples The abbreviation in parenthesis at the end of each reference is used to identify the source of the example sentences cited in the body of the paper. Anon. (1991). Malam Kemalangan. Jenakarama, 48, 54–61. 15/1/91. (MK) Azizi Haji Abdullah. (1982). Seladang. In H. Awang (Ed.), Sayembara III Cerpen Hadiah Esso-­ Gapena (pp. 27–51). Kuala Lumpur: Dewan Bahasa Dan Pustaka. (S) Aznoor. (1992). Anak Menantuku. Jenakarama, 77, 62–65. 1/7/92. (AM) Datin Masnah Rais. (1982). Tanahnya Masih Merah. In H. Awang (Ed.), Sayembara III Cerpen Hadiah Esso-Gapena (pp.14-26). Kuala Lumpur: Dewan Bahasa Dan Pustaka. (TMM) Sabda S. (1983). Tetamu dari Langit. In H. Awang (Ed.), Sayembara IV Cerpen Hadiah Esso-­ Gapena (pp. 16–32). Kuala Lumpur: Dewan Bahasa Dan Pustaka. (TDL) Zaidah Haji Baharuddin. (1990). Mega di Langit Barat. Selangor Darul Ehsan: Marwilis Publisher & Distributors Sdn. Bhd. (MLB)

References Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. Harlow: Pearson. Boucher, J. D. (1983). Antecedents to emotions across cultures. In S. H. Irvine & J. W. Berry (Eds.), Human assessment and cultural factors (pp. 407–420). New York, NY: Plenum. doi: 10.1007/978-1-4899-2151-2_28 Collins WordBanks Online. De Smet, H. (n.d.) The meaning of the English present participle. [https://lirias.kuleuven.be/ bitstream/123456789/278750/2/The+meaning+of+the+English+present+participle+] accessed 29 June 2014. Dixon, T. M. (2003). From passions to emotions: The creation of a psychological category. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511490514 Enfield, N. J., & Wierzbicka, A. (Eds.). (2002). The body in the description of emotion. Special issue of Pragmatics & Cognition, 10(1). Gladkova, A. (2010a). Russkaja kul’turnaja semantika: ėmocii, cennosti, žiznennye ustanovki [Russian cultural semantics: Emotions, values, attitudes] Moscow: Languages of Slavonic Cultures. Gladkova, A. (2010b). ‘Sympathy’, ‘compassion’, and ‘empathy’ in English and Russian: A linguistic and cultural analysis. Culture & Psychology, 16(2), 267–285. doi: 10.1177/1354067X10361396 9. As well, we reduce the danger of getting bogged down in fruitless arguments about technical terminology, an occupational hazard of linguistics.

47

48 Cliff Goddard

Gladkova, A., Vanhatalo, U., & Goddard, C. (2015). The semantics of interjections: An experimental study with Natural Semantic Metalanguage. Applied Psycholinguistics. First View. doi: 10.1017/S0142716415000260 Goddard, C. (1997). Contrastive semantics and cultural psychology: ‘Surprise’ in Malay and English. Culture & Psychology, 3(2), 153–181.  doi: 10.1177/1354067X9700300204 Goddard, C. (2003). Dynamic ter- in Malay (Bahasa Melayu): A study in grammatical polysemy. Studies in Language, 27(2), 287–322.   doi: 10.1075/sl.27.2.04god Goddard, C. (Ed.). (2008). Cross-linguistic semantics. Amsterdam: John Benjamins. doi: 10.1075/slcs.102 Goddard, C. (2010). Universals and variation in the lexicon of mental state concepts. In B. C. Malt & P. Wolff (Eds.), Words and the mind: How words capture human experience (pp. 72– 92). New York, NY: Oxford University Press. Goddard, C. (2011). Semantic analysis, Revised 2nd edition. Oxford: Oxford University Press. Goddard, C. (2014a). Interjections and emotions (with special reference to “surprise” and “disgust”). Emotion Review, 6(1), 53–63.  doi: 10.1177/1754073913491843 Goddard, C. (2014b). On “disgust”. In F. Baider & G. Cislaru (Eds.), Linguistic approaches to emotions in context (pp. 73–98). Amsterdam: John Benjamins.  doi: 10.1075/pbns.241.06god Goddard, C. (2015). Verb classes and valency alternations (NSM approach), with special reference to English physical activity verbs. In A. Malchukov & B. Comrie (Eds.), Valency Classes in the World’s Languages (pp. 1649–1680). Berlin: Mouton de Gruyter. Goddard, C., & Wierzbicka, A. (Eds.). (2002). Meaning and universal grammar: Theory and empirical findings. Amsterdam: John Benjamins.  doi: 10.1075/slcs.60 Goddard, C., & Wierzbicka, A. (2014a). Words and meanings: Lexical semantics across domains, languages and cultures. Oxford: Oxford University Press. Goddard, C., & Wierzbicka, A. (2014b). Semantic fieldwork and lexical universals. Studies in Language, 38(1), 80–127.  doi: 10.1075/sl.38.1.03god Goddard, C., & Ye, Z. (Eds.). (2016). “Happiness” and “pain” across languages and cultures. Amsterdam: John Benjamins. [First published 2014, as special issue of International Journal of Language & Culture, 1(2).] Harkins, J., & Wierzbicka, A. (Eds.). (2001). Emotions in crosslinguistic perspective. Berlin: Mouton de Gruyter.  doi: 10.1515/9783110880168 Heider, K. G. (1991). Landscapes of emotion: Mapping three cultures of emotion in Indonesia. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511527715 Jespersen, O. (1933). Essentials of English grammar. London: Taylor & Francis. Levin, B. (1993). English verb classes and alternations. Chicago: University of Chicago Press. Levisen, C. (2012). Cultural semantics and social cognition: A case study on the Danish universe of meaning. Berlin: Mouton de Gruyter. McAndrew, F. T. (1986). A cross-cultural study of recognition thresholds for facial expressions of emotion. Journal of Cross-cultural Psychology, 17(2), 211–224. doi: 10.1177/0022002186017002005 Peeters, B. (Ed.). (2006). Semantic primes and universal grammar: Evidence from the Romance languages. Amsterdam: John Benjamins.  doi: 10.1075/slcs.81 Russell, J. A. (1991). Culture and the categorization of emotion. Psychological Bulletin, 110, 426–450.  doi: 10.1037/0033-2909.110.3.426



The complex, language-specific semantics of “surprise”

Shweder, R. A. (2004). Deconstructing the emotions for the sake of comparative research. In A. S. R. Manstead, N. Frijda, & A. Fischer (Eds.), Feelings and emotions: The Amsterdam symposium (pp. 81–97). Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511806582.006 Wierzbicka, A. (1990a). Introduction (Special issue on “The semantics of emotions”). Australian Journal of Linguistics, 10(2), 133–138.  doi: 10.1080/07268609008599439 Wierzbicka, A. (1990b). The semantics of emotions: Fear and its relatives in English. Australian Journal of Linguistics, 10(2), 359–375.  doi: 10.1080/07268609008599447 Wierzbicka, A. (1992). Defining emotion concepts. Cognitive Science, 16, 539–581. doi: 10.1207/s15516709cog1604_4 Wierzbicka, A. (1995). Adjectives vs. verbs: The iconicity of part-of-speech membership. In M. E. Landsberg (Ed.), Syntactic iconicity and linguistic freezes: The human dimension (pp. 223–245). Berlin: Walter de Gruyter.  doi: 10.1515/9783110882926.223 Wierzbicka, A. (1998). “Sadness” and “anger” in Russian: The non-universality of the so-called “basic human emotions”. In A. Athanasiadou & E. Tabakowska (Eds.), Speaking of emotions: Conceptualisation and expression (pp. 3–28). Berlin: Mouton de Gruyter. Wierzbicka, A. (1999). Emotions across languages and cultures. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511521256 Wierzbicka, A. (2003). Cross-cultural pragmatics: The semantics of human interaction, Expanded 2nd edition. Berlin: Mouton de Gruyter.  doi: 10.1515/9783110220964 Wierzbicka, A. (2009a). Language and metalanguage: Key issues in emotion research. Emotion Review, 1(1), 3–14.  doi: 10.1177/1754073908097175 Wierzbicka, A. (2009b). Overcoming Anglocentrism in emotion research. Emotion Review, 1(1), 21–23.  doi: 10.1177/1754073908097179 Wierzbicka, A. (2014). Imprisoned in English: The hazards of English as a default language. New York, NY: Oxford University Press. Ye, Z. (2006). Why are there two “joy-like” ‘basic’ emotions in Chinese? Semantic theory and empirical findings. In P. Santangelo & D. Guida (Eds.), Love, hatred and other passions: Questions and themes on emotions in Chinese civilisation (pp. 59–80). Leiden: E.J. Brill. Ye, Z. (2013). Comparing the Natural Semantic Metalanguage (NSM) approach to emotion and the GRID paradigm. In J. J. R. Fontaine, K. R. Scherer, & C. Soriano (Eds.), Components of emotional meaning: A sourcebook (pp. 339–409). Oxford: Oxford University Press.

49

Grammatical evidentiality and the unprepared mind* Tyler Peterson

University of Auckland

The aim of this paper is to investigate how evidentials are used to reflect the unprepared mind of the speaker. Lexical evidentials are typically used by speakers to encode the kinds of evidence they have for making statements about states, events and actions they did not personally witness or were a part of. However, under certain conditions evidentials can be used to express the surprise of the speaker. This is known in the literature under the label mirativity. The empirical factors that condition the mirative use of an evidential are first determined, and then using an information and schema-theoretic analysis it is shown that mirativity is the linguistic reflex of a series of mental events involving the processing of new information, coupled with other contextual factors involving speaker knowledge. Keywords: evidentiality, mirativity, Schema Theory

1. Introduction Imagine that you are sitting in a restaurant. The waiter arrives at the table and asks which of the specials they are offering that evening: the halibut or the veal. What is your reaction? If you like fish, you order the halibut. The following week you decide to try out a new restaurant that has recently opened in your neighbourhood. The waiter approaches your table and asks you if you’d like to sing along with him. What is your reaction? Most likely surprise: this event does not ‘fit’ with what you * Special thanks to my Gitksan consultants Barbara Sennott and Louise Wilson, and my Turkish consultant Hande Ergun. Thanks also to the participants of the symposium Describing and Expressing Surprise at the University of Paris-Diderot, Agnès Celle, Laure Lansari, and two anonymous reviewers for their valuable comments. This research was made possible with a grant from the Endangered Languages Documentation Program (SOAS), awarded to the author. All errors and misinterpretations of cited data are my own. doi 10.1075/bct.92.03pet © 2017 John Benjamins Publishing Company

52

Tyler Peterson

know about how restaurants normally work: the role of a waiter is to take meal orders and serve food. In other words, you are ‘mentally unprepared’ for the waiter’s deviation from this role when they ask you to sing. Now, imagine you and Gwen are preparing a surprise birthday party for Sheila. You don’t expect her home for another hour. After an hour you look outside and see Sheila’s car in the driveway. What is your reaction? Perhaps to tell Gwen and the others ‘Looks like Sheila’s home (let’s hide)’. The following week you and Gwen are preparing a surprise birthday party again, this time for Alvin. You don’t expect him home for another hour. However, in the middle of your preparations, Alvin walks through the door. What is you’re reaction? Most likely (1) a. ‘Alvin!’ b. ‘I’m surprised you’re home so early.’ c. ‘What are you doing here?!’

or a variety of other words, expressions, sentence types, or information structure marking that other speech participants interpret as surprise. This event does not ‘fit’ with our expectations of the current situation: as with the singing waiter, you are mentally unprepared for Alvin’s arrival. The general goal of this paper is to enrich our understanding of the linguistic, contextual and cognitive properties of utterances that reveal the ‘unprepared mind’ of the speaker.1 Broadly speaking, new information that is not easily assimilated into a speaker’s current situational awareness is often coded differently linguistically than information that can be adapted into this awareness. This coding – often referred to as mirativity – is the linguistic reflex of what we commonly interpret as surprise (DeLancey 1997). Languages employ a variety of grammatical strategies for registering the surprise of the speaker. Using Sheila’s birthday party as an illustration, imagine again that you and Gwen are preparing a surprise birthday party for Sheila. You don’t expect her home for another hour. After an hour you look outside and see Sheila’s car in the driveway. A speaker of Gitksan (Tsimshianic) will likely use the sentence in (2): (2) ʼnakw=hl witxw=s Sheila EVID=CND arrive=PND Sheila ‘Looks like Sheila is here.’

In this context, Gwen does not actually see Sheila arrive, but she infers that she has arrived because she has indirect sensory evidence for her arrival: her truck is in the driveway. Gwen expresses her inference using the grammatical evidential ʼnakw.

1. The description ‘unprepared mind’ is adapted from Aksu-Koç and Slobin (1986).



Grammatical evidentiality and the unprepared mind

Turning to Alvin’s birthday party preparations, you don’t expect him home for another hour. However, in the middle of your preparations, Alvin walks through the door. You exclaim: (3) ʼnakw=hl witxw=s Alvin EVID=CND arrive=PND Alvin ‘[Looks like] Alvin is here!’

The sentence in (3) is grammatically identical to the sentence in (2). However, the indirect evidential meaning of ʼnakw is ‘short-circuited’ in this context: Gwen actually sees that it is the case that Alvin is here. This misuse of ʼnakw does not lead to infelicity, instead, in direct evidence contexts ʼnakw has a mirative translation in addition to its evidential one. This is a cross-linguistically robust phenomenon: in many languages which have grammatical evidentials that encode indirect evidence, mirative meaning is often signalled by the use of this evidential (DeLancey 1997, 2001; Aikhenvald 2004). An often cited example in the literature on evidentiality and mirativity is found in Turkish, where the evidential suffix -miş in (4) can be used to indicate surprise, in addition to its evidential meaning (Aksu-Koç & Slobin 1986): (4) Turkish (adapted from Aksu-Koç & Slobin 1986: 159) Kemal gel-miş Kemal came-EVID EVIDENTIAL TRANSLATION: ‘[Looks like] Kemal came.’ MIRATIVE TRANSLATION: ‘Kemal came!’

Similar to evidential n’akw in Gitksan, Aksu-Koç and Slobin (1986) report two interpretations of -miş. The first interpretation involves indirect evidence: the speaker sees Kemal’s coat hanging in the hallway, but hasn’t yet seen Kemal. Thus, the speaker infers the presence of Kemal from this evidence. The second interpretation involves the speaker’s surprise at Kemal’s arrival: the speaker hears someone approach, opens the door, and sees Kemal – a totally unexpected visitor. The use of the evidential -miş in this context signals the mirative: a speaker’s immediate experience of an event does not correlate well with the speaker’s awareness of the current situation they are in. Recently, the descriptive label ‘mirativity’ has fallen within the sight-lines of descriptive, functional and theoretical linguists. However, this attention has also revealed a number of fundamental issues regarding the kinds of meanings the term mirativity is intended to cover, where these meanings come from, and the nature of the relationship between mirativity and evidentiality. Because of this, the label ‘mirativity’ itself, as it has been defined and passed down through the various language descriptions and typological literature, is insightful but is still

53

54

Tyler Peterson

rather undefined. Linguists have long since recognized the connection between evidentiality and the kinds of meanings associated with mirativity, and we can benefit greatly from their insights. Nonetheless, the empirical and theoretical properties of mirativity as a category have yet to face a dedicated, systematic investigation. This leads to a number of basic questions that motivate this paper: what are the core ingredients that define the category of mirativity, and what is the source of these ingredients? Is mirativity a part of the semantics of evidentiality? If not, then what are the pragmatic, contextual and cognitive factors that drive the mirative meanings in (3) and (4) above? 1.1

Aims and proposals

The central aim of this paper is to deconstruct the putative category of mirativity, focus on why and how the kinds of meanings ascribed to mirativity – surprise and unexpectedness – are implicated by evidentiality, and what this can reveal about the unprepared mind of a speaker within a speech context. This is done by concentrating on the contextual features and the cognitive process that underpin the mirative use of a grammatical evidential. Our intuitions tell us that these various meanings seem to form a kind of natural class that reflects the mental state of the speaker, and how that state is transformed when faced with new information. Therefore, I show that an explanation of mirativity is facilitated by viewing it as a linguistic reflex of a cognitive process within a speech context, thus giving us the means to define this natural class and the context that licenses the mirative use of a grammatical evidential. I argue that the advantage of this approach lies in the fact that no new theoretical machinery needs to be developed in order to explain mirativity – it follows from what we already know about modeling information, speaker knowledge within a speech context, and a link to the theory of surprise as a common cognitive process. It is important to note that it is not the aim of this paper to provide a comprehensive study all of the various characterizations and grammatical realizations of mirativity found cross-linguistically.2 There are various excellent surveys of mirativity, both from a cross-linguistic perspective (for example Aikhenvald 2006), but also how mirativity manifests itself language-internally (for example Lazard 1999; Dickinson 2000; Peterson 2010a). However, I do draw heavily from these studies, complementing the data from these with primary language data from

2. The reader is referred to a special issue of Linguistic Typology (2012, Volume 16, Number 3) devoted to the debate surrounding the categorial status of mirativity from a descriptive and typological perspective.



Grammatical evidentiality and the unprepared mind

Gitksan and Turkish that specifically probes the triangulation between mirativity, evidentiality and context. Rather, the strategy of this paper is to deconstruct mirativity as we understand it now, and take steps to rebuilding it from a cognitive perspective – a task that has yet to be undertaken in the literature – by focussing attention specifically on the connection between evidentiality and mirativity. It is expected that this kind of study will shed light onto the categorial status of mirativity on one hand, and on the other it will provide an analysis that can be used to guide future research into other cross- and intra-linguistic encodings of mirativity. 1.2

Outline

Section 2 deconstructs mirativity as we currently understand it. This is done by first taking stock of the kinds of meanings the label mirativity is intended to cover, as found in the various language descriptions and studies that have mirativity as their focus. Using these insights as a foundation, I forward the generalization that the mirative use of an evidential results from the interpretative tension that arises from the use of evidential meaning when a speaker ‘witnesses’ the state, event, or action denoted by an evidentially marked sentence. In essence, this ‘short circuits’ the meaning of a grammatical indirect evidential, which is typically used when the speaker does not witness an event, but rather that they have evidence for that event. Section 3 takes steps towards reconstructing mirativity. By adapting certain information-­theoretic notions, I develop and motivate in detail the definition of new environmental information, which provides a way of framing states, events and actions in terms of information that is new to the speaker. New environmental information is then embedded within schema theory, the most prominent theory for modelling the situational awareness of a speaker and their network of beliefs and experiences, or what are called schemata. Schema theory also provides a framework for modelling the discrepancies caused between a speaker’s situational awareness and the real-time intake of new environmental information. This is the cognitive locus of surprise, and I show how a link can be made between this process and the linguistic realization of surprise, which, in turn, defines what we label as mirativity. Section 4 fills out the picture by outlining future directions of research on mirativity and how this analysis can be extended to another category mirativity is closely associated with, specifically, control marking.

55

56

Tyler Peterson

2. Mirativity and evidentiality An investigation into the overlap between evidentiality and mirativity is greatly aided by what we already know about evidentials and evidentiality (Chafe & Nichols 1986; Willett 1988; Rooryck 2001a, b; Aikhenvald & Dixon 2003; Aikhenvald 2004; Palmer 2006, McCready & Ogata 2007; Papafragou et al. 2007; Speas 2008; Peterson & Sauerland 2010; a.o.). We can also draw upon the many insights that have been gained on the mirative use of evidentials through the typological studies (cf. Aikhenvald 2004), functional analyses of mirativity (Slobin & Aksu 1982; Aksu-Koç & Slobin 1986; DeLancey 1997, 2001; Lazard 1999; Aikhenvald 2004; Temürcü 2007), and mirativity as the focus of a language description (cf. DeLancey 1990; Dickinson 2000), and also theoretical forays into the syntax and pragmatics of mirativity (Wiklund 2009; Peterson 2010b).3 However, the history of mirativity finds its roots in the study of evidentiality. For example, Jacobsen (1964) noted that the inferential evidential in Washo can have a semantic extension to include ‘surprise’. Although descriptions of mirativity have since appeared in various language descriptions and in the typological literature, discussions of mirativity usually begin with the observations made on the meanings and uses of the Turkish evidential -miş (Slobin & Aksu 1982; Aksu-Koç & Slobin 1986). An example of this was given in (2), which is elaborated on in (5): If a speaker has direct knowledge of Kemal’s arrival (for example, they saw Kemal walk through the door earlier), then a speaker uses the sentence in (5a) marked with the past tense suffix -di. If a speaker has only indirect knowledge of Kemal’s arrival, perhaps by observing Kemal’s coat hanging in the front hall, or from being told that Kemal has arrived, then the speaker uses a sentence that is marked with the inferential evidential -miş, as in (5b). (5) Turkish (Slobin & Aksu 1982: 187) a. Kemal gel-di Kemal came-PAST ‘Kemal came.’ b. Kemal gel-miş Kemal came-EVID ‘Kemal came.’ 3. I refer the reader to these sources for details on the finer grained manifestations of mirativity intra- and cross-linguistically. I won’t be going into mirativity in detail; rather, the purpose of this section is introduce the basic relationship between evidentials and mirativity, and to distill these features with the aim of developing the broader generalizations that capture and help us predict how and why mirativity is expressed through evidentiality.



Grammatical evidentiality and the unprepared mind

This can be described as the functionally unmarked use of evidential -miş, which is to encode the inference a speaker makes from observable or reported evidence. However, a speaker can also use (5b) if they have direct knowledge of Kemal’s arrival. The interpretative tension caused by using an inferential evidential when the speaker has direct knowledge does not lead to infelicity. Rather, the use of -miş in this context is a way of indicating the speaker is ‘not prepared’ for the arrival of Kemal and finds this surprising. Aksu-Koç and Slobin (1986) cast this in psychological terms, and how the alternation in (5) reflects the something of the conscious experience of the speaker: the use of -miş represents an experience for which the speaker lacks ‘premonitory awareness’. DeLancey (1997, 2001) picked up on these observations and connected them to similar phenomena he observed in two unrelated languages: Lhasa Tibetan and Hare. He developed a definition of mirativity as marking information which is ‘new to the speaker’, or more specifically: [Mirativity] marks both statements based on inference and statements based on direct experience for which the speaker had no psychological preparation… What these apparently disparate data sources have in common … is that the proposition is one which is new to the speaker, not yet integrated into his overall picture of the world. (DeLancey 1997: 35–36)

In his discussion of the Hare (Athapaskan) particle lõ, DeLancey provides a minimal pair, given in (6), that is similar to the Turkish evidential -miş (DeLancey 1997: 38–40, 2001: 375–378): (6) Hare (DeLancey 2001: 375) a. júhye sa k’ínayeda hereabout bear SG.go.around.3SG.SUBJ/PERF ‘There was a bear walking around here.’ b. júhye sa k’ínayeda lõ hereabout bear SG.go.around.3SG.SUBJ/PERF EVID ‘I see there was a bear walking around here.’

Example (6a) is a simple declarative statement used by the speaker if they actually witnessed a bear walking around that location. In another context, a speaker has just come out of the house in the morning and finds bear tracks around the door. Unlike (6a) there is no bear witnessing event, rather, she infers from tracks around the door that a bear was walking around there, and the sentence is marked with lõ in (6b). This is the primary function of the evidential lõ: to encode an inference based on the evidence a speaker has for a statement. If they know a statement is true, they use a sentence without an evidential, such as (6a).

57

58

Tyler Peterson

However, if the speaker actually witnesses the event denoted by the sentence, evidential lõ can still be used. This effect is illustrated in (7), where the hero of the story being told, Egadekini, has been sitting up in a tree throwing branches down on an ogre who has been hunting for him. The ogre finally looks up and sees him and says: (7) heee, gúhde daweda! ch’ifi dachída lõ hey, up.there SG.sit.3sg.IMPERF guy sitting MIR ‘Hey, he’s sitting up there! The guy is sitting up there!’ (DeLancey 2001: 376)

This use of lõ would be expected to lead to infelicity, as the unmarked function of an evidential is to encode the evidence they have for states, events or actions the speaker has no first hand knowledge of. However, as we observed with the evidentials in Gitksan and Turkish, infelicity does not result in (7), rather, this is the mirative use of evidential lõ: it implicates that a speaker has, to use DeLancey’s description, no ‘psychological preparation’ for the something that happens in the immediate context. 2.1

A functional definition

DeLancey’s description of the mirative use of evidentials has been both foreshadowed and since corroborated by various researchers working with primary language data. As a brief survey, mirativity of this kind has been characterized using a variety of related descriptors, such as ‘non-expected’ information (Egerod & Hansson 1974), ‘surprise at unexpected new information based on immediate observation’ (Friedman 2003: 197, 200), ‘just discovered’ (LaPolla 2003), and ‘new knowledge’ (DeLancey 2001: 369; see also Aikhenvald 2004: 195–215 for an overview of other references from the descriptive literature).4 Dickinson (2000) in her detailed study of mirative marking in Tsafiki, presents several refinements to the previous characterizations of mirativity, and claims that it involves two different construals concerning how a speaker’s mental state is conditioned by either their immediate experience of an event, or by previous experiences and expectations: In a mirative system, events and states that cannot be easily assimilated are coded differently than those that easily fall in with the speaker’s expectations. One construal is based on the speaker’s past experiences of similar situations and his 4. The related term admirative is used in the Balkan linguistics tradition for non-firsthand evidentials that have a mirative use. See Friedman (2003) for details. This has also been referred to as the médiatif in French linguistics tradition (Lazard 1999).



Grammatical evidentiality and the unprepared mind

general knowledge. This set of assumptions can range from knowledge about purely physical interactions to assumptions based on cultural and social norms. The second construal is based on the speaker’s immediate experience of an event or state. If the immediate situation does not correlate well with the speaker’s expectations, the proposition coding the event or state receives special marking.  (Dickinson 2000: 379)

The kind of special marking Dickinson is referring to is the congruent and noncongruent marking in Tsafiki, a Barbacoan language spoken in South America. In (8a) the speaker uses congruent suffix -yo on the verb to encode that he knows he has money. However, in (8b) the speaker suddenly discovers he has some money he did not think he had. In this context, the speaker instead uses the non-­ congruent suffix -i to encode that he was unaware of this, and is thus unexpected and perhaps surprising: (8) Tsafiki (Dickinson 2000: 401) a. kala ta-yo-e money have-CONGR-DECL ‘I have money.’ b. kala ta-i-e money have-NONCONGR-DECL ‘I have money!!’

Although Tsafiki uses control marking in (8) rather than grammatical evidentials to implicate mirativity, the relevant point here is how Dickinson’s construals ground mirativity in the psychological orientation of the speaker, both in terms of their individual knowledge (their personal experiences and fixed assumptions about the way the world works), and their knowledge of the context (their immediate experience of a state, event, or action). What is insightful about these construals is that surprise, the meaning most often associated with mirativity, is not stipulated. Rather, it shows how a speaker’s psychological orientation in any given context can be transformed upon encountering new information, such as the sensation of finding money in your pocket you weren’t expecting to have. This can then lead to a sense of surprise, among other similarly related notions such as sudden realization and lack of situational awareness. Following similar observations of control marking in Lhasa Tibetan made by DeLancey (1992, 1997, 2001), Dickinson claims that non-congruent verbal marking in Tsafiki lexically encodes mirativity as part of its meaning. Dickinson’s construals of mirativity also appropriately capture the mirative meaning of the Turkish evidential -miş in (5), and the Hare evidential lõ in (7). This is aptly characterized in terms of ‘distancing’, where an ‘inferential evidential marker indicates physical distancing from the event – the speaker did not see it, [whereas] a mirative marker

59

60 Tyler Peterson

indicates psychological distancing – the speaker did not anticipate the event or state’ (Dickinson 2000: 381). However, unlike non-congruent marking in Tsafiki in (8a), the evidential lõ does not have the mirative as a part of its semantics, as the minimal pair in (6) shows.5 In fact, no language that has evidentials has mirativity as part of its entailed meaning, rather, it is somehow an extension of evidential meaning (Lazard 1999). Although Dickinson’s descriptions of mirativity involve a different kind of grammatical category (which I return to in Section 4), it serves as a useful point of departure in untangling the relationship between evidentiality and the psychological characteristics found with mirative evidentials. In Section 3 I show how it directly connects to an information-theoretic analysis of mirativity. 2.2

Grammatical evidentials and speaker knowledge

The function of an indirect evidential is to give a speaker a way of talking about events they haven’t personally seen, heard, or otherwise taken part in. For example, in (9a) Gwen knows Alvin has arrived because she has first-hand knowledge of his arrival (i.e. she sees him standing in the doorway). However, by using of the indirect evidential ʼnakw in (9b) Gwen is communicating two things: first, she does not know or believe Alvin has arrived, and second, she has indirect evidence for his arrival: (9) a. Non-evidential: witxw=t Alvin arrive=PND Alvin ‘Alvin arrived.’ b. Evidential: ʼnakw=hl witxw=s Alvin EVID=CND arrive=PND Alvin ‘Looks like Alvin arrived.’

The basic semantic representations in (10) provide a way of capturing the specifics of the speaker’s belief in the truth of a proposition denoted by a sentence (BELs(p)), and the evidence a speaker has for that proposition (EVID(p)).6 Consider p as the proposition Alvin arrived expressed by both of the sentences in (9):

5. This observation challenges DeLancey’s (2001) analysis of lõ, where he claims that it is indeed a lexical mirative. I show below how this cannot be the case. 6. See Faller (2002), Geurts (2010) for a similar implementation of a belief operator BEL.



Grammatical evidentiality and the unprepared mind

(10) p = Alvin arrived. a. BELs(p) = s believes p in c b. ¬BELs(p) ^¬BELs(¬p) = s believes neither p nor ¬p in c EVID(p) = s has indirect evidence for p in c

(=9a) (=9b)

These representations are at the heart of the observations made above about the evidential-as-mirative sentences in Gitksan, Turkish and Hare: it is the speaker’s knowledge of the proposition embedded under the evidential that determines whether the indirect evidential has a mirative reading or not. This knowledge can only be acquired if the speaker has personally seen, heard, or otherwise taken part in this state, event or action of p (although this characterization is refined in Section 3). Reconsider (9b), repeated as (11) cast in these terms: the evidential sentence in (11a) matches the context for (9b): Gwen does not know if the proposition Alvin arrived is true or not; she only has evidence that he has arrived, and uses ʼnakw to encode this. The mirative interpretation of (11) emerges when the speaker knows or believes the proposition Alvin arrived is true. The semantic description of the mirative use of ʼnakw is given in (11b): (11) ʼnakw=hl witxw=s Alvin EVID=CND arrive=PND Alvin ‘Looks like Alvin arrived.’ p = Alvin arrived. a. Non-mirative evidential: ¬BELs(p) ^¬BELs(¬p) = s believes neither p nor ¬p in c EVID(p) = The speaker has indirect evidence for p in c b. Mirative evidential: BELs(p) = s believes p in c EVID(p) = The speaker has indirect evidence for p in c

Both the non-mirative and mirative scenarios for (11) can be characterized in terms of the speaker’s belief that p. In the evidential (non-mirative) scenario in (11a), where guests are arriving one by one at a party, Gwen does not know whether the proposition embedded under the evidential, EVID(p), is true. She uses the morpheme ʼnakw to indicate that she only has indirect sensory evidence to support the sentence in (11) (i.e. Alvin’s truck is in the driveway). This is the evidential use of ʼnakw, which lacks any surprise or unexpectedness on Gwen’s part. In the mirative scenario in (11b), Gwen knows that the statement is true, because she sees Alvin walking through the door. In this case, the use of ʼnakw indicates surprise at the unexpected arrival of Alvin at the party. The mirative encodes a discrepancy between John’s actual presence and Gwen’s expectation that John would not be present. This is the mirative use of ʼnakw.

61

62

Tyler Peterson

These descriptions reveal that mirative evidentials are semantically identical as non-evidential sentences with regards to speaker belief. This generalization is captured in (12), in particular by comparing (12a) and (12c): (12) a. Non-evidential: BELs(p) = s believes p in c b. Evidential: ¬BELs(p) ^¬BELs(¬p) = s believes neither p nor ¬p in c EVID(p) = The speaker has indirect evidence for p in c c. Mirative evidential: BELs(p) = s believes p in c EVID(p) = The speaker has indirect evidence for p in c

(= 9a)

(= 11a) (= 11b)

The generalization formulated in (12) is that it is the speaker’s knowledge of the truth of a proposition embedded under an evidential that is the pivotal feature that distinguishes between the plain evidential reading of a sentence and an evidential sentence which has a mirative use. Belief and context play a pivotal role determining the mirative use of an indirect evidential: it is the interpretive tension caused by the speaker’s use of an indirect evidential in a direct evidence context (where BELs(p)) that drives mirativity. The representations in (12) capture the descriptions of the mirative uses of the Turkish inferential evidential -miş in (5b), analyzed in (13) using the contexts provided in Aksu-Koç and Slobin (1986: 159): (13) Kemal gel-miş Kemal came-EVID ‘Kemal came.’ p = Kemal came. a. Non-mirative evidential: EVID(p) = The speaker has indirect evidence for p in c ¬BELs(p) ^¬BELs(¬p) = s believes neither p nor ¬p in c b. Mirative evidential: EVID(p) = The speaker has indirect evidence for p in c BELs(p) = s believes p in c

The context in (13a) involves indirect evidence: the speaker sees Kemal’s coat hanging in the hallway, but hasn’t yet seen Kemal. Thus, the speaker infers the presence of Kemal from this evidence and encodes this inference with -miş. The mirative interpretation in (13b) involves the speaker’s surprise at Kemal’s arrival: the speaker opens the door, and sees Kemal. The use of the evidential -miş in this context signals its mirative use: the speaker knows the proposition embedded under the evidential is true.



Grammatical evidentiality and the unprepared mind

With this generalization in place, we can now evaluate the other cases of evidentials used miratively. For example, in Tsafiki the indirect evidential verbal suffix -nu encodes ‘information inferred from direct physical evidence’ (Dickinson 2000: 407). In example (14a) the speaker infers from the fact she is on the floor, she must have fallen out of bed. However, in (14b) the ‘speaker heard what he thought was a car approaching. But when he saw it, he realized it was a motorcycle’ (Dickinson 2000: 407): (14) a. tse lowa=bi ne=chi keere-i-i-nu-e 1FEM bed=LOC from=LOC throw-become-NCONGR-EV/MIR-DECL ‘I must have fallen out of bed.’ (I’m on the floor).  (2000: 412) b. moto jo-nu-e motorcycle be-EVID-DECL ‘It’s a motorcycle!’ (2000: 411)

As with the mirative use of evidentials in Turkish, Hare and Gitksan, the speaker of (14b) sees that it is in fact a motorcycle. This direct knowledge of the motorcycle overrides the unmarked indirect evidential meaning of -nu, yet the result is not infelicity, but rather the kinds of meanings associated with mirativity: a sense of surprise and unexpectedness. This confirms the generalization in (12). In a similar case, the evidential suffix -k in Qiang (LaPolla 2003) has both an inferential evidential meaning and a mirative meaning. An inference may be based on evidence obtained visually or by some other sense. In (15a) the speaker knew the person was supposed to go to Chengdu, but wasn’t sure when, and then saw the person luggage gone, so assumed he had left for Chengdu. However, in (15b) the speaker sees that the door is open, but doesn’t know who opened it, and the evidential suffix -k has a mirative interpretation (i.e. ‘just discovered’): (15) a. theː ʑdʑytɑː ɦɑ-qi-k 3sg chengdu.LOC OR-go-INFER “He went to Chengdu.” b. dʑy de-ʑge-ji-k door OR-open-CSM-INFER “The door is open!”

The generalization in (12) is implicit in many of the descriptions of mirative evidentials, yet it has not been explicitly formulated in the literature. It can also be made into a useful heuristic in order to test whether an indirect evidential in any given language can be used miratively or not. However, it lacks the means to explain why evidentials are used this way, what exactly mirativity is, and how (12) connects to the mirative ‘process’ described by Aksu-Koç and Slobin, DeLancey

63

64 Tyler Peterson

and Dickinson (and others). I believe their observations are on the right track, and the main task in the remainder of this paper is to answer these questions. 3. Reconstructing mirative meaning The previous section showed that what distinguishes between the plain evidential reading and the mirative reading of a grammatical evidential is the speaker’s knowledge of the proposition embedded under the evidential: in order for an evidential sentence to express mirativity, the speaker must see or know by some other means that state, event or action expressed by an evidential sentence is in fact true (or they believe that it is true). The task of this section is to explain how evidentials acquire the kinds of mirative readings in the examples above, why they are used to express the kinds of meanings associated with mirativity, and, ultimately, to give shape to mirativity as a linguistic category. We can start with the generalization that mirativity is about a specific kind of psychological (as well as physiological) response to the speaker’s encounter with new information. In order to explain this, it is necessary to have in place two analytical components: (1) a way of modelling information, and (2) a framework for processing that information. There are two general approaches that can handle these tasks. First, I adopt a general theoretical notion of information and show how it applies to evidentiality. Secondly, I utilize schema theory (i.e. Mandler 1984; Rumelhart 1984), and show how we can embed an information-theoretic analysis of information within a schema-theoretic explanation of surprise. Using this analysis I claim that we can gain an understanding of what it means to be ‘surprised’ at new information, and how this leads to the linguistic realization of mirativity. It also explains a critical feature of mirative sentences: a speaker is not surprised at the proposition embedded under a miratively used evidential, rather, a speaker is surprised at the new information they encounter in some speech context. This point is often obscured in the literature, and it is explained in detail in this section. With these theoretical tools on hand, mirativity can be reconstructed in a principled way – without having to appeal to a separate theory of linguistic mirativity – and show how its various characteristics (surprise, new information, unpreparedness etc.), as well as Dickinson’s two construals of mirativity, can be shown to come from a single source: the everyday cognitive activity of processing of new information.



Grammatical evidentiality and the unprepared mind

3.1

New information

From a broad perspective, the intake of information is a function of our every day life. The perception of states, events, and actions from the world around us feeds this intake, as does the presentation of new information from every day linguistic communication. Our encounters with new information often, but not necessarily, trigger a sensation of surprise or unexpectedness. However, a necessary ingredient for surprise is new information. This observation reflects the intuition that we can’t be surprised by information that is ‘old’. But what exactly does new information consist of? In order to clarify the connection between surprise and new information, it is important to have in place a definition of both information, and information that is ‘new’. As a first step, I distinguish two kinds of information: environmental information and linguistic information.7 3.1.1 Environmental vs. linguistic information The notion of environmental information is an adaptation of a notion from information theory, as defined in (16): (16) Environmental information: For an agent with prior knowledge k, two systems α and β are coupled in such a way that α’s being of type or in a state, event, or action F(α) is correlated to β being of type or in a state, event, or action G(β), thus carrying for the information agent the information that β is G if the agent can infer that β is G from α being F together with k (but could not from k alone).  (Adapted from Barwise & Seligman 1997; McKinlay 2008)

Consider the following scenario as an illustration of (16): you drive by John’s house to ask a favour of him, but you don’t know if he is at home or not. Standing in his driveway, you can see John’s light (α) which is in the state of being on (F(α)). The state of his light provides information from the environment about John (β) in the state of being home (G(β)). This simple scenario shows how environmental information is defined relative to an observer (an information agent), who is supposed to have no direct knowledge that β is G (i.e. John is home). Rather, a kernel of information emerges from the inference that is performed that α and β are coupled in a kind of way: the fact at John’s light is on provides information that John is home. This inference is performed by an agent that must have some prior knowledge (k) of John’s habits with conserving electricity in order for the

7. The distinction (and variations on it) is of course a classic one, with roots in Bar-Hillel and Carnap (1953).

65

66 Tyler Peterson

observation that ‘John’s light is on’ to be meaningful.8 This information is less meaningful to someone who lacks this prior knowledge, such as a man out walking his dog. When an agent becomes a speaker within a speech context, this state of affairs can be linguistically realized in the one of many different ways, a sample of which is given in (17). These utterances are conditioned by the content of F(α) coupled with k, and include, for example, a simple assertion, modal auxiliaries such as must and might, evidential verbs such as looks like and I see, and the raising verb seem: (17) John is home; John must be home; John might be home; Looks like John’s home; I see that John’s home; It seems John’s home; etc.

Environmental information can be conventionalized, or engineered, or it may be entirely natural, as when litmus – a natural colouring matter from lichens – is used as an acid-alkali indicator because it turns red in acid solutions and blue in alkaline solutions. Other typical examples include the correlation between fingerprints and personal identification (see Floridi 2011 for details). Now, consider how this same scenario is linguistically realized in the following Gitksan sentence in (18): (18) ʼnakw=hl ta’a=s John EVID=CND at.home=PND John ‘Looks like John is home.’

As we learned about the meaning of the grammatical evidential ʼnakw from the previous section, the speaker of (18) does not know first hand whether it is true that John is in fact home or not. Rather, the speaker is performing an inference based on the information available in her environment. In languages which have grammatical evidentials such as Gitksan, this inference is encoded by an evidential, which encodes what specific kind of evidence this is. In (18), the use of ʼnakw grammatically encodes the fact the speaker is making an inference from perceptual evidence (seeing that John’s light is on).9 8. See Barwise and Seligman (1997) for a similar discussion along these lines. Within a standard possible worlds analysis of modality, F(α)-knowledge and k-knowledge correspond to the modal base of an ordering source, respectively (Kratzer 1991; Peterson 2012; a.o.). In the following subsection I show how k-knowledge corresponds to a person’s activated cognitive schema. Also see McCready and Ogata (2007), Davis et al. (2007) for probabilistic accounts of information and evidentiality. 9. This line of reasoning involves considering the difference between knowledge and belief. The speaker knows John is home because this belief is justified by her knowledge of John’s



Grammatical evidentiality and the unprepared mind

Depending on contextual factors and the speaker’s prior knowledge (k), environmental information may be very ‘direct’, in the sense that the inference performed is minimized to the simple mental activity of processing visual (or other sensory) data. Using a variation on the context above, you can see John standing at the window of his house (F(α)). This state provides information from the environment about John in the state of being home (G(β)). At first blush this may appear trivial, but this also highlights the importance of k-knowledge in the definition of environmental information in (16): the speaker must know who John is, and that it is his house that he is standing in. Only against the backdrop of this k-knowledge is a relevant assertion such as ‘John’s home’ felicitous. Or, using the same k-knowledge in another context, it may be the case that this same environmental information (John standing in his window) is used to infer that he is watching out for the speaker’s arrival. This direct quality of environmental information is crucial when looking at the kinds of sentences used in this situation, especially with the mirative use of evidentials. I return to this point in more detail below. Standing in contrast to environmental information is what I call linguistic information. This is intended to have a broad coverage, and includes any kind of information that engages the speaker’s (or information agent’s) language faculty. The most straightforward example of linguistic information, as defined here, are the assertions speakers make in conversational exchanges. But linguistic information can include other kinds of illocutionary forces and sentence types, or even more indirect forms of linguistic communication such as conversations overheard at a bus stop, or an article printed in a newspaper. Evidentials often grammaticalize the contrast between environmental and linguistic information. In (19), a speaker makes a statement that John is home. He believes this, not because he has any direct knowledge of John being home, but because he overheard someone at the gas station saying so: (19) ta’a=kat=s John at.home=REP=PND John ‘[I hear(d)] John is home.’

habits with leaving lights on or off, and other relevant considerations. However, John’s light may be on because he left them on before he went away on holidays. The speaker’s belief and its justification would remain the same, but she would not know John is home. It is likely that evidentials will prove to be an interesting testing ground for belief and knowledge, especially as lexical evidentials explicitly reveal a speaker’s knowledge for a proposition. As with non-evidential languages such as English, the difference between knowledge and (true) belief is conditioned by the reliability of information (see Peterson 2012 for a formal analysis of this with evidential languages such as Gitksan). I leave a fuller investigation of this to future research.

67

68 Tyler Peterson

Reported information is encoded in Gitksan using the reportative evidential =kat, which requires an inference in the form of a report (i.e. linguistic information). Contrast this with example (18), which is infelicitous in a reportative context, as ʼnakw requires an inference from sensory evidence (i.e. environmental information). The relevance for mirativity of both environmental and linguistic information, as well as the distinction between them, is expanded upon below, where I show that, whereas environmental information can be used miratively (i.e. the surprise expressed through evidentiality), linguistic information normally cannot be. 3.1.2 What it means for environmental information to be new With these two definitions of information in place, the next step is for us to determine what it means for information to be ‘new’. The distinction between environmental and linguistic information provides a useful means of defining what new information is. In a nutshell, what we are concerned with is new environmental information, and not new linguistic information. There is also another distinction related to this that requires further clarification: The following series of scenarios help illustrate the distinction between environmental vs. linguistic information, and why this is important. You and a friend have been planning on going berry-picking for some time, but you’ve been waiting for the berries to get ripe enough. Your friend has just returned from the berry picking grounds and calls you on the phone; she says ‘The berries are ripe.’ This is new information to you, but according to the definitions of information introduced above, this statement is new linguistic information: ‘the berries are ripe’ is an assertive speech act. The addressee interprets this sentence as a commitment on the part of the speaker that she has knowledge that the berries are, in fact, ripe. It is an entirely linguistic transaction. In Gitksan, this assertion is expressed in (20): (20) mukw=hl ’maay ripe=CND berries ‘The berries are ripe.’

Now consider a variation on this scenario very similar to the one discussed in the previous section: you and your friend observe people returning from the forest with buckets full of ripe berries; their boots are muddy, and their hands are purple. Crucially, the assertion in (20) is infelicitous in this context: the speaker does not actually know if the berries are ripe or not. Rather, a speaker infers from the evidence – the environmental information – they observe (i.e. full buckets, purple hands). The use of the evidential ʼnakw in the sentence in (21) is felicitous in this scenario:



Grammatical evidentiality and the unprepared mind

(21) ʼnakw=hl mukw=hl ’maay EVID=CND ripe=CND berries ‘It looks like the berries are ripe.’ ‘The berries must be ripe.’

But what precisely makes this environmental information new? As a basic characterization, environmental information is new when it has just come into awareness and not yet been integrated to the current knowledge state of a speaker (Dickinson 2000). However, this characterization can be sharpened by identifying two conditions on environmental information that make it new, as defined in (22): (22) Environmental information (as defined in (16)) is new relative to an information agent iff i. It has not been previously observed or perceived by that information agent. ii. It is spatio-temporally bound to the context it is observed in by that information agent.

(22) defines what I call new environmental information (NEI). In order to illustrate the effects of the conditions in (22), consider another scenario: Alvin is driving into the town to run some errands. He passes by Bob’s house and observes smoke rising from Bob’s wilpsehon (smokehouse). This observation provides environmental information (evidence) that Alvin uses to infer that Bob is smoking fish. It is also new information, as Alvin did not previously observe the smoke. In this context Alvin says to the friend accompanying him: (23) ʼnakw=hl se-hon-(t)=s Bob EVID=CND CAUS-fish-3=CND Bob ‘Looks like Bob is smoking/preparing/doing up fish.’ ‘Bob must be smoking fish.’

The evidential statement in (26) satisfies the first condition of (22). The second condition on the spatial-temporal boundedness is required to account for an effect that is predominant with evidentiality and the linguistic encoding of inference that is provided by the new environmental information. Upon returning from town, Alvin again passes by Bob’s house, and again observes the smoke. (26) is still felicitous, but the environment information is no longer new. Using this same scenario, the evidential sentence in (22) is markedly less felicitous if Alvin utters it to his friend just as they are returning to Alvin’s house, well out of sight of Bob’s smokehouse. This is because the felicity of evidential ʼnakw in (22) requires that the new information provided by the sight of smoke coming out of the smokehouse be spatially and temporally bound to the context of the utterance of (22).

69

70 Tyler Peterson

I suggest that the condition on spatio-temporal boundedness in (22ii) reflects a general property of indirect evidentials that encode inferences made from perceived environmental information (unlike reportative evidentials such as =kat in (19), which encode inferences based on linguistic information). It should also be noted that the condition on spatio-temporal boundedness is flexible in the sense that violations of this condition do not lead to outright infelicity. For example, evidentials are frequently sensitive to the spatial context they are used in, and the effects of the spatial boundedness condition can be felt in how an addressee can challenge a statement such as (26) when the conversation occurs outside of the context of the new information.10 Applying the smokehouse context in English also reveals the effect of the sub-condition in (22ii): while driving by Bob’s smokehouse the statement ‘It looks like Bob is smoking fish’ is less likely to be challenged than when it is uttered after returning to Alvin’s house, well out of sight of Bob’s smokehouse. This effect is examined in more detail in Section 5.11 The definition of NEI developed here is also consistent with what Nichols (1986) calls ‘immediate meaning’. Nichols identifies this meaning in the use of the Chinese Pidgin Russian marker est, which is used by a speaker “in a more or less spontaneous reaction to a new, salient, often surprising event as it just happens” (1986: 248). For example, the Chinese Pidgin Russian sentence in (24a) is uttered by a speaker while he is “standing at a riverbank as a storm approaches, [and] notices high water and remarks on it” (Nichols 1986: 250). Compare this to the sentence in (24b) without est, which does not convey any immediate reaction to a newly perceived situation: “[The speaker] has been aware for some time that a flood might take place” (Nichols 1986: 205–251; cited in Temürcü 2007: 71–72): (24) a. Voda pribavljaj est water rise est ‘The water’s RISING.’ (capitals in original)

10. See Peterson (2010a) for an analysis of this effect as a reflex of presupposition failure: a speaker using a sentence of the form ʼnakw(p) pragmatically presupposes that there is sensory evidence in the context which is verifiable by other discourse participants. 11. The effects of the boundedness condition can also be felt with expressions beyond evidentials. For example, as is well known, the interpretation of certain event expressions such as the present perfect are also constrained temporally when involving NEI. Upon looking out the window and seeing a taxi pull up, one can say ‘The taxi has arrived’, where the event of the taxi’s arrival provide NEI that has a certain ‘shelf life.’ After a period of time or further away from the original context, this sentence becomes less felicitous in favour of the simple past ‘the taxi arrived.’



Grammatical evidentiality and the unprepared mind

b. Kamni smotrju: voda pribavljaj rocks look water rise ‘I’m looking at the rocks: the water’s rising.’

The contrast between a simple assertion in (24b), and the same assertion marked with est in (24a) shows how NEI can be grammatically marked in a language and distinguished from non-new information.12 In fact, there are languages in which there is morphology that obligatorily marks the status of new information (among other properties such as control). Among these are the Tibeto-Burman languages (DeLancey 1992), and Tsafiki (cf. (8)). For example, the congruent morphology on the verb in Lhasa Tibetan in (25a) indicates that the speaker has just reached into his pocket and discovered he has some money: (25) Lhasa Tibetan (DeLancey 1992: 43–44) a. Old information (Congruent) ngar dngul tog=tsam yod I:DAT money some EXIST.CONGR ‘I have some money.’ b. New information (Non-congruent) ngar dngul tog=tsam ’dug I:DAT money some EXIST.NCONGR ‘I have some money!’

As observed earlier in this section, a sense of surprise often accompanies new information, as can be observed in the translation of (25b). Defining NEI achieves two things that are relevant for the examination of mirativity: first, it provides a way of contrasting information that a speaker acquires 12. It is likely that every language has some linguistic way of indicating what is new and what is old information by linguistic means (Keenan 1974). For example, one strategy for indicating the split between new and old information is pseudo-cleft constructions in Germanic languages, as in (1): (1) a. A: What did John buy? b. B: [What John bought]OLD was [a brand new car]NEW In (1b), the old information What John bought is presupposed in the pseudo-cleft, which syntactically precedes the new information a brand new car is at the end of the sentence. This provides the listener with a linguistic cue used to differentiate between what is old information and what is new information. Focus-sensitive expressions such as only, even, and also are also associated with new information. Beaver and Clark (2008), in a survey of focus-sensitive expressions in English, suggest that only, even, and also encode reference to the pragmatic category of NEW in their meaning, since they must be interpreted as associating with syntactically marked new information.

71

72

Tyler Peterson

through their own experience of the world (environmental information), and the information acquired by others (linguistic information). This distinction can be linguistically coded in languages with evidentials such as Gitksan, as (26) shows: (26) a. mukw=kat=hl ’maay ripe=REP=CND berries ‘[I heard] The berries are ripe.’ b. ʼnakw=hl mukw=hl ’maay EVID=CND ripe=CND berries ‘It looks like the berries are ripe.’ ‘The berries must be ripe.’

The reportative sentence in (26a) is felicitous if a speaker overhears a conversation at the coffee shop that the berries are ripe and ready for picking – the speaker is inferring the berries are ripe based on linguistic information in the form of a report. The sensory evidential ʼnakw in (26b) is infelicitous in a reportative context such as this. Rather, in (26b) the speaker is inferring the berries are ripe based on environmental information in the form of sensory evidence (i.e. people returning with muddy boots and purple hands). Because =kat is an evidential, we might expect it to be used miratively. However, this is not the case: (26a) cannot be used to express surprise. Distinguishing between environmental and linguistic information captures this observation, which also reflects a fairly robust cross-linguistic generalization: only first-person, non-firsthand or inferential evidentials are used miratively (Aikhenvald 2004: 207).13 As described above, it is this interpretative tension – the use of a non-first hand evidential with the first hand knowledge of the speaker – that drives the mirative use of an evidential. Secondly, there is an asymmetrical relation between NEI and surprise: surprising information always entails NEI, but NEI isn’t always surprising. Intuitively speaking, it is difficult to imagine a situation where some piece of information we’re already aware of can also be surprising. For example, the utterance of (26b) may be no more surprising than its English translation: both involve some kind of NEI, which may not be surprising at all if the current month is August, and the berries are usually ripe at this time.14 So, why is NEI sometimes surprising? 13. Under Aikhenvald’s classification of evidentials this excludes the reportative. However, Aikhenvald also provides some cases of the reportative used miratively (2004: 205–206). 14. It is possible to recreate a sense of surprise at a piece of information that has already been mentally assimilated by a speaker. For example, one might say ‘You’re smoking!’ as a surprised reaction (encoded by ‘surprised’ intonation in English) upon encountering a friend who recently stopped smoking. Having assimilated this surpising new information, one can later report this event to another friend by exclaiming ‘he was smoking!’ (also with the surprised



3.2

Grammatical evidentiality and the unprepared mind

Surprise and the unprepared mind

The central characteristic found in the various descriptions of mirativity is surprise. Given this, little attention has been given to the link between the linguistic encoding of surprise and the psychology of surprise (although see Slobin & Aksu 1982; Aksu-Koç & Slobin 1986). What is known about surprise originates mainly from the literature in psychology (for example Meyer & Niepel 1994; Meyer et al. 1997; Niepel et al. 1994; Reisenzein 2000), and more recently it has also been the focus of numerous investigations in the cognitive and computer sciences (Schützwohl 1998; Itti & Baldi 2005; Lorini & Castelfranchi 2007), language acquisition (Kagan 2002), and artificial intelligence (Macedo 2009). As a starting point, surprise can be characterized in two complementary ways. Firstly, surprise is one of the core human emotions, along with happiness, sadness, anger, disgust, contempt, and fear (Ekman 1980, 1984; Izard & Malatesta 1987).15 These specific core emotions are biologically determined emotional responses whose expression and recognition – typically through a combination of verbal and non-verbal cues seen in association, or what is called a syndrome – is fundamentally the same for all individuals, regardless of ethnic or cultural differences. Secondly, surprise can be viewed from an evolutionary and cognitive perspective: the concept of surprise is central to sensory processing, adaptation and learning, attention, and decision making. We can gain a novel insight into the linguistic nature of surprise through engaging in the vast body of research and analysis on the emotion of surprise, and the study of surprise involving non-­ linguistic behaviour. 3.2.1 Schema theory and surprise as a cognitive process Within psychology and cognitive research streams, the properties of the emotion of surprise have been effectively charted through the use of schema theory. According to schema theories (i.e. Mandler 1984; Rumelhart 1984), schemata are non-­linguistic formal objects that are organized knowledge structures representing concepts such as situations, objects, events, and actions at various levels of abstractness. Schemas can be thought of as categorical rules or scripts that we use to interpret the world. Information that does not fit into these schema may not be comprehended, or may not be comprehended correctly. Under this definition, schemas precisely correlate with the kind of speaker-level k-knowledge, as part of intonation). However, I believe this is akin to ‘reporting’ surprise, and not as a reaction of surprise to the previously assimilated information. 15. See Schützwohl (1998), Reisenzein (2000) for a comprehensive overview on the history and evaluation of research on surprise.

73

74

Tyler Peterson

the definition in (16) above. Recall the restaurant scenario from the introduction: if the waiter in a restaurant asked you if you would like to sing along with him, you might have a difficult time interpreting what he was asking and why, since singing is not something that patrons in a restaurant normally do. However, if you had been to the restaurant in the past and knew that it has an opera theme and staffed by opera students, and singing to customers is part of the job, you would have incorporated that information into your schema and not be surprised when the waiter asked if you’d prefer to sing. NEI is processed according to how it fits into these schema, which are used not only to interpret our environment through the inferences performed on this information, but also to predict outcomes of events or situations occurring in our immediate environment. This was shown in the restaurant scenario. However, the world is often a highly dynamic place, and in order to maintain a proper representation of reality there is a continuous feedback loop, where an individual continuously checks (mostly unconsciously) whether their schema matches the inputs coming from the surrounding environment (Scherer 1984). Within the framework of this process, it is fairly straightforward to see how as soon as inputs from this feedback loop diverge from the schema, a sense of surprise is a natural response. Let us examine this process more closely, as there are several notions that serve as points of contact with the linguistic encoding of surprise. New events – modelled here as NEI – can involve a deviation to some degree from activated cognitive schema, which are schemata that are immediately relevant and currently activated in the mind of the speaker (akin to situational awareness). This deviation creates a schema-discrepancy (Meyer et al. 1997; Schützwohl 1998; Reisenzein 2000). Using these building blocks, a cognitive model in Figure 1 on the following page outlines a series of six mental activities that occur during surprise. The initial state can be thought of as the state of the speaker at any given moment; it represents the current situational awareness (active cognitive schema), along with the capabilities of perception (sight, sound, tactile etc.) that constantly take in NEI. Surprising NEI is evaluated within this process as exceeding some threshold value of schema-discrepancy. In other words, upon perceiving some NEI, one or more of the person’s current expectations is violated to a significant degree.16 The consequence that follows is that the person experiences surprise. Motivated by the feeling of surprise, the person analyzes the surprising event, consciously verifying its unexpectedness, assessing the cause of the event, and

16. There has been a considerable amount of research on quantifying these degrees. See Lorini and Castelfranchi (2007) for details.



Grammatical evidentiality and the unprepared mind

Activated Cognitive Schema

Perception of NEI

Schema discrepancy check

Evaluation of NEI

Schema discrepancy causes surprise

Schema revision

Figure 1.  A cognitive model of the mental processes elicited by surprising events (adapted from Reisenzein 2000: 265)

deciding whether some response or change in plans is necessary. If a response is necessary, the person begins responding. The schema discrepancy check follows, and is responsible for monitoring and evaluating the active cognitive schema. When NEI is perceived by the individual it is evaluated as either being assimilable or unexpected. If it passes under a certain threshold, then it feeds back and updates the active cognitive schema. However, if this NEI counters an individual’s previous expectations, or is an event that they are not mentally prepared for, then it triggers a schema discrepancy. It is at this step that surprise is an elicited response to a discrepancy. The NEI is then eventually assimilated and fed back into the active cognitive schema. (27) 1. Schema: The speaker’s past experiences of similar situations and his general knowledge. This set of assumptions can range from knowledge about purely physical interactions to assumptions based on cultural and social norms (Dickinson 2000) → The role of a waiter in a restaurant. Activated cognitive schema: → Sitting in a restaurant on Davie street, assigning this role to various people waiting on tables.

75

76

Tyler Peterson

2. Unexpected events may involve a deviation to some degree from an activated cognitive schema: schema discrepancy check and evaluation of NEI 3.  Schema discrepancy causes surprise: The waiter asking you to sing along with him → Revision: → (Oh, this must be that opera-themed restaurant where the waiters sing.)

Schema theory provides us with a way of calculating the cognitive steps that elicit the emotion of surprise. It is also where we can explicitly account for the asymmetric relation observed above between new information and surprise: old information is a part of a person’s active cognitive schema and is used to appraise NEI, which may or may not lead to schema discrepancy. When it does, surprise is elicited; when it doesn’t, it simply updates an individual’s active cognitive schema. However, this is only part of the picture: the framework set out in Figure 1 treats surprise as a cognitive process; it isn’t explicit as to the presentation of surprise. Thus it is important to probe further these properties in order to shed light on why the elicitation of surprise can be a prominent enough feature to be linguistically encoded. 3.2.2 The linguistic correlates of schema discrepancy The emotion of surprise is often viewed as a syndrome, which is commonly defined as the association of recognizable physiological and behavioural features that often occur together to form a pattern. The patterns that comprise the surprised reaction manifest themselves on three distinct levels simultaneously, given in (28) (Reisenzein et al. 1996; Meyer et al. 1997; Schützwohl 1998; Reisenzein 2000): (28) The syndrome of surprise 1. THE SUBJECTIVE LEVEL: the subjective feeling of surprise; verbal exclamation of surprise. 2. THE PHYSIOLOGICAL LEVEL: different cortical response wave patterns; changes in heart and respiration rates; increased neural activation; etc. 3. THE BEHAVIOURAL LEVEL: subsequent curiosity/exploratory behaviour; specific facial expression; interruption of ongoing activities; etc.

It is at the subjective level that we find a direct channel for the linguistic expression of surprise. Unlike the physiological, and, to a lesser extent, the behavioural levels, the subjective level mostly involves volitional responses to a surprising event. This may come in the form of simple spontaneous vocalizations (i.e. ‘wow!’, ‘oh!’, etc.), or more complex constructions at the utterance level. This volitionality



Grammatical evidentiality and the unprepared mind

at the subjective level is reflected in the variety of linguistic strategies a speaker can use to express surprise. (29) Context: You and Gwen are preparing a surprise birthday party for Alvin. You don’t expect him home for another hour. However, in the middle of your preparations, Alvin walks through the door 1. Schema: “The speaker’s past experiences of similar situations and his general knowledge. This set of assumptions can range from knowledge about purely physical interactions to assumptions based on cultural and social norms” → Surprise birthday parties are only surprising if the birthday boy or girl doesn’t know you’re planning it, and you can surprise them when they show up. Activated cognitive schema: The party is for Alvin, the birthday boy; the party preparations are underway while Alvin is not there 2. “If the immediate [experience of an event or state] does not correlate well with the speaker’s expectations, the proposition coding the event or state receives special marking.” Schema discrepancy check: Alvin is there

The sentences in (30) are a sample of the variety of linguistic choices a speaker has to express surprise when a friend unexpectedly shows up at a party: (30) You made it! (Surprised intonation marked by ‘!’) I don’t believe you made it! (Referencing the schema-discrepancy) Looks like you made it! (Evidential verb, with/without ‘!’ intonation) Wow, you’re here! (Surprise vocalization, plus ‘!’ intonation) I’m amazed you made it!  (Verb of surprise) That can’t be who I think it is! (Root modality, negation, plus ‘!’ intonation) What a surprise (you’re here)! (Wh-exclamative, verb of surprise)

In addition to giving us a glimpse of the intra-linguistic diversity in marking surprise, the utterances in (30) also show us how a single schema discrepancy can have multiple linguistic realizations of surprise, for example, in the way information structure marking intonation (‘!’) can co-occur with an evidential verb. All of these expressions of surprise are projected from the same schema discrepancy: the speaker was unable to assimilate the NEI they were faced with in a specific context. It is important to elaborate on this claim. The cognitive locus of surprise is schema discrepancy caused by NEI, and the surprise conveyed through the sentences in (30) is a linguistic reflex of this. For example, in the English sentence ‘Alvin’s here!’, the speaker is not surprised at the proposition denoted by the sentence (Alvin’s here), rather, the speaker is surprised by the actual presence of

77

78

Tyler Peterson

Alvin – the NEI. The speaker then makes a subjective, linguistic choice in the special marking a sentence that registers this surprise – among the many other choices in (30). In the case of the sentence ‘Alvin’s here!’, this marking is a specialized intonational pattern. A cognitive treatment of surprise precisely connects Dickinson’s construals of mirativity (and by extension the other characterizations ascribed to mirativity given at the beginning of this section) in the following way: we now have the theoretical means to explain the cognitive features that underpin mirativity, specifically, a speaker’s current mental state (active cognitive schema), and how this state is transformed when faced with NEI that does not correlate well with the speaker’s current mental state (schema discrepancy). Schemata not only provide structures for a speaker’s existing network of assumptions and beliefs, but they also structure knowledge about oneself, grounded both in the present and past experiences of similar situations, and general knowledge (i.e. k-knowledge). Within any context these schemata can be activated at any time and used to evaluate NEI. It is now clearer that the term ‘surprise’, as it has been used here and in the literature on mirativity, is just a handle for what is essentially a single mental event (in a series of events) that occurs when a schema discrepancy is triggered. This single event corresponds to a variety of related emotions and psychological notions, including (but not limited to) surprise, unpreparedness, and unexpectedness.17 Thus, these meanings now form a natural class, which in turn reflects and confirms the kinds of meanings that researchers have used to describe mirativity in the beginning of Section 2. An analysis of mirativity as schema discrepancy predicts that, because all of these notions are derived from a single event (the discrepancy), these notions should not be mutually exclusive from each other. For example, a mirative marker in one language may have surprise as its dominant meaning, while implicating a secondary sense of unexpectedness, or vice versa. This variability can be viewed as a reflection of the individual nature of some schemata, the universality of others, and the always unique nature of a discourse context.18 Additionally, because NEI is what triggers the schema discrepancy, NEI is necessarily one of the meanings of 17. What is notably missing from this list is lack of intention and control – two kinds of meaning frequently claimed to be a part of mirativity (cf. Dickinson 2000). I set this issue aside for future research, although see Dickinson (2000) for an explanation of how control marking implicated surprise. 18. This variability may be further conditioned by the semantics of the morpheme or construction that that language uses to encode mirativity. For example, the mirative use of an exclamative construction, as in ‘What a nice guy he is!’ also implicates a scale of ‘niceness’, in addition to encoding a speaker’s sense of surprise or unexpectedness.



Grammatical evidentiality and the unprepared mind

mirativity, and, in fact, in some languages it may even be the dominant interpretation. This explains the observation that surprising or unexpected information always entails NEI. As another illustration of this prediction, consider a situation in Turkish, where Hande and her friend walk through the door and see Hakan sitting on the couch. Hande has one of two linguistic options for responding to this situation in (31): (31) a. Hakan bura-da, şimdi gid-ebil-ir-iz Hakan here-LOC now go-MODAL-PRES-1pl ‘Hakan’s here, so we can go now.’ b. Hakan bura-da(y)-mış, şimdi gid-ebil-ir-iz Hakan here-LOC-EVID now go-MODAL-PRES-1pl ‘Hakan’s here, so we can go now.’

In (31a), using the simple past, Hande expected Hakan to be there, since they made previous plans for the three of them to go out together; the fact that Hakan is there is not new information to her. In (31b), using evidential -miş, Hande was not expecting Hakan to be there. Hande and her friend had an extra ticket to the theatre, and were looking for a third person to go to the theatre with; since Hakan happened to be there, they can go together. In this context Turkish consultants report that the sense of surprise with this use of -miş in (31b) is secondary, nor is its evidential meaning prominent. Rather, in this context (31b) is simply marking that this information is new to Hande, which may further optionally implicate that this was unexpected for her. Thus, the sentence denoting the proposition that Hakan came receives special marking using -miş instead of the regular past tense. Framed within the cognitive analysis of surprise, a schema discrepancy is triggered by the new information Hande encounters, as, based on her past experience, Hakan is usually at school at this time. Using the detailed contexts provided by Nichols (1986), a similar result obtains in the Chinese Pidgin Russian in (24) above, a paraphrase of which is repeated below in (32). In the context of watching the rising water a speaker has a choice of using a simple declarative in (32a), or they register the fact that this is new information with (32b), which is the same sentence formally marked with the particle est: (32) Chinese Pidgin Russian (Nichols 1986: 250) a. … voda pribavljaj … water rise ‘… the water’s rising.’

79

80 Tyler Peterson

b. Voda pribavljaj est water rise est ‘The water’s RISING.’

Recall that Nichols describes the sentence in (32b) as uttered by a speaker while he is standing at a riverbank as a storm approaches, and notices high water and remarks on it, whereas (32a) without est, does not convey any immediate reaction to a newly perceived situation, as the speaker has been aware for some time that a flood might take place (1986: 250). Similar to Turkish -miş, the particle est may be viewed primarily as a marker of new information, which may also be surprising to the speaker. Example (32) supports the claim that the speaker is surprised at the NEI perceived in this context, and not at the proposition denoted by the sentence in (32b). More specifically, the NEI is the difference in water levels, and the speaker infers from this that the water is rising. This NEI is then embedded within a cognitive process which triggers a schema discrepancy, perhaps given the speaker’s experience with water levels, it’s too early in the season for such high water (the activated cognitive schema). The evidence for the distinction between being surprised at NEI (the schema discrepancy), and the marking of a sentence as being new or surprising (mirative marking) is the one-to-many correspondence shown in the English example (30). In the speech context of (30) there is one schema discrepancy – the NEI of the friend’s arrival at the party – and many different kinds of propositions that can reflect this discrepancy. The sentences that denote these propositions are where the special linguistic marking of this discrepancy is made, which we label mirativity. This analysis of the one-to-many correspondence between surprising NEI and the linguistics expressions of surprise also re-enforces the point made about the ‘directness’ of NEI in Section 3.1. Returning to the scenario in (17) above, you can see John standing at the window of his house, which provides the NEI that John’s at home, which is then linguistically realized through the miratively marked sentence ‘John’s home!’ (perhaps because you expected him to be at work). However, the exact same NEI can lead to the mirative expression ‘He’s ready already!’. This NEI is instead surprising because you had made an appointment to go out with John, and he’s always late getting ready. This shows the one-to-many correspondence between one kind of NEI and the many kinds of discrepancies that can arise, each of these with their own mirative outcomes.



Grammatical evidentiality and the unprepared mind

4. Conclusion The aim of this paper was to explain how evidentials are used to express the unprepared mind. Evidentially is the linguistic encoding of first person, speaker-­ oriented information that is used to infer the truth of a statement, without knowing if this statement is true or not. Mirativity, as it is currently understood in the literature, is also a first person, speaker-oriented phenomenon which marks a statement as being surprising or unexpected. Evidentials can be used to express mirativity, but it is not a part of their primary meaning. Therefore, it was shown how an explanation of how evidentials can be used miratively requires the theoretical tools that can model information, speaker knowledge, and a process that can explain the cognitive transformation a speaker undergoes when perceiving new and surprising information. The outcome of this approach is an analysis of mirativity as the cognitive process of surprise, and the marking of mirativity is the linguistic reflex of this process. Using this, the mirative use of an evidential is licensed by the speaker’s knowledge of a proposition in a context, which has this process embedded within it. This explains how evidentials are used miratively. The next step was to explain why evidentials are used this way. It was claimed that the mirative use of an evidential has a similar pragmatic function to a rhetorical question in conversation: evidential-as-mirative statements do not necessarily contribute to the discourse. In addition, miratives use evidential meaning as a kind of gestural deixis, the purpose of which is to draw the attention to the other conversational participants. 4.1

Future prospects

The empirical focus of this paper was on the mirativity implicated through evidentiality. However, mirative interpretations can project from a wide variety of syntactic and morphological constructions, interjections, lexical items, control marking, illocutionary adverbs, information structure marking such as intonation, wh-exclamatives, expressives, and different speech acts. Even language-­ internally, the quick survey of how mirativity is conveyed in English in (30) revealed a wide variety of ways of how one can express surprise when a friend unexpectedly shows up at a party. Many of these expressions appear at face value to be independent of any other semantic or grammatical category, or what can be defined as non-­parasitic mirativity. If this is indeed the case, we expect non-­parasitic miratives to have complementary properties to parasitic miratives (where mirativity is ‘parasitic’ on another grammatical category such as evidentiality), while sharing many of the same properties. This opens up another avenue

81

82

Tyler Peterson

of research into mirativity, as well as providing a point of contact with research already underway on the formal syntactic, semantic and pragmatic properties of constructions such as exclamatives and exclamations (for example Zanuttini & Portner 2003; Portner & Zanuttini 2005; Rett 2011). 4.2

Mirativity and other categories

This investigation has direct implications for the definition of mirativity through an entirely different grammatical and semantic category: control. DeLancey (1992: 43–44) suggests that the conjunct/disjunct pattern of Lhasa Tibetan and some other Tibeto-Burman languages is a grammaticalization of a non-mirative/ mirative distinction, an example of this is given in (33):19 (33) a. ngar dngul tog=tsam yod I:DAT money some EXIST.CONGR ‘I have some money.’ b. ngar dngul tog=tsam ’dug I:DAT money some EXIST.nCONGR ‘I have some money!’

Dickinson (2000) extends this analysis to Tsafiki. According to Dickinson, the congruent/incongruent distinction in Tsafiki (which basically corresponds to the ‘congruent’ and ‘non-congruent’ in Lhasa Tibetan) primarily codes the degree to which the state or event is in agreement with the speaker’s general knowledge system. Congruent indicates that the information contained in the proposition is congruent with the speaker’s general knowledge. Non-congruent indicates that the information is not congruent with the speaker’s general knowledge. For example, both congruent and non-congruent forms in example (34) code an accident. But with the congruent form (34a), the speaker was in a situation where a fall was quite likely, i.e. walking on slippery ground. The speaker was conscious of the danger and conscious of the fall while it was occurring. The non-congruent form (34b) indicates that the speaker had no premonitory consciousness that he was going to fall. He suddenly slips and falls. (34) a. la ke-ere-i-yo-e 1MASC throw-send-become-CONGR-DECL ‘I fell.’ (non-mirative, expected) 

(Dickinson 2000: 393)

19. Although see Curnow (2000) who argues that conjunct/disjunct is not grammaticalized mirativity.



Grammatical evidentiality and the unprepared mind

b. la ke-ere-i-i-e 1MASC throw-send-become-NONCONGR-DECL ‘I fell!!’ (mirative, unexpected) 

(Dickinson 2000: 394)

The surprise at the event of falling naturally follows from the speaker’s lack of awareness at the dangers of falling. This sense of surprise is even made more acute by the use of an unaccusative verb, in which the grammatical subject is the semantic patient. However, with unergative and some transitive predicates, a speaker exercises a degree of agentivity over the event. In (35a) the congruent form indicates that the behaviour was intentional. The speaker intended to kill the pig. The non-congruent form (35b) indicates that the speaker accidentally killed the pig. Perhaps he ran over it with his car: (35) a. la kuchi=ka tote-yo-e 1MASC pig-ACC kill-CONGR-DECL ‘I killed the pig.’ (intentionally) b. la kuchi=ka tote-i-e 1MASC pig-ACC kill-NONCONGR-DECL ‘I killed the pig.’ (unintentionally)

The occurrence of non-congruent constructions Lhasa Tibetan and Tsafiki indicates surprise, accident, ignorance or irony, and occurs in complementary distribution with the congruent constructions (Dickinson 2000: 389). A non-congruent marker indicates that there is a contradiction between the speaker’s set of established expectations and assumptions and one or more factors in the immediate experienced situation. The contradiction is most commonly, but not necessarily, surprising. The meanings associated with (non)congruent morphology cast mirativity in a different light, and reveal how surprise reflects something of the conscious experience of a speaker and their control over states, events and actions. In the normal course of experience, an observer has a set of expectations concerning these, but when an event or state does not meet those expectations, the speaker of a language with congruent/incongruent morphology must explicitly mark the information contained in the proposition as occurring outside the normal course of experience. It is beyond the scope of this paper to investigate the semantic properties of congruent and non-congruent marking in detail. However, I suggest that the form and function of (non)congruent marking is not grammaticalized mirativity, but

83

84

Tyler Peterson

instead connects to the wider cross-linguistic category of control (Haller 2000).20 Many languages morphologically encode the volitionality of a subject and their ability to exercise ‘control’ over state, event, or action. For example, in the Salish language St’át’imcets, there is verbal morphology dedicated to marking predicates with a similar range of meanings, such as accident, suddenly, and non-controllable (Davis et al. 2009). Depending on the semantics of the verb, eventualities denoted by a verb that are non-controlled correlate with notions such as the lack of intention or involvement on the part of the speaker. Control is a different kind of meaning from evidentiality that is often associated with mirativity. So, is there a way of unifying the mirative use of an evidential with the kinds of mirativity described for Lhasa Tibetan and Tsafiki? From a purely conceptual perspective, control can be viewed as standing in a relation with evidentiality and mirativity when framed from the perspective of the speaker, the knowledge they have for making statements about the world around them, and the perceived control they have over actions, events, or states they experience. These kinds of meanings, and how mirativity fits in with these meanings are summarized in (36): (36) i. EVIDENTIALITY: encodes the source of the speaker’s knowledge for a state, event, or action. a.  Lexical evidentiality: encodes the specific type of evidence (i.e. perception, report, etc.) a speaker has for an assertion. b.  Tense-aspect: encodes the temporal flow in an action, event, or state denoted by a proposition, from the speaker’s perspective. ii. CONTROL: encodes the speaker’s ability to act as an agent or with volitionality over a state, event, or action. iii. MIRATIVITY: encodes the relationship between a proposition and the speaker’s overall expectations and assumptions with regards to an immediate state, event, or action. (Dickinson 2000: 382)

Languages have various strategies for linguistically encoding the kinds of meanings in (36). This also means that grammars may also conflate these meanings to various degrees through their linguistically encoding. From this perspective, there are aspects of both evidentiality and control meaning that are amenable to conveying mirativity. However, an information-­ based analysis of mirativity as presented in the previous section predicts only a subset of the meanings associated with control marking can be considered as truly mirative, such as example (33) and (34). As mirativity is analyzed as a cognitive 20. See Häsler (2001) for a useful overview of the approaches to the semantics of the Tibetan verb stem.



Grammatical evidentiality and the unprepared mind

process involving NEI, control meanings involving the volitionality of the agent of an action such as (35) are excluded, as these kinds of meanings do not involve information of any kind; rather, they involve a speaker’s intentions or abilities. As such, the mirative readings of non-congruent morphology in both Lhasa Tibetan and Tsafiki are parasitic on the primary control meaning of incongruent morphology in exactly the same way as mirativity is parasitic on the primary evidential meaning of lexical evidentials. Further research on control in these languages is required to see if these predictions hold. Mirativity can also arise via other grammatical categories that have evidential uses. It is well documented that in many languages the perfect aspect has an indirect evidential interpretation in addition to its aspectual one (Izvorski 1997). Thus, it is not surprising that we find mirativity associated with perfect-as-­ evidential constructions as well. This can be observed in unrelated languages such as Bulgarian (37), where the perfect morphology can carry a mirative meaning in addition to its aspectual and evidential meaning: (37) Maria celunala Ivan Maria kiss-PE Ivan ‘Maria (has) kissed Ivan!’

In another case where temporality plays a role in projecting mirativity, Faller (2003) argues that the evidential interpretation of the past tense marker -sqa in Cuzco Quechua has a mirative interpretation. In example (38) the speaker has direct evidence for the described eventuality, and would usually be uttered at the time the speaker is perceiving the surprising situation: (38) Kay-pi-(má) ka-sha-sqa Marya-qa this-LOC-SURP be-PROG-NX.PST Marya-TOP ‘Marya is here!’

What these examples from two unrelated languages indicate is that mirativity is parasitic on evidentiality, whether realized through lexical evidentials, or the covert evidentiality associated with tense or the perfect aspect. The mirative use of aspect also provides supporting empirical evidence for the definition of NEI (cf. (22)), specifically, the second sub-condition: in order for environmental information to be new, it must be spatio-temporally bound to the context it is observed in by that information agent. The semantics of the perfect are ideally suited to fulfilling this condition (see Koev 2011 who also addresses this issue). Surprise is one of the few universal human emotions. Likewise, mirativity is probably a linguistic universal: all languages have the means for linguistically encoding mirativity. This paper represents one of the first studies dedicated to the

85

86 Tyler Peterson

theoretical analysis of mirativity, and also supports this hypothesis. In addition to filling this gap, this study is important because of the bridge it creates – which, at the moment, does not exist – between the linguistic encoding of surprise and the purely psychological and cognitive streams of research on the emotion of surprise.

References Aikhenvald, A., & Dixon, R. M. W. (Eds.). (2003). Studies in evidentiality. Amsterdam: John Benjamins.  doi:  10.1075/tsl.54 Aikhenvald, A. Y. (2004). Evidentiality. New York: Oxford University Press. Aksu-Koç, A., & Slobin, D. (1986). A psychological account of the development and use of evidentials in Turkish. In W. Chafe & J. Nichols (Eds.), Evidentiality: The linguistic coding of epistemology. Advances in Discourse Processes XX (pp. 159–167). Norwood, NJ: Ablex. Bar-Hillel, Y., & Carnap, R. (1953). An outline of a theory of semantic information. Reprinted in Bar-Hillel [1964]. Barwise, J., & Seligman, J. (1997). Information flow: The logic of distributed systems. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511895968 Beaver, D., & Clark, B. (2008). Sense and sensitivity: How focus determines meaning. Oxford: Blackwell.  doi: 10.1002/9781444304176 Chafe, Wallace L., & Nichols, J. (Eds.). (1986). Evidentiality: the linguistic coding of epistemology, Advances in Discourse Processes XX. Norwood, NJ: Ablex. Curnow, T. (2000). Why ‘first/non-first person’ is not grammaticalized mirativity. In Proceedings of als2k, the 2000 conference of the australian linguistic society. Davis, C., Potts, C., & Speas, M. (2007). The pragmatic values of evidential sentences. In M. Gibson & T. Friedman (Eds.), Proceedings of the 17th Conference on semantics and linguistic theory, (pp. 71–88). Ithaca, NY: CLC Publications. Davis, H., Matthewson, L., & Rullmann, H. (2009). A unified modal semantics for out-of-­ control in St’át’imcets. In L. Hogeweg, H. de Hoop, & A. Malchukov (Eds.), Proceedings of the tamtam conference. DeLancey, S. (1990). Notes on evidentiality in Hare. International Journal of American Linguistics, 56(1), 152–158.  doi: 10.1086/466142 DeLancey, S. (1992). The historical status of the conjunct/disjunct pattern in Tibeto-Burman. Acta Linguistica Havniensia, 25, 39–62.  doi: 10.1080/03740463.1992.10412277 DeLancey, S. (1997). Mirativity: The grammatical marking of unexpected infor- mation. Linguistic Typology, 1, 33–52.  doi: 10.1515/lity.1997.1.1.33 DeLancey, S. (2001). The mirative and evidentiality. Journal of Pragmatics, 33, 369–382. doi: 10.1016/S0378-2166(01)80001-1 Dickinson, C. (2000). Mirativity in Tsafiki. Studies in Language, 24, 379–422.   doi: 10.1075/sl.24.2.06dic Egerod, S. & Hansson, I. L. (1974). An Akha conversation on death and funeral. Acta Orientalia, 36, 225–284. Ekman, P. (1980). Asymmetry in facial expression. Science, 209, 833–834. doi: 10.1126/science.7403851



Grammatical evidentiality and the unprepared mind

Ekman, P. (1984). Expression and the nature of emotion. In Approaches to emotion (pp. 319– 344). Hillsdale, NJ: Lawrence Erlbaum. Faller, M. (2002). Semantics and pragmatics of evidentials in Cuzco Quechua. Stanford dissertation. Faller, M. (2003). Propositional- and illocutionary-level evidentiality in Cuzco Quechua. In Proceedings of semantics of the under-represented languages of the americas 2. Amherst, MA: GLSA. Floridi, L. (2011). Semantic conceptions of information. In E. N. Zalta (Ed.), The stanford encyclopedia of philosophy, Spring 2011 edn. Friedman, V. A. (2003). Evidentiality in the Balkans with special attention to Macedonian and Albanian. In A. Y. Aikhenvald & R. M. W. Dixon (Eds.), Studies in evidentiality Typological Studies in Language 54. Amsterdam: John Benjamins.  doi: 10.1075/tsl.54.11fri Geurts, B. (2010). Quantity implicatures. Cambridge: Cambridge University Press.   doi: 10.1017/CBO9780511975158 Haller, F. (2000). Dialekt und erzählungen von shigatse. In D. Schuh (Ed.), Beiträge zur tibe­ tischen erzählforschung 13. Bonn: VGH Wissenschaftsverlag. Häsler, K. (2001). An empathy-based approach to to the decription of the verb system of the Dege dialect of Tibetan. Linguistic of the Tibeto-Burman Area, 24(1), 1–34. Itti, L., & Baldi, P. F. (2005). A principled approach to detecting surprising events in video. In Proceedings of ieee conference on computer vision and pattern recognition (cvpr) (pp. 631– 637). San Siego, CA. Izard, C. E., & Malatesta, C. Z. (1987). Perspectives on emotional development I: Differential emotions theory of early emotional development. In Handbook of infant development (pp. 494–554). New York: Wiley. Izvorski, R. (1997). The present perfect as an epistemic modal. In Proceedings of salt, vii, 222–239. Jacobsen, W. (1964). A grammar of the washo language. Berkeley: University of California dissertation. Kagan, J. (2002). Surprise, uncertainty, and mental structures. Cambridge, MA: Harvard University Press. Keenan, E. L. (1974). Logic and language. In M. Bloomfield & E. Haugen (Eds.), Language as a human problem (pp. 187–196). New York: W.W. Norton & Co. Koev, T. (2011). Evidentiality and temporal distance learning. In A. Chereches, N. Ashton, & D. Lutz (Eds.), Proceedings of semantics and linguistics theory (salt), 21, 95–114. Kratzer, A. (1991). Modality. In D. Wunderlich & A. von Stechow (Eds.), Semantics: An international handbook of contemporary research (pp. 639–656). Berlin: de Gruyter. LaPolla, R. (2003). Evidentiality in Qiang. In A. Y. Aikhenvald & R. M. W. Dixon (Eds.), Studies in evidentiality, Typological Studies in Language 54, Amsterdam: John Benjamins. doi: 10.1075/tsl.54.06lap Lazard, G. (1999). Pour une terminologie rigoureuse: Quelques principes et propositions. Mémoires de la Société de Linguistique de Paris, (6), 111–133. Lorini, E. & Cristiano C. (2007). The cognitive structure of surprise: looking for basic principles. Topoi: An International Review of Philosophy, 26(1), 133–149. Macedo, C. A., Reisenzein, R., Lorini, L., & Castelfranchi, C. L. (2009). Artificial surprise. In J. Vallverdú & D. Casacuberta (Eds.), Handbook of research on syn- thetic emotions and sociable robotics: New applications in affective computing and artificial intelligence (pp. 267– 291). Hershey, PA: IGI Global.

87

88

Tyler Peterson

Mandler, G. (1984). Mind and body. New York: Norton. McCready, E., & Norry O. (2007). Evidentiality, modality and probability. Linguistics and Philosophy, 30(2), 147–206.  doi: 10.1007/s10988-007-9017-7 McKinlay, S. (2008). Information, knowledge and confirmation holism. In A. Briggle, P. Brey, & K. Waelbers (Eds.), Current issues in computing and philosophy (collected essays). IOS Press. Meyer, W.-U., & Niepel, M. (1994). Surprise. In Encyclopedia of human behavior, vol. 3, California: Academic press. Meyer, W-U., Reisenzein, R., & Schützwohl, A. (1997). Toward a process analysis of emotions: The case of surprise. Motivation and Emotion, 21(3), 251–274.  doi: 10.1023/A:1024422330338 Nichols, J. (1986). The bottom line: Chinese Pidgin Russian. In W. Chafe & J. Nichols (Eds.), Evidentiality: The linguistic coding of epistemology. Advances in Discourse Processes XX (pp. 239–257). Norwood, NJ: Ablex. Niepel, M., Rudolph, U., & Schötzwohl, A. (1994). Temporal characteristics of the surprise reaction induced by schema discrepant visual and auditory events. Cognition and Emotion, 8(5), 433–452.  doi: 10.1080/02699939408408951 Palmer, F. 2006. Mood and modality, 2nd edn. Cambridge: Cambridge University Press. Papafragou, A., Li, P., Choi, Y., & Han, C. (2007). Evidentiality in language and cognition. Cognition, 103, 253–299.  doi: 10.1016/j.cognition.2006.04.001 Peterson, T. (2010a). Epistemic modality and evidentiality in Gitksan at the semantics-pragmatics interface. University of British Columbia dissertation. Peterson, T. (2010b). Examining the mirative and non-literal uses of evidentials. In T. Peterson & U. Sauerland (Eds.), Evidence from evidentiality, Vol. 28, Vancouver: University of British Columbia Working Papers in Linguistics. Peterson, T. (2012). The role of the ordering source in Gitksan modals. In E. Bogal-Allbritten (Ed.), Proceedings of sula 6: Semantics of under-represented languages in the Americas. Amherst, MA: GLSA Publications. Peterson, T., & U. Sauerland (Eds.). (2010). Evidence from evidentiality, Vol. 28. Vancouver: University of British Columbia Working Papers in Linguistics. Portner, P., & Zanuttini, R. (2005). Nominal exclamatives in english. In R. Stainton &R. Elugardo (Eds.), Ellipsis and non-sentential speech, Studies in Linguistics and Philosophy (pp. 57–67). Dordrecht: Kluwer Academic Publishers. Reisenzein, R. (2000). The subjective experience of surprise. In H. Bless & J. Forgas (Eds.), The message within: The role of subjective experience in social cognition and behavior. Philadelphia, PA: Psychology Press. Reisenzein, R., Meyer, W.-U., & Schützwohl, A. (1996). Reactions to surprising events: A paradigm for emotion research. In N. H. Frijda (Ed.), Proceedings of the 9th conference of the international society for research on emotions (pp. 292–296). Toronto: ISRE. Rett, J. (2011). Exclamatives, degrees and speech acts. Linguistics and Philosophy, 34(5), 411–442.  doi: 10.1007/s10988-011-9103-8 Rooryck, J. (2001a). Evidentiality, Part I. GLOT International, 5, 3–11. Rooryck, J. (2001b). Evidentiality, Part II. GLOT International, 5, 161–168. Rumelhart, D. E. (1984). Schemata and the cognitive system. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition, Vol. 1 (pp. 161–188). Hillsdale, NJ: Erlbaum. Scherer, K. R. (1984). On the nature and function of emotion: A component process approach. In Approaches to emotion (pp. 293–317). Hillsdale, NJ: Lawrence Erlbaum. Schützwohl, A. (1998). Surprise and schema strength. Journal of Experimental Psychology, Learning, Memory and Cognition, 24(5), 1182–1199.  doi: 10.1037/0278-7393.24.5.1182



Grammatical evidentiality and the unprepared mind

Slobin, D., & Aksu, A. (1982). Tense, aspect, and modality in the use of Turkish evidential. In P. J. Hopper (Ed.), Tense-aspect: Between semantics and pragmatics, Typological Studies in Language, (pp. 397–405). Amsterdam: John Benjamins.  doi: 10.1075/tsl.1.13slo Speas, M. (2008). On the syntax and semantics of evidentials. Language and Linguistics Compass, 2(5), 940–965.  doi: 10.1111/j.1749-818X.2008.00069.x Temürcü, C. (2007). A semantic framework for analyzing tense, aspect, and mood – an application to the ranges of polysemy of -xr, -dir, -iyor and -∅ in Turkish. Universiteit Antwerpen dissertation. Wiklund, A.-L. (2009). The syntax of surprise: Unexpected event readings in complex predication. Working papers in scandinavian syntax 84, 181–224. Willett, T. (1988). A cross-linguistic survey of the grammaticization of evidentiality. Studies in Language, 12(1), 51–97.  doi: 10.1075/sl.12.1.04wil Zanuttini, R., & Portner, P. (2003). Exclamative clauses: At the syntaxsemantics interface. Language, 79(1), 39–81.  doi: 10.1353/lan.2003.0105

89

Operationalizing mirativity A usage-based quantitative study of constructional construal in English Karolina Krawczak and Dylan Glynn

Adam Mickiewicz University, Poznań / University of Paris 8 Saint Denis

This study focuses on the conceptual category of mirativity and its constructional construal in English. We propose an operationalization of mirativity with a view to investigating the phenomenon within the usage-based quantitative methodology of multifactorial analysis (Geeraerts, Grondelaers, & Bakema, 1994; Gries, 2003). The proposed operationalization is founded on two usage dimensions, i.e., the degree of performativity of the utterance and the degree of incongruity of the described event. It is argued that mirativity, in its prototypical form, can be operationally defined as a combination of high levels of these two variables. The feasibility of this operationalization in usage-based quantitative research is tested in a case study investigating three surprise-encoding constructions in English: [what + the + np], [what + a + np] and [to + my + np]. The data, amounting to 350 observations of the three constructions, were extracted from dialogic online diaries and submitted to detailed manual annotation and subsequent multivariate statistical modeling. The results reveal a usage continuum ranging from [what + the + np] through [to + my + np] to [what + a + np] relative to the high degrees of performativity and incongruity. Keywords: corpus linguistics, grammatical constructions, mirativity, multivariate statistics, profile-based approach, usage-feature analysis

doi 10.1075/bct.92.04kra © 2017 John Benjamins Publishing Company

92

Karolina Krawczak and Dylan Glynn

1. Introduction: Operationalizing the conceptual-functional category of mirativity The present study is a usage-based investigation of the constructional construal of mirativity in English.1 Mirativity, understood here to be a subcategory of epistemicity, encodes the speaker’s attitudinal state of surprise with respect to a given aspect of the interactive event (DeLancey, 1997, pp. 35f., 2012, p. 533; Dendale & Tasmowski, 2001, p. 343). More specifically, mirative markers indicate that some information is not readily integratable with the speaker’s overall epistemic system and expectations (DeLancey, 2001, p. 380). In other words, in some respect or other, it diverges from the picture of reality projected by the experiencer. On the epistemic cline, running from certainty, through disbelief (including mirativity), to lack of confidence and, finally, total rejection, the latter three categories lie in the domain of irrealis (Akatsuka, 1985, p. 625). By this, it is meant that these categories encode different degrees of the speaker’s uncertainty with regard to a given aspect of the surrounding world. In the present study, this notion of an epistemic cline leads us to argue that mirativity can be perceived as a scalar phenomenon. We propose three arguments for this. Firstly, a given aspect of the unfolding reality can exhibit different degrees of incongruity with the speaker’s established understanding of the world. This conceptual dimension of incongruity is here used to refer to the divergence of the conceptualized situation from what the speaker would consider the expected and normal state of affairs: The greater this epistemic divergence, the higher the incongruity. Secondly, the category of mirativity, verging between epistemic certainty and uncertainty, can manifest varying degrees of enactment of the state of incongruity, which will here be referred to as performativity. This functional parameter of usage can be defined in terms of the presentation or stylization of the experienced situation: the more emotionally-charged, engaging and emphatic this presentation, the more performative the utterance. Thirdly, the perspective adopted by the speaker to construe the surprising event is also likely to vary, exhibiting different degrees of subjectivity. This latter notion denotes the orientation of the utterance on and its grounding in the speaker’s standpoint. This third argument will not be taken up in the present study but warrants investigation in future research. Mirativity, as a conceptual-functional category, is inherently onomasiological in nature. Although onomasiological, or function-first, approaches to grammar and lexis are fundamental to many theoretical paradigms (cf. Bondarko, 1991; 1. We wish to express our gratitude to two anonymous reviewers for their constructive comments. Any remaining shortcomings are our own.



Operationalizing mirativity

Talmy, 2000; Grondelaers & Geeraerts, 2003, inter alia), they all suffer from an inevitable weakness: since the descriptive point of departure is a non-­observable concept or function, delimiting the object of study exactly is problematic. Such an approach necessarily entails lengthy debates as to what constitutes an instance of the object of study, here a mirative construction. For this reason, operationalization, or definition in terms of measurement instead of truth-values (Stefanowitsch, 2010), is arguably a necessary step in mirativity research. This study does not claim to propose an adequate operationalization of the category that would be valid across all languages, but does seek to make the first steps in that direction. The exact specifications of mirativity will naturally depend on its linguistic encoding and the overall context. One may assume, however, that the more enactive/performative the instantiation of the epistemic state of surprise and the higher the level of conceptual incongruity, the more mirative the given expression should be. This is the central claim of the present article forming the basis upon which to understand mirativity and formulate testable hypotheses. Accordingly, the semantic-pragmatic category of mirativity, in its most typical form, is here operationalized as a combination of: i. a high degree of functional performativity; ii. a high degree of conceptual incongruity. It should be stressed that the mirative category is understood as the simultaneous instantiation of both incongruity and performativity. In order to test the feasibility of the operationalization, we examine three lexico-­syntactic constructions that express surprise, or epistemic incongruity. The aim will be to quantitatively describe their use and, therefore, the conceptual-­ functional profiles of each construction in their relation to each other. An inherent limitation lies in the fact that we have no way of determining the mirative nature of the constructions independently of the study. This means that both the accuracy of the descriptions and the accuracy of the operationalization in capturing mirative structures cannot be categorically determined. Instead, the operationalization will be deemed successful if its application proves systematic and coherent and if the results are accurate, that is, quantitatively predictive.

93

94 Karolina Krawczak and Dylan Glynn

2. Object, data and method 2.1

Object: Three constructions for encoding epistemic incongruity

The descriptive study is based on three lexico-syntactic constructions that can be used to express epistemic incongruity. The aim here is to determine what degree of mirativity the constructions encode. The three constructions are: Construction A [what + the + np], Construction B [what + a + np], and Construction C [to + my + np]. The nominal slot in Construction A typically licenses a range of taboo lexemes such as fuck, hell, heck. These lexemes are also used as expletive exclamatives and it is likely that there is some constructional inheritance involved. Indeed, the original construction may simply have been the exclamative Fuck!, where the lexical alternatives are non-taboo substitutes for the taboo term. This interpretation would explain why many taboo items, such as shit, are not licensed in this position. It is further supported by the fact that simple exclamatives can also encode epistemic incongruity and indeed, arguably, mirativity. However, explaining the lexical licensing of the constructions is not the goal of the present study and so such questions can be left open. Although obvious, it should be stressed that a construction is not necessarily associated with a single usage anymore than a lexeme. Therefore, any statement about the semantics of the constructions assumes that this form-meaning pairing involves semasiological variation. Construction B licenses a range of nouns that literally or figuratively designate surprise. In this, the construction is lexically close to Construction C, which also takes surprise nouns. It follows that, formally at least, the three constructions are closely related: Construction A is similar to Construction B syntactically, while Construction C is similar to Construction B lexically. If one accepts Construction Grammar’s Inheritance Theory (Goldberg, 1995), one might argue that B inherits syntactic structure from A and lexical structure from C, making it a hybrid of A and C. Another possible developmental path would be for Construction A to inherit its syntactic structure from B, the latter still receiving its lexical structure from C. Figure 1 is a visualization of these potential inheritance relationships. Of course, it might well be that the three constructions developed largely independently of each other, over time coming to encode, among other functions, the speaker’s epistemic state of surprise. Further discussion about the relationship between the constructions is offered in Section 3. However, it should be noted here that there is no suggestion that a given construction is uniquely associated with any given usage.



Operationalizing mirativity

Cx. A [what + the + exclamative]

Cx. C [to + my + surprise]

Cx. B [what + a + surprise]

Figure 1.  Formal relationships between constructions

2.2

Data: Interactive online language use

The corpus employed in this study is the LiveJournal Corpus of American and British English (Speelman, 2005). This is a genre specific corpus made up entirely of online personal diaries. Although the social dimensions of speaker age, gender and social background are unknown, the corpus is strictly composed of two components, British and American. The language is highly informal, spontaneous2 and typically consists of diary entries describing personal events. Although the text is entirely composed of diary entries, the language is characterized by dialogic use. This latter feature is due to the interpersonal nature of the texts, where the authors are aware that their blogs are publically available and open to comments and discussion.3 This fact, combined with the spontaneous nature of the language, offers some similarity to live interaction. This is especially important when dealing with such intersubjectively negotiated categories as epistemicity in general or mirativity, in particular. Nonetheless, due to the written nature of the data, the utterance is by default distinct from the event that triggers the state of surprise. This means that purely performative verbal responses to conceptually incongruent situations are a rarity. The mirative event is, therefore, in a sense, nearly always reported or re-enacted for the addressee post factum. The speaker-cum-writer chooses to employ an expression that simulates a mirative utterance, assumedly for stylistic effects and better expressivity. Interestingly, however, there are some examples in which the

2. The language used in LiveJournal can be considered spontaneous, as it tends to approximate natural language production. This is true in two respects. Firstly, the blogs do not seem to be edited by the authors, as evidenced by frequent typos, rewording or what may be regarded as unintended repetition of words (i.e., clearly not for emphatic purposes). Secondly, the language appears to reflect the speaker’s naturally unfolding stream of thought or consciousness. Although this is a subjective impression of the text, being impossible to verify, the impression is systematic across the majority of entries. 3. The dialogic nature of the blogs is evidenced, for example, by the common use of linguistic devices that are normally characteristic of spoken dialog whose function involves inviting replies or addressing specific followers of the journal, etc.

95

96 Karolina Krawczak and Dylan Glynn

expression of epistemic incongruity is coextensive with the event causing it, as evidenced, to varying degrees, by (1)–(3): (1) I am starting my next assignment for our writer’s group, even though I was sure I would do it earlier this time. Where does the time go. I am doing some research, wow what a surprise that is for me. (2) what the heck, this is the 4th journal that i have posted and the other two arn’t there!!! Uggg. this one better be there when im done. (3) Okay so i’m trying to register for classes and it’s not letting me on! what the heck!!! I guess every other freshman on this campus is tryin to as well. Ugggg! Well, I got bored and decided to go to facebook … nothing special so I checked my email … once again nothing

In these examples, the state of surprise concerning the speaker’s online activity is expressed simultaneously (or nearly so) with the experience itself. This is especially true in the case of (1) and (2), where it may be claimed that the state of epistemic incongruity is expressed at the same time as it arises.4 It should be noted though that in (1), the temporal scope of the experience may well go beyond the here-and-now of its externalization, incorporating the extended present time in which the research is being conducted. In (3), on the other hand, there is likely to be a small delay between the experience and its verbalization. Methodologically, this study bridges quantitative corpus-driven research and close qualitative text-discourse analysis. For this reason, it is important to stress that the manual annotation necessitated substantial contextualization of the investigated construction. Despite all due care, the identification and subsequent annotation of the categories under investigation remain highly subjective. To illustrate the importance and difficulty of this point, consider examples (4) and (5): (4) Whoa! what the heck was that? Did you just feel the earth move under your feet?

The immediate impression one is likely to get after reading the example is that this utterance is indeed nearly-simultaneous with the event causing it, an earthquake, which also shows that the statement is liable to receive a literal reading. However, once we take more of the surrounding contextual information into account, as is 4. In uses such as those presented in (2) or (3), it might be argued that the state of epistemic incongruity is no longer a case of unadulterated surprise; rather, it may well be mingled with that of annoyance, which commonly follows negative instances of surprise. However, a comparable epistemic-emotional fusion will be true of those instances of surprise where the evaluation of the event is positive. In such cases, the epistemic state of surprise will be likely to be tinged with happiness. The emotive state that the incongruity evokes should be a separate concern from the incongruity itself.



Operationalizing mirativity

the case in (5), it becomes apparent that the utterance should be understood figuratively and that the expression of surprise is not at all co-extensive with whatever causes it: (5) today’s horoscope: Capricorn: Whoa! what the heck was that? Did you just feel the earth move under your feet? Lately it feels like foundations that you take for granted no matter what are being awfully shifty and unreliable. It’s important to remember that in this world, the only constant element is change. So while it feels like everything is in a state of upheaval, hold onto a sense of perspective and you’ll be just fine.

The data for the study, comprising approximately 350 examples, were extracted proportionally – equal numbers of occurrences per construction and per dialect. Although the constructions license a wide range of lexemes in the nominal slot, this study is restricted to a small set of the most frequent lexical instantiations. These lexemes include fuck, hell and heck for construction A, surprise and shock for construction B, and surprise, shock and horror for construction C. In addition, construction A subsumes the acronym wtf, standing for ‘what the fuck’, which is commonly used in informal written communication that makes up the corpus. The examples were also checked to make sure they encoded the experience of the speaker. Although this could be performed automatically for Constructions B and C, there were many instances of quotatives in Construction A. This was especially common with the construction-lexeme pairing what the fuck, where speakers would use the expression to describe a third person’s reaction. An example of this kind of use would be ‘He was like “what the fuck”’. Therefore, it was necessary to check the sample manually to ascertain that all the occurrences encoded the speaker’s own experience. 2.3

Method: Constructional profiles and multifactorial feature analysis

The aim of this study is firstly to test an operationalization of the category of mirativity and secondly to offer a quantitative usage-based description of three constructions that express epistemic incongruity in English. The descriptive study employs Multifactorial Usage-Feature Analysis or the Profile-Based Approach (Dirven, Goossens, Putseys, & Vorlat, 1982; Geeraerts, Grondelaers, & Bakema, 1994; Gries, 2003; Divjak, 2010; Glynn & Fischer, 2010; Glynn & Robinson, 2014). This method comprises three steps: 1. Corpus – A large sample of contextualized examples of a given linguistic phenomenon 2. Usage-Features – Manual annotation of examples for a range of usage-features

97

98 Karolina Krawczak and Dylan Glynn

3. Behavioral Profile – Multivariate statistics is used to establish the frequency-­ based patterns of the use of the linguistic phenomenon under investigation. Importantly, the final step, given sufficient data, can also ascertain statistical significance, the effect-size of the specific variables, and the predictive power (descriptive accuracy) of the identified patterns. The present study uses three statistical techniques. Patterns of usage, indicative of the behavioral profile, are first identified with the use of multiple correspondence analysis (cf. Glynn, 2014). The statistical significance of the correlations that underpin these patterns is tested for with loglinear analysis (cf. Krawczak & Glynn, in press). Finally, the descriptive accuracy of the patterns is calculated with use of fixed-effects polytomous logistic regression (cf. Krawczak, 2014a, 2014b, 2014c). The underlying assumption behind the use of Multifactorial Feature Analysis and the Profile-Based Approach is that contextualized language structure, as registered in corpora, affords an index of the conceptual structuring of the linguistic phenomenon under scrutiny, here mirativity. For a comparable method, based on elicited data rather than usage data, components analysis has been developed in psychology (Scherer, 2005; Fontaine, Scherer, & Soriano, 2013; Soriano, Fontaine, & Scherer, this volume). Although there is an established tradition of applying this method for the analysis of grammatical constructions (Gries, 1999; Heylen, 2005; Bresnan, Cueni, Nikitina, & Baayen, 2007, inter alia), these studies have usually focused on syntactic alternations. A syntactic alternation results in a binary response variable for which quantitative techniques have been established. Previous research on a network of constructions has been restricted to exploratory multivariate statistics (Glynn, 2008, 2009, 2010). Multinomial outcomes, as is required for the confirmatory modeling of a constructional network, have thus far only been applied to lexical relations (Arppe, 2008; Nordmark & Glynn, 2013; Krawczak, 2014a, 2014b). Although extending these statistical techniques to a field of near-­ synonymous constructions should be unproblematic, it has yet to be undertaken. 3. Analysis. Manual annotation of sample The present study, aiming to determine the usage profiles and measure the mirative status of the investigated constructions relative to a specific operationalization of the phenomenon, examines two major factors that are directly informative in this regard, namely, the degree of incongruity and the degree of performativity. The former conceptual variable concerns the extent to which a given event is incongruent or surprising, while the functional dimension of performativity refers



Operationalizing mirativity

to how (re)enactive the presentation of the given situation is. In both cases, three levels are distinguished, i.e., strong, neutral and weak. All the examples in the dataset were manually annotated for these two factors as well as a range of other functional variables (Speaker Evaluation, Speaker Engagement, Stimulus, Temporal Scope of the Stimulus), which, due to space limitations, are not included in the present study. It must be stressed that the qualitative analysis, consisting in manual annotation, is a meticulous procedure. It is for this reason that the sample is restricted to 350 occurrences. 3.1

Conceptual-functional mirative profile of constructions

Having established the formal relations between the constructions, let us move to the qualitative description of how these constructions might profile the conceptual-functional category of mirativity. This will permit the formulation of a set of hypotheses that will then be tested quantitatively in Section 3. At first glance, constructions A and B could both be taken to represent the more performative instances of expressing conceptual incongruity, whereby the speaker encodes his or her epistemic stance vis-à-vis a given unexpected development or situation in a dynamic and direct manner. Such a dynamic and enactive construal could be claimed to be imposed by the exclamative what, which is most likely to be selected to encode the experience of surprise in the immediate circumstances in which it arises. It is as if the subject, by choosing this exclamative expression, (re-)enacts his or her disbelief, rather than describing it to the interlocutor. Given this, these two constructions could be viewed as prototypically mirative, especially when accompanied by other means of expression, such as capitalization, the use of exclamation marks or multiple question marks and other emphatic elements, e.g., the use of direct speech, repetition, emoticons or some emotionally-charged acronyms such as LOL. These highly expressive features are illustrated in examples (6) and (7), both instantiating Construction A: (6) she doesn’t wanna be married any more?? She then proceeded to make me feel worse by telling me that she doesn’t see me as her husband because she is not as into marriage as I am. WHAT THE FUCK IS THAT??!! If any of you need someone to make you feel like complete shit, gimmie a call and I’ll let u talk to my wife. (7) I went into the back to count my till and stuff and then Stephanie comes in (and I had never worked with her before) she like screams ““Oh my gosh, you have the most perfect nose”” I was like ““what the heck?”” LoL it was really funny lol… im like ummm okay never heard that one before lol.

99

100 Karolina Krawczak and Dylan Glynn

In example (6), the speaker discusses a highly emotionally-loaded issue and employs capitalization as well as multiple question marks and exclamation marks to intensify their pragmatic import. In example (7), on the other hand, the reported incident is not so much particularly significant to the speaker as clearly highly unexpected. This is evidenced not only by the use of what the heck, but also by the use of direct quotes of what was being communicated and what the speaker’s thoughts were, which is accompanied by the acronym LoL, repeated three times. As mentioned in Section 2.1, Construction B is formally related to Construction A, the main difference being the definite – indefinite article alternation.5 Intuitively, there might be some reason to believe that Construction A carries a higher performative load. This could be argued to be a result of the lexical slot. In contrast to Construction B, Construction A licenses expletives such as fuck or hell, which are also used as exclamatives in their own right. Although the contribution of the lexical item to the performativity of the construction appears self-evident, the role of the definite article is less obvious. At this point, a closer investigation of the structure of the construction would need to consider both its diachronic development and its relationship with daughter constructions such as ‘what the exclamative is that (doing here)’ etc. Despite the fact that such a line of inquiry would be interesting and might well be informative, it is beyond the purview of the current study. Construction B differs from Construction A on two accounts. Syntactically, it uses the indefinite article, which renders the described situation or, rather, the epistemic experience, much less focused, i.e., less determinate or specific. Lexically, it takes, in the nominal slot, one of the synonymous expressions instantiating surprise. In the present study, these include surprise and shock. These lexemes clearly lack the expletive element typical of the lexemes associated with Construction A, thus being much more descriptive of the speaker’s epistemic state. These two lexico-grammatical characteristics can be argued to attenuate the performativity of Construction B. It could also be assumed that this construction will be favored over Construction A in circumstances that are less incongruent with the speaker’s epistemic system of expectations and assumptions. For these reasons, Construction B may be expected to figure lower on the scale of mirativity. Its usage is illustrated in (8) and (9):

5. The other difference between the two constructions, as discussed in Section 2.1 and elaborated on below, is not syntactic in nature. It relates to the specific noun licensed. The divergence consists in how in Construction A, the nouns have a clearly expletive function, while in Construction B, they are more descriptive of the epistemic state experienced by the speaker.



Operationalizing mirativity 101

(8) I had totally forgotten about an exam I had taken a few weeks back which would allow me to do new procedures at work. Well today I got an envelope with the results in.. what a shock. Overall I got 80% but unbelievably for my practical exam I got 11/10?!? How is this possible I hear you ask? Well so did I, so I asked my manager who explained it to me. The Home Office examiner (nice lady that she is) was so impressed with my technique ; ) she gave me an extra point. (9) In the morning met up with Adri and ended up going to tardy tank with her and Zuly. what a surprise eh? And of course same people as always were there hahah.

The situation described in (8) indeed appears to come as a shock to the speaker – s/he is genuinely surprised by it – but the immediate co-textual features in which what a shock occurs are considerably less performative than those we have observed in (6) and (7) above. The expression itself is not accompanied by multiple exclamation or question marks, capitalization or other parallel emphatic features. Such characteristics do, however, occur in the sentence that follows, where it becomes clear that what a shock really does encode the speaker’s disbelief. Another feature that is noteworthy is the question directed at the addressee. This is particularly interesting in that such an addressee-engaging element in the utterance, which we might refer to as heightened addressee-orientation, is also present in (9) in the form of the interactively employed exclamative eh. The presence of such features, perhaps at the expense of a higher degree of performativity, might suggest that Construction B is oriented more on the interactive event and the addressee, while Construction A is more readily focused on the speaker, his/her subjective experience and the venting thereof. Despite this commonality present in both (8) and (9), in the latter example, the expression what a surprise clearly encodes a very low level of incongruity, being used in a rather ironic manner. As a matter of fact, this non-literal use is not uncommon in the observations of [what + a + np] found in the dataset compiled for this study. Such sarcastic usage is hard to imagine in the case of Construction A, which might well be taken as another characteristic determining the more prototypical status of [what + the + np] vis-à-vis the category of mirativity. Construction C, on the other hand, could be said to instantiate a more objective and descriptive designation of a given incongruent event. Here, the speaker adopts a less dynamic encoding system and by direct reference to him- or herself by means of the possessive adjective my, treats him- or herself as an object of conception, an onstage participant of the conceptualized scene, to use Langacker’s

102 Karolina Krawczak and Dylan Glynn

(1987) parlance.6 It may be assumed that such a relatively low level of enactive dynamism and high level of descriptivity is likely to be accompanied by an equally low level of incongruity, or, at least, a considerable distance, temporal or conceptual, toward the conceptualized event. In this respect, construction C, lacking in performativity and low in immediate incongruity, might be regarded as marginally, if at all, mirative. Sentences (10)–(11) can serve as an example of the low level of performativity and relatively high descriptivity of [to + my + np]. (10) I got talking to the new boy; his name was Jeph Howard. He was really nice to my surprise even though he had somewhat of a strange look about him. (11) I bought the Sigma lens on Saturday, much to my surprise (another £320 gone).

In both these sentences, the speaker reports on the surprising event, without enacting the situation. The incongruity of the event might have been strong as the event unfolded, but its immediacy has since waned to the extent where the speaker can distance him/herself from it and approach it from an external vantage point, treating him/herself as a participant of the described circumstances. Other than passage of time, which removes the intensity of the experience of surprise, another dimension that may be decisive in the choice of a less dynamic and enactive construal is the event itself and its importance to the conceptualizer. The more significant the conceptualized event is to the speaker, the more likely it becomes that the construal will tend toward high performativity. The use of the descriptive expression to my surprise in (10) and (11), rather than a more performative phrase, may be argued to be the result of a less emotionally engaging theme and an event that, in the end, is not all that surprising to the speaker at all. In other words, the overall importance and incongruity of the described situation can be ranked as low. It, therefore, comes as no surprise that the construal imposed by the speaker is less dynamic and enactive. These proposed differences between the three constructions are presented in Table 1. The first column lists the constructions, while the following two columns specify the hypothesized relations obtaining between each construction vis-àvis the functional dimension of performativity and the conceptual dimension of epistemic incongruity. 6. By the same token, Constructions A and B can both be claimed to impose a more subjective construal on the experiencer in the sense that their expression of epistemic incongruity does not make explicit reference to the speaker and so the link must be inferred. This feature may also be among the conditions that permit the ironic use of Construction B, where its meaning encoding surprise can be removed or reversed. Arguably, such use is impossible in the case of prototypically mirative Construction A and prototypically non-performative Construction C. This subjectivity-objectivity discussion, however, is beyond the scope of the present study.

Operationalizing mirativity 103



Table 1.  Hypotheses for conceptual-functional profile of epistemic incongruent constructions Construction

Functional dimension – Degree of Performativity

Conceptual dimension – Degree of Epistemic Incongruity

A: what + the +np B: what + a + np C: to + my + np

High Performativity Moderate Performativity Low Performativity

High Incongruity Moderate Incongruity Low Incongruity

In accordance with the above discussion, Construction A is hypothesized to rank relatively high on the functional scale of performativity and equally high for the conceptual parameter of epistemic incongruity. Constructions B and C, on the other hand, are each expected to be comparatively lower on this conceptual-­ functional continuum. This continuum is put forward here as a yardstick against which to gauge the degree of mirativity encoded in a given expression. 3.2

Operationalizing performativity and incongruity

Let us now illustrate these two conceptual-functional categories and their respective levels with examples from the corpus. We start with incongruity. Below are three entries from the online blogs, each representing a different degree of incongruity of the conceptualized circumstances relative to the speaker’s overall accepted world knowledge or expectations. (12) This weather is weird. I went out last night, after dark, to take the dogs on a potty break. I was in a sleeveless shirt and my pajama bottoms. AND I WASN’T COLD! what the heck? Crazy crazy craziness. I should probably be out taking advantage of the beautiful weather, out playing with the kids or something more fun. But we have been total hermits this week. (Strong Incongruity) (13) i said geography is probably the most important one, and she said so whens that in for, i said tomorrow, shes like so its in for today? and im like no. shes like so okay its in for today, what lesson? im like WTF (in my head) and i just shrug. shes like ““well i suggest you go and speak to your teacher at break”” i was like okay… (Neutral Incongruity) (14) im failing physics..woah what a surprise -note the sarcasm- and im failing geometry -gasp- yeah so this week was just boring man. (Weak Incongruity)

In example (12), there are a number of usage-features that implicate the relatively high level of incongruity between the specific situation presented by the speaker, i.e., the fact that the temperature was warm enough for him/her not to feel the

104 Karolina Krawczak and Dylan Glynn

cold, and his/her expectations to the contrary. The post begins with the description of the weather as weird, then explains what the speaker was doing and wearing. This is followed by the exclamation AND I WASN’T COLD!, whose pragmatic leverage is additionally reinforced by the use of capitalization. Next comes the expression of surprise, encoded by what the heck?, strengthened in its import by the words ensuing it, which involve emphatic repetition, Crazy, crazy craziness. In (13), the speaker’s disbelief in his or her interlocutor’s capacity for misunderstanding what is being communicated is moderate. The speaker does not seem to attach too much importance to the event. This is evidenced by the modification in my head, which indicates that the experience of epistemic incongruity was not verbalized in front of the other person. What is more, the description of the physical response to the interlocutor’s behavior demonstrates that the short-lived state of epistemic incongruity was shrugged off. Finally, in example (14), despite the use of the exclamative woah, which might be taken as a highly performative form of giving vent to one’s surprise, the construction what a surprise is followed directly by the negating element of note the sarcasm. This cancels altogether the encoded meaning of incongruity, showing that failing the specific subject at school does not at all come as a surprise for the speaker. It was to be expected. The other functional variable, performativity, is also subdivided into the same three levels specifying its intensity, i.e., strong, neutral and weak. Example (15) below illustrates the first degree of performativity. (15) *STATE OF PANIC* HEART ATTACK PEOPLE … my chinese written test is NEXT TUESDAY WTF?!?! I thought it was like a week or two away, not 5 days! BLODDDDY HELL! (Strong Performativity)

Here, the speaker puts on a real show for the readers of the blog, expressing not only the mental state of incongruity, in this case a strongly negative experience, but also, in an exaggerated manner, the other accompanying effects, such as panic. Special effects are added by the use of such emphatic features as capitalization or repeated exclamation and question marks. The exclamative, BLODDDDY HELL, accentuates the (re-)enactive character of the utterance even further. The next level of neutral performativity is exemplified by (16) below: (16) today I got this official letter thru the post. When I opened (to my shock) it was a letter about my future pension! I was like ““WTF? I’m gonna be 21 tomorrow! Not bloody 61!”” Still, the words haunt me even now. (Neutral Performativity)

Compared to (15), this post is rather moderate in its performativity. It contains two realizations of the state of incongruity, both referring to the same event, i.e., the unexpected and rather premature reception of a pension portfolio. The former surprise-encoding construction, to my shock, is parenthesized, which significantly

Operationalizing mirativity 105



weakens its pragmatic thrust, so much so that it might well qualify as instantiating weak performativity. The latter expression, WTF, accompanied by a single question mark, is relatively lower on the performativity scale than its counterpart in (15). Two other features are noteworthy here. On the one hand, WTF is preceded by I was like, which is a rather descriptive phrase, commonly used to describe the observation-based surprise-induced reaction of other people rather than that of the speaker. On the other hand, however, WTF is embedded in a direct quote allowing the addressee to enter the immediate scene which is epistemically incongruent with the speaker’s expectations. This example, therefore, combines less performative usage-features, i.e., I was like, with more performative ones, i.e., the use of direct speech. The presence of these latter characteristics is further foregrounded by the use of two exclamative comments concerning the speaker’s age and the age of a person eligible for retirement, with the latter containing the emphatic lexeme bloody. This does not come close to the performative nature of (15), but, altogether, it does mean that the expression of surprise is moderate in its re-enactment/performativity, rather than weak. Finally, in (17) below, the utterance can be said to be weakly, if at all, performative, as it does not contain any special characteristics that would make it re-enactive in any way. An additional characteristic is that the expression is positioned medially in the utterance, functioning primarily as a modifying agent of the feeling of sadness, rather than an expression of the epistemic state of surprise considered in its own right. (17) I drove home out of seaford for the last time on friday night, with much to my surprise a distinct feeling of sadness. (Weak Performativity)

The two descriptive dimensions, discussed and exemplified above, serve as the basis upon which mirativity is operationalized. In line with this operationalization, it is hypothesized that a direct proportional relation will obtain between the increase in the semantic category of mirativity and the degrees of performativity and incongruity. Before turning our attention to the quantitative modeling of the linguistic behavior of the three constructions, let us consider the functional category of descriptivity and its relation to that of performativity. 3.3

The problem of the functional category of descriptivity

The discussion thus far suggests that performativity and descriptivity are related. One assumption is that this relation could take the form of two opposed poles on a conceptual-functional continuum. Although it may be possible that performativity and descriptivity are related, in this section, we argue that the two language

106 Karolina Krawczak and Dylan Glynn

functions do not lie on a mutually exclusive continuum. Since the role of descriptivity is not included in the analysis, its omission warrants justification. Nuyts (2001), in his analysis of epistemic predicates such as think and believe, sees the distinction as not only opposed but discrete. For example, the utterance I think is performative, but the utterance she thinks is descriptive. Although we are using the terms in a slightly different way, the idea that these ‘kinds of functions’ are opposed, be that discretely or upon a cline of degree, enjoys wide currency in Functional Linguistics. Indeed, it is precisely this kind of difference that Halliday’s (1985) discrete and opposed ideational and interpersonal meta-functions are designed to capture. If performativity and descriptivity are mutually exclusive, then including descriptivity as a parameter in the analysis of mirativity would be crucial. However, do we have reason to juxtapose these two functions a priori? Deductively, is there an argument that demonstrates that an utterance cannot be both performative and descriptive simultaneously? Despite the difficulty in providing a clear definition of the category of mirativity, at a theoretical level, there would be no disagreement over the fact that descriptivity is clearly distinct from mirativity. If this is the case, understanding the role of descriptivity and identifying its presence should be crucial to the identification and characterization of mirative structures. However, empirically, it would appear that performativity does not automatically preclude descriptivity, and vice versa. To illustrate this important theoretical and descriptive point, let us consider example (18): (18) After just two minutes, I was nearly bald above my ears, but had bushels of hair billowing over the side of my head like a bad bowl cut. She tried to compensate by evening out the sides with the top using scissors, so I waited patiently, watching my changing reflection in the mirror. Then, to my horror and quite without warning, she pronounced my haircut finished. Although I probably should’ve just left before this butcher could do any more damage, I asked her to take a little more off the top, so that it would be even with the sides instead of looking like I had extra brains hanging off the sides of my head. Resigned to my fate, I watched in dismay as she took a razor to the top of my head.

Arguably, the speaker here is simultaneously expressing his/her epistemic state of surprise in the performative sense, by staging the scene in a most vivid and humorous manner, and describing it, by using what seems to be the objective-descriptive encoding of surprise, to my horror. Among the phrases that add to the highly expressive character of the utterance, we can list butcher referring to the hairdresser, in dismay denoting the speaker’s emotional state or the witty description of the dreaded haircut as looking like … extra brains hanging off the sides.



Operationalizing mirativity 107

Examples such as this would suggest that a discrete opposition between the two notions should not be assumed. In contrast to example (18), the utterances in (19) and (20), although employing the more performative construction of [what + a + np], are much weaker in their performative import. (19) the pictures she posts, apparently not realizing that i can also see her bulletins. and she is a survey freak so i get a lot of insight into my darling eldest childs world. there’s a lot of drinking going on. hmmm… what a shock. in my defense, i never drank around my kids except on special occasions and then never to excess. (20) Some talk show, mom and leslee are watching..Lol Okay.. this weekend?! Leslee popped into town.. what a surprise, in a way. On Saturday, we hung out around the house, and then mom wants to trade in her neeeew truck!!

Here, despite weaker performativity, there is no evidence that descriptivity increases. In other words, if the two functions were at two ends of a continuum, one would expect that a decrease in one function would result in an increase in the other. Although such an important claim warrants further research, the examples presented here would suggest that the categories of descriptivity and performativity are independent of each other. An additional observation should be made here with regard to the conceptual dimension of incongruity encoded in examples (19) and (20). This is a point that refers us back to what has already been stated about Construction B in Section 3.1, namely that, based on the examples gathered for the present study, [what + a + np] clearly manifests potential for semantico-pragmatic weakened or even reversed usage, where instead of designating surprise, the construction may come to denote complete lack thereof. In the case of these two examples, the incongruity is not denied altogether, but rather its extent is toned down by the use of hmmm in (19) and in a way in (20). On the basis of such observations, one might conclude that lower degrees of epistemic incongruity might be reflected in lower levels of performativity, which, as indicated above, does not necessarily imply higher levels of descriptivity. Overall, this would suggest a direct proportional relation between the conceptual dimension of incongruity and the functional parameter of performativity, regardless of the descriptivity. This could be taken as further evidence for the conceptual-functional independence of these performativity-­incongruity categories and the category of descriptivity. For the purposes of the current study and its aim to develop an operationalization of the category of mirativity, we will leave aside the problem of the relationship between the ideational – descriptive function and the interpersonal –

108 Karolina Krawczak and Dylan Glynn

performative function. Due to the uncertainty surrounding the role of descriptivity, the case study presented below omits this dimension in its analysis. However, the above examples strongly suggest that future research will need to explicitly bring this dimension into the understanding and description of mirativity. Let us now proceed to the next step of multivariate quantitative modeling. This stage will enable us to identify usage-patterns in the multidimensional metadata that results from the detailed manual annotation of the examples. 4. Results This section presents the results of the quantitative analysis of the annotated data. This is done in two steps. We first employ an exploratory method in the form of multiple correspondence analysis, which is then followed by the confirmatory stage, where log-linear analysis and polytomous logistic regression modeling are used. The former exploratory technique, which reduces the multiple-dimensional space to two dimensions, identifies patterns of language use based on association of features, where the relative proximity of data points is indicative of their degree of correlation. This method, however, does not tell us whether the correlations (associations or disassociations) are statistically significant. To establish how descriptively accurate the analysis is and whether we can make any predictive generalizations, we use the two confirmatory methods. It is their role to determine the significance and importance of the individual usage-features in predicting a given outcome level, i.e., one of the three constructions. 4.1

Identifying usage patterns

The plot in Figure 2 visualizes the correlations emerging between the three constructions, i.e., [what + the + np], [what + a + np] and [to + my + np] relative to the degree of performativity and the degree of incongruity. The size of the individual data points is indicative of their contribution to the overall structuring of the distribution of the data. In Figure 2, from a coarse-grained perspective, we can see that the three constructions are located in three different quadrants of the plot. This means that they form distinct groupings, determined by relatively clear correlations of usage-­ features. In the bottom left-hand quadrant, [what the np] is closely associated in use with both strong incongruity and strong performativity. Recalling that our operationalization of mirativity proposed a definition based on a combination of a high degree of both these pragmatic variables, this correlation may be taken to

Operationalizing mirativity 109



Degree of Incongruity – neutral 2

To PRO NP 1

Degree of Performativity – neutral Degree of Performativity – weak 0

Degree of Incongruity – strong

W a NP

Degree of Performativity – strong W the NP

–1

Degree of Incongruity – weak 0

1

2

3

Figure 2.  Multiple correspondence analysis: Construction, performativity and incongruity

mean that this construction is highly mirative in its use. Example (21) is typical of the kind of usage identified by the correspondence analysis here. (21) AND the wierdest thing just happened. My mother called a lil while ago, and we were just doing the normal talk about what we both did today and whatever and out of nowhere I started CRYING. I don’t even know why. She wasn’t being mean to me or ne thing! I think I REALLY miss my mom and am REALLY homesick. what the hell doode?! I don’t even LIKE my mom, but maybe she’s growing on me.

This utterance is both emphatic and emotionally charged, as evidenced by the use of capitalization and intensifying adverbials to express some of the more important landmarks of the conceptualized event. The speaker here goes well beyond mere description of the situation. She puts it onstage and re-enacts it for the readers. We can nearly ‘hear’ her complete shock as she enunciates the capitalized words. She also chooses to additionally justify her epistemic state of incongruity, expressed by what the hell doode?! with an explanation as to why the event is incompatible with her overall expectations. As we go further up in the plot, we find the second construction, [to my np], situated in what may appear as isolated conceptual context, given the absence of

110 Karolina Krawczak and Dylan Glynn

other features in this quadrant of the plot. This, however, is not at all the case. What the position indicates is that the construction occupies a middle conceptual field stretching from neutral incongruity, to be found in the top right-hand quadrant, to its stronger instantiations, as placed in the left-hand quadrant along the x axis. This latter aspect can be exemplified by (22), where the speaker clearly expected a different outcome of his or her efforts: (22) For the first time in 3 years i decided to actually properly revise instead of cramming, and to my surprise i didnt do too well lol. I got 3 D’s and in the only exam i crammed i got an A. I think that is meant to tell me not to try and revise and be organized for my exams in the future.

Overall, the construction emerges as neutral in its performativity, a correspondence to be found in the top right-hand quadrant . This means that, by and large, this construction is not characterized by distinctly enactive properties. This finding is in accordance with our hypothesis, which posits a considerably less performative profile of this encoding of surprise. Finally, in the right-hand quadrant, the last grouping emerges for the construction [what a np], which is still lower on the scale of mirativity, being closely related to weak epistemic incongruity and equally weak or neutral performativity, as evidenced in utterances (23) and (24), respectively. This is an interesting finding, which stands in sharp contrast to what we postulated in Section 3.1, where this construction was expected to be relatively close to [what the np] in its mirative value. As the plot demonstrates, it here emerges as the least mirative realization of the state of epistemic incongruity. Examples (23) and (24) illustrate how these associations of usage-features come together to produce the emergent behavioral tendency. (23) I had so much on my mind this morning. We had an early appointment so there was no time to put in an LJ post and now most of it is gone from my head. what a surprise (24) im failing physics..woah what a surprise -note the sarcasm- and im failing geometry -gasp- yeah so this week was just boring man.

In both examples, the conceptualized event clearly does not come as a surprise to the speaker, who, given either general expectations or specific awareness of his/ her own skills, uses the expression what a surprise with a purely ironic intent. It is completely bleached of the sense designating epistemic incongruity. This usage characteristic relates to our discussion in Section 3.1, where it was noted that unlike either Construction A or Construction C, the instantiations of [what + a + np] are the most likely candidates for ironic uses. In utterance (24), this pragmatic function of ironizing is additionally augmented by the explicit reference to



Operationalizing mirativity

sarcasm. In both cases, the report can also be said to be much more descriptive than performative with respect to the conceptualized scene. There is, however, one feature that diverges from such a descriptive account, namely the use of the exclamative woah. In natural speech, it would, in all likelihood, be marked by appropriate prosody, such as lengthened, nasal and stressed intonational patterns with a clear negative coloring (Winner et al., 1987). The same sort of prosodic contour would accompany the surprise-encoding expression itself. The absence of such means of expression, typically reserved for the speech mode, may be the main motivating factor for the blogger to make the sarcastic character of the statement overt through the interjected cue note the sarcasm. This way, any misunderstanding is avoided. 4.2

Confirming usage patterns

We will now turn to two confirmatory methods in order to evaluate the reliability of the patterns identified above. We begin with loglinear analysis, a technique that identifies statistically significant correlations and anti-correlations. We then test the overall descriptive accuracy of those correlations with a polytomous logistic regression. Figure 3 presents a mosaic plot that visualizes the results of the loglinear analysis, i.e., the interactions obtaining between the three constructions relative to the functional dimension of performativity and the conceptual dimension of incongruity. In this visualization, significant associations and disassociations are represented by means of colors, where blue indicates the degree of attraction, while red stands for repulsion. A three-way set of correlations is included, where the interaction of each of the levels of the degree of incongruity and performativity and each of the constructions is represented. The size of the ‘boxes’ is indicative of the relative frequency of the correlation. Taking the first construction, [to my np], represented in the first column, we can see that it is associated with neutral degrees of both incongruity and performativity, as indicated by the blue cell. It is also disassociated from weak incongruity and strong performativity, as evidenced by the red cell at the bottom of this column. This conforms fully to what we have seen in the multiple correspondence analysis in Figure 2. The usage profile of this construction can thus be described as neutral with respect to mirativity – it is most commonly associated with contexts that constitute a middle conceptual field, being neither particularly performative nor predominantly descriptive. This neutral, rather than weak, character of the construction in both these regards may be attributed to its occasional employment in a relatively enactive manner, whereby the speaker stages the conceptualized scene in most vivid terms, as can be witnessed in (18) above or (25) below:

111

112 Karolina Krawczak and Dylan Glynn

Construction To my NP

W a NP

W the NP

STRONG

S

Pearson residuals: 8.25

WN S N W

WEAK

S

NEUT

W

Performativity

N

Incongruity

4

2

0

–2

–4 –4.94 p-value = 2.22e-16

Figure 3.  Loglinear analysis: Incongruity, performativity and construction

(25) Finally I got everyone out of my room, took a deep breath, then just sat there for a second looking at the mess… as soon as I felt calm enough to continue I felt something crawling up the inside of my pants leg and I just imagine this huge discussing [sic] black widow getting ready to stick its nasty little fangs into my leg, so I get up and try shaking it out but this ““bug”” just kept on climbing so I grabbed it from the out side of my pants and brought it down but to my surprise it wasn’t a massive ferocious spider but a sweet innocent little lady bug (I was mesmerized! lady bugs are my favorite bug they are my ““lucky bug”” so to speak …

In this utterance, the speaker chooses a highly performative and emotionally engaged way of reporting the situation, which serves to show how strongly incongruent the event was with his/her expectations. It is more of a performance than a mere description. So much so that the addressee can imagine the conceptualized event in quite vivid terms and experience it him- or herself in a vicarious manner, which is greatly facilitated by the re-enacting skills of the blogger. The spider is not just a spider, it is a huge discussing [sic] black widow or a massive ferocious spider, and the ladybug is not just a ladybug, but a sweet innocent little lady bug, and finally the speaker is not just surprised, s/he is mesmerized. Nevertheless, despite this rather highly enactive nature of this specific example, overall, the construction



Operationalizing mirativity 113

emerges as neutral in its performativity and incongruity alike. This is to say it remains somewhere in between highly performative and incongruent uses, on the one hand, and marginally performative and less incongruent ones, on the other. The construction might perhaps even be referred to as descriptive to the extent that, rather than enacting the situation in a dynamic manner, most of the time, it merely reports the event without much engagement on the part of the speaker. The second column shows the features linked to [what a np], which is repelled by both strong incongruity and performativity, as indicated by the red box at the top of the plot. This relates directly to what we have seen in the previous analysis presented in Figure 2. As we go to the bottom of the mosaic plot, we can see three cells of different shades of blue indicating that this construction is also linked to weak instances of incongruity in conjunction with all degrees of performativity, which is an interesting pattern that has not emerged to this extent in Figure 2. This regularity may be related to the fact that this construction tends to be used sarcastically relatively more often than the other encodings of surprise, where the level of incongruity is low, but that of performativity may vary, as illustrated in examples (26)–(29). In the first three instances (26)–(28), the level of performativity can be described as relatively high, given the use of such usage-features as capitalization, exclamation marks, multiplied question marks, interjections like wow and gee, the exceptionally emotional discourse manner adopted in (27) or the interactive-dialogic tone established in the first utterance below. The last example (29), on the other hand, is much weaker in enactive properties, its only distinctive feature being perhaps the smiley, which might be claimed to position it slightly higher on the performativity scale than would be the case, for example, with utterance (23) lacking any such characteristics. (26) Guess what I am doing????……WORKING AGAIN…..Wow what a surprise! Even though I only work 5 days a week, I feel like I am here all the time!! But anyways, I am going to my friend Dee’s, boyfriends’,mom’s house tonight for dinner. (that was a mouth full) (27) because half of your mother’s blood runs through your vains therefore you are inferior to them. your not all italian so you mean nothing until they need something then all of a sudden i’m alive gee what a shock. last time some relative needed a gift for some anniversary we had to go to the damn celebration. wahoo. dad and mom dissappeared on me and left me there with them. never in my life had i wanted to cry so hard so i’d go blind.then rip out my ear drums with a knife. all these relatives that couldn’t give so much as a rat’s anus about me asking me my age what school i’m in if i’m still in high school, keep in mind at the time i was a freshman. yeah i think i’m still in high school. some of them actually forgot who i was. they had to ask me who i was! one of my aunt’s had to remind

114 Karolina Krawczak and Dylan Glynn

them all that i’m “mike’s daughter”. that’s all i am mike’s fucking daughter, then the aunt that at least remembered that i’m “mikes daughter” had to ask what my name was!!! i felt so worthless. (28) Yeah thats right… today was JUST AS BAD AS YESTERDAY!!!! wow what a shock! ive been looking for someone alll day and can’t find them!!! ugh and i d k waht to do. I got grounded so i can’t go to the Senses fail concert. My quad broke down so i couldnt quad. I have horrible grades. (29) Chelsea won 2-0 over Middlesborough today! what a surprise, Chelsea wins everything! :)

Finally, the last construction [what the np] emerges here as being highly related to strong levels of incongruity and performativity, while, at the same time, being disassociated from weak incongruity in combination with strong performativity. This further supports our exploratory results registered in Figure 2 and demonstrates that Construction A can be regarded as the prototypical expression of mirativity in light of the definition proposed in Section one. Such a highly performative and incongruent profiling imposed by this construction can be seen in utterances (6), (15) or (21) above or (30) below.7 (30) omg i am so angry right now:@:@:@:@:@ i could literally strangle something. bloody hell. why is it that when people want something they act all nice when the rest of the time they act like total arseholes???? and then expect you to help them regardless of what theyve said before!!!!!!!!????????? why the fuck should i have to put up with that????? seriously WTF did i do to deserve that?????

Here, the epistemic state of incongruity, the state of complete shock at the behavior of other people is combined with intense anger, as expressed both verbally by the adjective angry and graphically by the multiplied use of the emoticon :@. It is further reinforced by the directly ensuing emphatic use of language that specifies, in a somewhat exaggerated manner, this emotional state of the speaker. The epistemic incongruity, in turn, is encoded farther in the example by the use of the expressions why the fuck, which has not been investigated in the study, and wtf. The expressive force of both these encodings of negative surprise is intensified by multiple question marks and exclamation marks as well as the emphatic adverb seriously. 7. Arguably, examples such as (18) or (30) might be considered to express not only the epistemic state of incongruity of the conceptualized situation and the speaker’s expectations (it is evident that what has happened is incongruent with what would be expected), but also his/her anger and annoyance at it. As mentioned in Note 4, this does not necessarily have any bearing upon the functional dimension of incongruity.

Operationalizing mirativity 115



Let us move now to the confirmatory modeling of the data. The results of the polytomous logistic regression analysis are presented in Table 2. Polytomous regression is a form of logistic regression that can be used when the dependent variable consists of more than two nominal outcomes. Arppe (2008) offers a detailed explanation of the method. In our case, since we have three constructions, binary logistic regression, typically used in the literature for constructional alternations, cannot be employed. The first column in Table 2 lists the two explanatory variables and their specific features with respect to which we expect to be able to account for the linguistic behavior of the three constructions. The next three columns provide the specific estimates (log-odds) for each construction. The estimates that are statistically significant are put in bold. The positive values are indicative of association, the negative of disassociation. That said, we can now consider the results. The first construction, [to my np] is strongly disassociated from weak incongruity, which corroborates what we have seen in Figure 3 or Figure 2, where this feature, is maximally distanced diagonally from the data-point representing this construction. No other significant predictors emerge here. The second construction, [what a np], has three significant predictors that are also sizable in terms of their effect/impact on the overall structuring of the data. The most important predictor here is the positive correlation with weak incongruity, followed by repulsion from strongly performative uses and strongly incongruent events. These correspondences converge fully with the Table 2.  Polytomous Logistic Regression: Cx ~ Degree of Performativity + Degree of Incongruity Predictors

to my np

what a np

what the np

(Intercept)

 (–0.3738)

(–0.3951)

 –1.306

Degree of Incongruity: strong

 (–0.3853)

 –1.035

   1.215

Degree of Incongruity: weak

 –3.62

   3.159

 –1.449

Degree of Performativity: strong

 (–0.08263)

 –1.167

   0.66

Degree of Performativity: weak

 (–0.02686)

  (0.6232)

(–0.5226)

Null deviance: 742.3 on 1047 degrees of freedom Residual (model) deviance: 576.2 on 1032 degrees of freedom McFadden R2: 0.22 Nagelkerke R2: 0.43 C statistic (reference level: to my np): 0.41 C statistic (reference level: what the np): 0.27 C statistic (reference level: what a np): 0.86

116 Karolina Krawczak and Dylan Glynn

emergent patterns revealed in Figure 2 and the correlations found in Figure 3. Finally, the last construction, [what the np], is significantly disassociated from weakly incongruent situations, while being clearly attracted to strong incongruity. It is also associated with strong performativity, but in a less pronounced way, given the smaller value of this estimate. Similarly to the two other constructions, in this case too, the respective patterns identified in the multiple correspondence analysis visualized in Figure 2 or the correlations observed in Figure 3 find their support here. The overall performance of the model can be determined on the basis of the pseudo R2 scores provided at the bottom of Table 2, with the values of 0.22 and 0.43 for McFadden R2 and Nagelkerke R2, respectively, where, as a rule of thumb, 0.2 and 0.3 are considered indicators of good model fit (Lattin, Carrol, & Green, 2003, p. 486; McFadden, 1978, p. 307). This, put more simply, means that the results can be treated as reliable. Another important indicator of the performance of the model is the C statistic, given at the bottom of Table 2. This value is informative with respect to the overall predictive power of the model. In other words, it tells us to what extent the generalizations emerging from the analysis can be taken to be indicative of the usage tendencies exhibited by the constructions. Three such scores are provided here, depending on the reference level, i.e., depending on which one of the three constructions is taken as the yardstick against which to compare the usage behavior of the remaining two constructions. As the C statistics demonstrate, the most predictive model is the one, where [what a np] serves as the reference level with an excellent value of 0.86. The C statistic obtained when the reference level is [to + my + np] is considerably lower, while that of [what + the + np] is still weaker in its predictive power. The immediate conclusion one can draw here is that the strongest model is yielded when the construction that has been identified as the least mirative in the correspondence analysis, i.e., [what a np], is taken as the reference level and the behavior of the two other more mirative constructions is juxtaposed with it. To recapitulate, both the exploratory and confirmatory results point toward [what the np] construction as the most prototypically mirative, which fully corroborates our hypothesis in this regard. This construction emerges in the study as commonly occurring in highly performative contexts, where the speaker, being frequently emotionally engaged, also expresses a high degree of incongruity between the conceptualized event and his/her overall system of knowledge, expectations and assumptions. The [to my np] constructional profiling of surprise also confirms our hypothesis that this construction will be weaker in its mirative value. However, contrary to our predictions, it does occur in situations where the speaker both re-enacts the conceptualized scene and communicates



Operationalizing mirativity 117

his or her great disbelief. This pattern was not expected. Finally, the [what a np] construal of epistemic incongruity emerges here as the weakest in its mirative value. It is most frequently used to encode weak incongruity and is significantly disassociated from strongly performative uses. This finding, which also diverges from our predictions, may be linked to the distinct tendency manifested by this construction to be used in ironic contexts. Naturally, no suggestion is made that these constructions are associated with a single usage and, like all formal structures in language, this usage is assumed to represent considerable semasiological variation. 5. Summary The primary goal of the present study was to ascertain the applicability of the proposed operationalization of the category of mirativity. Secondly, a descriptive case study examined three constructions encoding epistemic incongruity in English, i.e., [what the np], [what a np] and [to my np]. The results of this descriptive study were then interpreted employing the operationalization in order to determine if and to what degree these constructions can be regarded as mirative constructions. The operationalization of mirativity was based upon the interaction of two dimensions of usage: the conceptual degree of incongruity and the functional degree of performativity. Although there exists no independent means to determine the status or degree of mirativity in the constructions (making it impossible to actually test the accuracy of the operationalization), the operationalization proposed here proved straightforward to apply and appeared to produce coherent intuitively sound results. These results were, in turn, treated quantitatively. In this, we have demonstrated that such an inherently intersubjective phenomenon in language as mirativity, embedded in interaction and negotiated through context, can be operationalized and investigated with corpus-based quantitative methodology. The results of this descriptive case study reveal a reasonably clear picture of the encoding of mirativity in these constructions. The statistical modeling indicated that there emerges a cline of mirativity with respect to the two factors postulated as critical to its definition. The emergent continuum progresses from [what the np] through [to my np] to [what a np], with decreasing degrees of performativity and incongruity. This is a finding that corroborates our hypothesis concerning the first construction, which was expected to be the prototypical encoding of mirativity, but it also points to unexpected patterns of language use in the case of the two other constructional profilings of this epistemic state. These revealed

118 Karolina Krawczak and Dylan Glynn

tendencies, which are, at least to some extent, incongruent with our hypothesized paradigm, definitely merit further analysis. By way of conclusion, it should be stressed that the role of addressee-­ orientation and descriptivity, raised in Sections 3.1 and 3.3, are two lines of investigation that may impact upon mirativity but which were not included in this study. Further research is needed in order to determine whether these usage dimensions should be added to the operationalization. Finally, and perhaps most importantly, the proposed operationalization needs to be tested in other languages and text types, especially in true dialogic language and in languages where it is known that mirative constructions (morphemes or otherwise) exist. If this operationalization can adequately distinguish such forms from other non-mirative, yet epistemic incongruent forms, then it opens up the possibility for quantitative descriptive research. This, in turn, will allow the conceptual-functional category of mirativity to further enter the main stream of grammatical description.

References Akatsuka, N. (1985). Conditionals and the epistemic scale. Language, 61(3), 625–639. doi: 10.2307/414388 Arppe, A. (2008). Univariate, bivariate and multivariate methods in corpus-based lexicography; A study of synonymy. PhD dissertation. Helsinki: University of Helsinki. Bondarko, A. (1991). Functional grammar. A field approach (trans. I. Chulaki). Amsterdam: John Benjamins.  doi: 10.1075/llsee.35 Bresnan, J., Cueni, A., Nikitina, T., & Baayen, H. (2007). Predicting the dative. In G. Bouma, I. Krämer, & J. Zwarts (Eds.), Cognitive foundations of interpretation alternation (pp. 69– 94). Amsterdam: Royal Netherlands Academy of Arts and Sciences. DeLancey, S. (1997). Mirativity: The grammatical marking of unexpected information. Linguistic Typology, 1, 33–52.  doi: 10.1515/lity.1997.1.1.33 DeLancey, S. (2001). The mirative and evidentiality. Journal of Pragmatics, 33, 369–382. doi: 10.1016/S0378-2166(01)80001-1 DeLancey, S. (2012). Still mirative after all these years. Linguistic Typology, 16, 529–564. Dendale, P., & Tasmowski, L. (2001). Introduction: Evidentiality and related notions. Journal of Pragmatics, 33, 339–348.  doi: 10.1016/S0378-2166(00)00005-9 Dirven, R., Goossens, L., Putseys, Y., & Vorlat, E. (1982). The scene of linguistic action and its perspectivization by speak, talk, say, and tell. Amsterdam: John Benjamins. doi: 10.1075/pb.iii.6 Divjak, D. (2010). Structuring the lexicon: A clustered model for near-synonymy. Berlin: Mouton de Gruyter. Fontaine, J., Scherer, K., & Soriano, C. (Eds.). (2013). Components of emotional meaning: A sourcebook. Oxford: Oxford University Press.  doi: 10.1093/acprof:oso/9780199592746.001.0001 Geeraerts, D., Grondelaers, St., & Bakema, P. (1994). The structure of lexical variation: Meaning, naming, and context. Berlin: Mouton de Gruyter.  doi: 10.1515/9783110873061



Operationalizing mirativity 119

Glynn, D. (2008). Lexical fields, grammatical constructions and synonymy. A study in usage-­ based Cognitive Semantics. In H.-J. Schmid & S. Handl (Eds.), Cognitive foundations of linguistic usage-patterns: Empirical studies (pp. 89–118). Berlin: Mouton de Gruyter. Glynn, D. (2009). Polysemy, syntax, and variation. A usage-based method for Cognitive Semantics. In V. Evans & S. Pourcel (Eds.), New directions in Cognitive Linguistics (pp. 77– 106). Amsterdam: John Benjamins.  doi: 10.1075/hcp.24.08gly Glynn, D. (2010). Testing the hypothesis: Objectivity and verification in usage-based Cognitive Semantics. In D. Glynn & K. Fischer (Eds.), Quantitative methods in Cognitive Semantics: Corpus-driven approaches (pp. 239–269). Berlin: Mouton de Gruyter. doi: 10.1515/9783110226423 Glynn, D. (2014). Correspondence analysis: Exploring data and identifying patterns. In D. Glynn & J. Robinson (Eds.), Corpus methods for semantics. Quantitative studies in polysemy and synonymy (pp. 443–486). Amsterdam: John Benjamins.  doi: 10.1075/hcp.43.17gly Glynn, D., & Fischer, D. (Eds.). (2010). Quantitative Cognitive Semantics: Corpus-driven approaches. Berlin: Mouton de Gruyter.  doi: 10.1515/9783110226423 Glynn, D., & Robinson, J. (Eds.). (2014). Corpus methods for semantics: Quantitative studies in polysemy and synonymy. Amsterdam: John Benjamins.  doi: 10.1075/hcp.43 Goldberg, A. (1995). Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Gries, S. T. (1999). Particle movement: A cognitive and functional approach. Cognitive Linguistics, 10, 105–145.  doi: 10.1515/cogl.1999.005 Gries, S. T. (2003). Multifactorial analysis in corpus linguistics: A study of particle placement. London: Continuum. Grondelaers, S., & Geeraerts, D. (2003). Towards a pragmatic model of cognitive onomasiology. In H. Cuyckens, R. Dirven, & J. Taylor (Eds.), Cognitive approaches to lexical semantics (pp. 67–92). Berlin: Mouton de Gruyter.  doi: 10.1515/9783110219074.67 Halliday, M. A. K. (1985). An introduction to functional grammar. London: Edward Arnold. Heylen, K. (2005). A quantitative corpus study of German word order variation. In S. Kepser & M. Reis (Eds.), Linguistic evidence: Empirical, theoretical and computational perspectives (pp. 241–264). Berlin: Mouton de Gruyter.  doi: 10.1515/9783110197549.241 Krawczak, K. (2014a). Shame and its near-synonyms in English: A multivariate corpus-driven approach to social emotions. In I. Novakova, P. Blumenthal, & D. Siepmann (Eds.), Emotions in discourse (pp. 84–94). Frankfurt/Main: Peter Lang. Krawczak, K. (2014b). Epistemic stance predicates in English: A quantitative corpus-driven study of subjectivity. In D. Glynn & M. Sjölin (Eds.), Subjectivity and epistemicity: Corpus, discourse, and literary approaches to stance (pp. 355–386). Lund: Lund University Press. Krawczak, K. (2014c). Corpus evidence for the cross-cultural structure of social emotions: Shame, embarrassment, and guilt in English and Polish. Poznań Studies in Contemporary Linguistics, 54, 441–475. Krawczak, K., & Glynn, D. (in press). Operationalising construal. Of / about prepositional profiling for cognition and communication predicates. In C. M. Bretones Callejas & C. Sinha (Eds.), Construals in language and thought: What shapes what? Amsterdam: John Benjamins. Langacker, R. (1987). Foundations of Cognitive Grammar. Stanford: Stanford University Press. Lattin, J. M., Carrol, J. D., & Green, P. E. (2003). Analyzing multivariate data. Pacific Grove: Thomson Brooks.

120 Karolina Krawczak and Dylan Glynn

McFadden, D. (1978). Quantitative methods for analyzing travel behavior of individuals: Some recent developments. In D. A. Hensher & P. R. Stopher (Eds.), Behavioral travel modeling (pp. 279–318). London: Croom Helm. Nordmark, H., & Glynn, D. (2013). Anxiety between mind and society: A corpus-driven cross-cultural study of conceptual metaphors. Explorations in English Language and Linguistics, 1, 107–130. Nuyts, J. (2001). Subjectivity as an evidential dimension in epistemic modal expressions. Journal of Pragmatics, 33, 383–400.  doi: 10.1016/S0378-2166(00)00009-6 Scherer, K. (2005). What are emotions? And how can they be measured? Social Science Information, 44, 693–727.  doi: 10.1177/0539018405058216 Soriano, C., Fontaine, J. R. J., & Scherer, K. R. (this volume). Surprise in the GRID. Speelman, D. (2005). LiveJournal corpus of British and American English. Leuven University. Stefanowitsch, A. (2010). Empirical Cognitive Semantics: Some thoughts. In D. Glynn & K. Fischer (Eds.), Quantitative Cognitive Semantics: Corpus-driven approaches (pp. 355– 380). Berlin: Mouton de Gruyter. Talmy, L. (2000). Towards a Cognitive Semantics. Cambridge: MIT Press. Winner, E., Windmueller, G., Rosenblatt, E., Bosco, L., Best, E., & Gardner, H. (1987). Making sense of literal and nonliteral falsehood. Metaphor and Symbolic Activity, 2, 13–32. doi: 10.1207/s15327868ms0201_2

The computer-mediated expression of surprise A corpus analysis of chats by English and Italian native speakers and Italian learners of English Laura Ascone

Université Paris Diderot

This paper investigates how Italian native speakers express surprise in English as their second language on Facebook. A qualitative study was conducted on a corpus of forty English utterances by Italian native speakers conveying surprise and two control corpora composed of forty Italian and forty English native speakers’ expressions. First, a systemic approach will be adopted: by analysing the order in which the speaker reacts to, comments on, and wonders about new information, the objective is to determine a pattern peculiar to the verbal expression of surprise, and to ascertain how the mother tongue and the language-learning background are influential when expressing an instinctive reaction such as surprise in a foreign language. Attention will then be paid to the lexical expression of surprise. In particular, the analysis will focus on the features specific to non-native speakers (i.e. use of verbs and code-switching), on the codes peculiar to CMC (i.e. smileys and punctuation), and on how these codes are employed to convey surprise disruption, valence and intensity. By examining all these aspects, this research examines how English non-native speakers express surprise in chats. Keywords: surprise, CMC, Facebook, code-switching, intensity

1. Introduction This research paper analyses how Italian native speakers express surprise in their mother tongue and in English as their second language when communicating in chats. The objective of this research is to see how this inter-language positioning influences the expression of such an instinctive reaction. A qualitative study was conducted on forty English non-native speakers’ utterances and on two control doi 10.1075/bct.92.05asc © 2017 John Benjamins Publishing Company

122 Laura Ascone

corpora composed of forty English and forty Italian natives’ expressions. Attention was focused on conversations where surprise was aroused. Since surprise is so short in time (Grandjean & Scherer, 2008), people are not given the time to think about how to express it. In a second language, this lack of time may lead the non-native speaker to use short and fixed expressions or to literally translate a first-language expression into the second language. Moreover, since surprise is a reaction that makes conversation participants lose control over the situation for only some milliseconds, it is difficult, if not impossible, to define and study what is said at the very moment when the speaker is surprised. These particular characteristics make the expression of surprise in a foreign language even more complex. Because of the high variability of language-learning backgrounds, and because of the subjective dimension of emotions – different persons may not have the same emotional reaction to the same stimulus (Scherer, 1984) –, research to date has tended to study the expression of surprise in monolingual and bilingual speakers,1 based on data collected in instructed environments such as laboratories or texts (Krashen, 1981). This paper presents a novel investigation of the expression of surprise in late-bilingual speakers in a more natural context. The objective of this study is to examine how influential a mother tongue and the language-learning background – i.e. natural versus instructed environment – are, when expressing an instinctive reaction like surprise in a foreign language. In order to obtain the most spontaneous and natural expressions possible, the data was randomly collected from original Facebook conversations in which speakers (aged between 20 and 30) took part between 2010 and March 2014. This way, it is possible to investigate the language specific to chats and social networks – in opposition to spoken and written language – which has increasingly become an important means of expressing emotions, despite being neglected in most of the prior studies on emotions. The entire corpus (120 utterances) was analysed by taking into account the speakers’ level of intimacy – i.e. how long they had known each other2 –, the subject of the conversation and the order in which the speaker reacts to (reaction), comments on (comment), and wonders about (question) unexpected information. For non-native speakers, both the level of English and the language-learning 1. Contrary to bilinguals, who have two mother tongues, late-bilingual speakers have become bilinguals later than childhood. 2. Since the corpus relies on personal conversation, it was possible to define the level of intimacy of the two interlocutors. Aware of the fact that this parameter cannot be used for larger corpora, I took it into account with the objective of investigating the expression of surprise in depth, which might prompt further research in this field.



The computer-mediated expression of surprise 123

background were also taken into account. As far as the English native speakers are concerned, they are either American or British, except for one speaker who comes from South Africa. Despite the limits imposed by the choice of the parameters, namely the level of intimacy, this study allowed us to conduct a deeper analysis of the computer-mediated expression of surprise. When surprised, people’s responses occur in a split second, which means that they do not have the time to think about how to convey a message (Zajonc, 1980). When speaking in a second language, a speaker may use fixed expressions or literally translate a first-language utterance into his or her second language due to the lack of time. Even though some translation equivalents exist, it is not always possible to use them because of the cultural context (Pavlenko, 2008). By taking into account all these ‘obstacles’, this research aims at providing an analysis of how English non-native speakers convey surprise when communicating in chats. Firstly, a systemic approach will be adopted: by examining the combination of the reaction-, comment- and question-segments, the aim is to determine whether the expression of surprise varies according to the language used, and whether the verbal expression of surprise follows the cognitive experience of this reaction. Attention will therefore be focused on the lexicon used to express surprise. Compared to English native speakers, non-natives appear to mainly use verbs and code-switching to convey surprise. Smileys and punctuation will be argued to play a specific role in the computer-mediated expression of surprise and, in particular, in the expression of surprise valence and intensity. Our focus will be on how they are used to convey surprise disruption as a result of a discrepancy between the speaker’s expectations and what s/he is told, and how they may contribute to valence and intensity. After having discussed the theoretical issues and the link between surprise, cognition and emotions, attention is paid to the specificities of the corpus collected, and to the way in which the chat-codes are employed to express surprise. 2. Theoretical issues: Surprise, cognition, and emotion Everyone knows what an emotion is, until they are asked to give a definition of it (Fehr & Russell, 1984). As both our bodies and minds react to emotional stimuli, emotions have been widely examined from a physiological as well as a psychological point of view. More recently, linguists have started analysing how emotions are expressed verbally.

124 Laura Ascone

2.1

Temporal dimension of surprise

Ekman (1992) suggests another feature that differentiates surprise from other emotions. According to him, standard emotions such as joy or anger last only some seconds. On the contrary, surprise lasts no more than some milliseconds. Macedo, Reisenzein, and Cardoso (2004, p. 1) describe the mental processes that occur when surprise is elicited as follows: Surprising-eliciting events initiate a series of mental processes that (a) begin with the appraisal of a cognised event as exceeding some threshold value of unexpectedness or schema discrepancy, (b) continue with the interruption of ongoing information processing and the reallocation of processing resources to the surprise-eliciting event, and (c) culminate in the analysis and evaluation of that event plus immediate reactions to it and/or schema (belief) updating/revision.

The person realises that something is incongruent with his or her expectations and, because of this unbalance, communication cannot continue. The subject tries to interpret unexpected information in order to establish the pre-existing balance again. A surprise reaction may thus be said to be followed by the agent’s interpretation of the emotional stimulus. This very particular mental process suggests that surprise does not follow the stimulus–interpretation–reaction pattern as any other emotion does (see Gross & Thompson, 2007). When someone tells us something unexpected, we are surprised at the very moment when the new piece of information is announced; when we react to this unexpected event, however, surprise has already vanished (see Scherer, 2001). When we ask for more information, our questions vary according to the emotion triggered by the surprising news. Because of its short duration (see Ekman, 1992), surprise starts with the reception of the stimulus, which is followed by the arousal of the emotion. When, for instance, two people are talking, a sort of balance is established; the two persons speak in turn, carrying on the conversation. When one of them, Speaker 1, says something unexpected to his interlocutor, Speaker 2, the balance is broken. The regular interaction ‘movement’ is interrupted by the latter who does not accept what Speaker 1 has just told him/her. Then, by asking questions, Speaker 2 may try to re-establish the pre-existing balance. This exchange of questions and explanations may be seen as the process of interpretation that, in emotions like fear and anger, takes place between the reception of the stimulus and the emotion itself. Then, analysing the new piece of information in relation to his/her expectations, the surprised speaker tries to understand why s/he is surprised. To sum up, while emotions like fear and anger can be represented in stimulus–interpretation–response sequences, the scenario that applies to surprise is stimulus–­

The computer-mediated expression of surprise 125



response–­interpretation. Since interpretation comes only after the speaker’s reaction, surprise can be defined as the instinctive response to an unexpected event. Furthermore, contrary to any other emotion, the reaction of surprise lasts only some milliseconds. All these differences between surprise and other emotions challenge the assumption that surprise is an emotion; therefore, in this paper, surprise is considered as an instinctive reaction emotions are generated from. 2.2

Valence

According to Rusinek (2004), the origin of emotions is found in the breaking of one’s psycho-physiological balance. Since surprise breaks the balance between someone’s expectations and the real situation, it may be seen as a source of emotions, rather than an emotion by itself. Moreover, contrary to all emotions, which can be characterised by a positive or negative valence, surprise is the only reaction that can be either positive or negative (see Zammuner, 1998). The experience of anger, for instance, comes from the negative perception of an event. On the contrary, a surprising event may be judged positive or negative, according to how one feels about the event eliciting surprise. Furthermore, in some situations, surprise has a neutral valence: an unexpected event may be judged neither positive nor negative. To sum up, a series of factors – namely, the differences in the duration and in the cognitive patterns followed by surprise and other emotions, and the fact that, contrary to any emotion, surprise may be judged as neither positive nor negative –, give way to the hypothesis that surprise is not an emotion by itself, but rather a source of emotions. This view of surprise as the unbalance that generates psycho-­physiological reactions is at the core of this study, which investigates how non-­native speakers involved in computer-mediated conversations convey this disruption. As stated before, not all psycho-physiological responses can be classified as positive or negative; words like, for instance, surprise or amazement “do not imply any evaluation” (Wierzbicka, 1999). In other words, the lexicon of surprise is deprived of valence, in contrast with the valence attached to the evaluation of the stimulus. Surprising events and news may be judged as either positive or negative by the agent. However there are some cases where it is difficult, if not impossible, to determine the valence of the reaction triggered by unexpected stimuli. When a person who is experiencing a given emotion is seen or heard, the valence of that emotion is usually identified through his/her facial expressions or the pitch of his/ her voice. We usually do not only pay attention to words in order to understand whether a stimulus has been judged as being positive or negative. On the contrary,

126 Laura Ascone

when we text or e-mail someone, we realise that words are actually a fundamental tool that allows us to understand whether the other person is reacting positively or negatively. 2.3

Description vs. expression

All those characteristics that distinguish surprise from emotions like fear and anger are reflected in the way surprise is expressed verbally. The main reason why much attention has focused on the verbal expression of emotions is that language is the only real way to access them. Ortony, Clore, and Foss (1987) describe language as “the most convenient nonphenomenological access to emotions” (p. 342). This idea of a strong relationship between language and emotion was supported by Shaver, Schwartz, Kirson, and O’Connor (1987, in Reeve, 2012), who claimed that emotional experiences are part of language. Even when we do not explicitly express what we feel, what we say is permeated with our impressions and considerations about what surrounds us. Caffi and Janney (1994) also argued that language is strongly linked to feelings. This link between language and emotions has led many researchers to highlight the importance of cognition in the experience of emotions. However, the obvious difficulty to express what we feel has opened the way to a completely opposite position, according to which there is no relationship between emotions and cognition (Marsella & Gratch, 2009), and the obstacles encountered in the verbal expression of emotions are supposedly evidence of this separation. In the literature, a crucial distinction is made between the description of emotions and the expression of emotions. Descriptive language is used by the person who experiences emotions to comment on what and/or how they felt. It involves emotion lexemes, which can be classified as basic words denoting emotions (fear, anger, …), and non-basic words describing how the person who experiences the emotion feels (scared, astonished, …) (LeDoux, in Kövecses, 2003). Within descriptive language, another distinction is generally made between literal and figurative language. Figurative language involves metaphors and metonymies (e.g. I was overwhelmed by surprise) (Kendrick-Murdock, in Kövecses, 2003). As for the expression of emotions, it is linked to disfluency markers such as exclamatives, interjections, etc. – markers that occur when the emotion is being experienced as we shall see in our corpus analysis. This shows that analyzing how people convey surprise cannot be reduced to a study of the emotion lexemes used but has to take into account a whole variety of linguistic markers. As for the relationship between language and emotions, it is generally argued that there are two different ways of expressing our emotions and feelings.



The computer-mediated expression of surprise 127

Plantin (2011) distinguishes between emotional discourse, through which we express what we feel at the very moment when emotions are experienced, and emotive discourse, which is defined as referring to the situations in which emotions are described once they have already faded away. We could then make the distinction between expressing emotions – i.e. emotional discourse – and describing and commenting on them – i.e. emotive discourse. Arndt and Janney (1991, in Caffi & Janney, 1994) examine this distinction in a more detailed way. While emotional discourse is seen as “a type of spontaneous, unintentional leakage or bursting out of emotion in speech” (Caffi & Janney, p. 325), emotive discourse is regarded as being “inherently strategic, persuasive, interactional, and other-directed by its very nature” (Caffi & Janney, p. 329). This definition of emotive discourse shows that interaction is a relevant component of emotion. The hypothesis that emotional and emotive discourse are only used by speakers in order to explain what generated their emotions, is challenged by Plantin (2011, p. 189), who claims that emotion is not an expressive strategy, but rather an interactive one. Descriptive language – as opposed to expressive language – gives us information about the valence, intensity and cause of emotions. Since surprise lasts no more than some milliseconds, when we use descriptive language, surprise has already faded away, which means that utterances like “I am surprised” do not express surprise; rather, they inform our interlocutor that what we experienced was surprise. In contrast, expressive discourse conveys this instinctive response caused by a disruption between the speaker’s expectations and an event. Some parts of speech and expressions convey the speaker’s emotions directly. “Certain parts of speech – evaluative adjectives, interjections and expletives as well as other intensifying expressions (e.g., adverb) – have become specialised for affective use” (Hakulinen & Sorjonen, 2012, p. 147). So, interjections and exclamatives have become specific to the verbal expression of surprise. More precisely, interjections reflect the idea of disruption which, as already shown, is the main feature of this instinctive reaction. In fact, the speaker usually utters interjections involuntarily. They are verbal evidence of the experiencer’s loss of control over the situation caused by the introduction of new information or events. Let us now focus on a reaction taken from my data (example (1)). (1) “Oh that’s sooo cool I wish I could do that” (Intimacy: 1; Topic: video of a musician; Sequence: reaction-comment)

This English native’s utterance shows how computer-mediated communication includes characteristics of both oral and written language. For instance, the speaker uses the orthographic form ‘that’s’, which is more common in oral productions than in written ones. Furthermore, the additional vowels in the intensifier ‘sooo’ reproduce the speaker’s intonation (Ko, 1996), which can be perceived in spoken

128 Laura Ascone

language only. The fact that the person emphasises the adverb so is evidence of the role played by the reaction’s intensity in the experience and expression of surprise (Zammuner, 1998). The absence of punctuation makes the utterance appear as a flow of sounds as in spoken productions. As far as the surprise reaction is concerned, this example reveals the three most crucial stages of the surprise process. First, the speaker reacts by uttering the interjection ‘oh’, which reproduces the disruption caused by the video. Aijmer (2002) and Grinshpun (2008) focus on the different affective meanings of oh. This interjection is usually employed to convey agreement, acceptance and endorsement. It establishes a link between the new information and the receiver’s adjustment to the ‘new reality’. “More frequently oh and ah occur in elicitational contexts with the function to keep the conversation going” (Aijmer, 2002, p. 123). Yet, even though interjections like oh are usually uttered instinctively, they are likely to occur in conversations between speakers who do not know each other – i.e. when the level of intimacy is low – (Jucker & Smith, 1998, p. 178). Second, the speaker comments on the video. Despite the fact that surprise has already faded away, the reaction’s intensity – which is the main dimension linked to surprise prototypicality (Zammuner, 1998) – may still be perceived through the adverb ‘sooo’. By contrast, the final statement ‘I wish I could do that’ and its complex grammatical structure show that the speaker has taken the control of the situation again. This example confirms the tendency of not responding to unexpected information with interjections only (Hakulinen & Sorjonen, 2012). Rather, interjections help speakers carry the conversation on. Yet, it does not mean that they are uniquely used as interactive strategies (Wilkinson & Kitzinger, 2006). They are used as a transition between surprise disruption and stimulus interpretation, which gives weight to our hypothesis, according to which surprise is the source of emotions rather than an emotion by itself. 3. Presentation of the corpus 3.1

Characteristics of CMC

Even though CMC differs from both written and spoken communication, “users experience CMC in fundamentally similar ways to spoken conversation, despite CMC being produced and received by written means” (Herring, 2010). It is not coincidental that there are codes peculiar to chats whose objective is to imitate oral conversations (Ko, 1996). For instance, smileys are used to reproduce facial



The computer-mediated expression of surprise 129

expressions. This way, interlocutors may try to reduce the physical distance between each other; they may try to create and place themselves in a virtual situation of utterance common to both of them. Since CMC is, at the same time, similar to and different from both spoken and written language, it has been defined as “written speech” (Maynor, 1994) and “visible conversation” (Colomb & Simutis, 1996). Murray (1985) identified elements in CMC that are typical of oral conversations such as “active voice and personal pronouns; emotive and informal diction” (Murray, 1985, p. 217), contractions and ellipsis. At the same time, CMC presents features that characterise written productions like “attributive adjectives, participles and complement and relative clauses” (Murray, 1985, p. 220). However, the language used in computer-mediated conversations is likely to be considered “less correct, complex and coherent than standard written language” (Herring, 2001, p. 616). One of the features that makes CMC appear unconventional is the way people reproduce “auditory information such as prosody, laughter and other non-language sounds” (Herring, 2001, p. 616). The exclamative prosody, for instance, is reproduced using ‘visual elements’ – e.g. exclamation marks, additional and/or capital letters, etc. In CMC the use of exclamatives, which is typical of spoken language, is evidence of the proximity of spoken and computer-­mediated language. There are some expressions that have an intrinsic exclamative value like wow! and hurrah!. According to Zanuttini and Portner (2003), exclamatives can be analysed only if their syntactic and semantic components are taken into account. As far as the semantic aspect is concerned, exclamatives convey the speaker’s evaluation of an event. More specifically, the event that produces the exclamative reaction is usually evaluated as more or less ‘non canonical’ (Sorianello, 2012, p. 296). The ‘more or less’ can be established by the use of specific words (mainly adjectives and adverbs). In fact, intensity is sometimes conveyed with words like so (example (1)) and amazing (e.g. ‘That’s amazing!’). Sorianello adds that intonation is the only means receivers have to determine the right interpretation of the exclamative utterance. As stated before, in CMC, where exclamatives and proposition exclamations (Rett, 2008) are deprived of their prosodic component, users employ punctuation and smileys to reproduce intonation. Therefore, the objective of this study is to examine how all these elements characterising surprise are conveyed in computer-mediated conversations by English and Italian native speakers, and by Italian learners of English. More precisely, by analysing a corpus of forty non-natives’ utterances, the goal is to see how the language spoken (i.e. first vs. second language) and the context in which the conversation takes place influence the expression of this instinctive reaction.

130 Laura Ascone

4. Corpus study 4.1

Methodology

Valence and intensity are the most relevant factors characterising emotions (Zammuner, 1998). Yet, some researchers like Ortony et al. (1988) consider surprise as a reaction deprived of valence; rather, surprise refers to something unexpected and, once the person has evaluated the surprising event, it becomes a positive or negative surprise (Ortony et al., 1988, p. 32). This analysis investigates the codes specific to chats that are chosen by speakers to convey valence and intensity when experiencing surprise. Firstly, a systemic approach will be adopted to analyse the order in which speakers react to, comment on, and wonder about unexpected information. Every example is accompanied by a description of the level of intimacy of the two speakers (Intimacy: from 1 to 4, where 1 stands for two persons who have just met, and 4 for two persons who have known each other for a long time), the subject of the conversation, and the order in which the speaker reacts to (reaction), comments on (comment), and wonders about (question) the surprising information. For non-native speakers, the level of English (English: from 1 to 5; 1: beginner; 2: pre-intermediate, people who learnt English at school only and who do not study it anymore; 3: intermediate, people who continued studying English but not in an academic context; 4: upper-intermediate, people who study foreign languages at university, or who have spent at least one year in an English speaking country; 5: bilinguals, people who have one parent who is an English-native speaker, or who have been attending immersion classes since an early age) and their language-learning background (instructed versus natural environment, and the number of years spent in an English speaking country) are taken into account as well. Then, attention is paid to some features specific to native and non-natives’ utterances (e.g. the use of verbs and code-switching), and the distinguishing features of chat codes, such as smileys and punctuation. Lastly, the chat lexicon is examined in order to see how valence and intensity are expressed. 4.2

Sequences

Analysing the different examples in a detailed way, it has been noticed that every utterance is composed of ‘segments’. In the same expression, the speaker may react to the stimulus, comment on it, and ask for more information about the unexpected event. Consider example (2). The whole expression can be divided into three parts: ‘veramente??’ can be considered an instinctive reaction; ‘da quando??’



The computer-mediated expression of surprise 131

can be seen as a question asked to have more details; and ‘sono contentissimo! Che bella notizia!” can be considered the comment made by the speaker who analysed the unexpected information in relation to his expectations. We then have a reaction-­question-comment sequence. (2) “Veramente?? Da quando??Sono contentissimo! Che bella notizia!” (“Really?? Since when?? I’m so happy! What good news!”) (Intimacy: 4; Topic: talking about a friend’s boyfriend; Sequence: reaction-­question-comment) (3) “Aw wow!! I love Paris, never been to Rome though. Have you ever been to Scotland?” (I: 1; T: his interlocutor lives in Paris; S: reaction-comment-question)

In these two examples, it is possible to see the difference between the reaction-segment and the comment- and question-ones. The first part of the two utterances reproduces the speaker’s instinctive reaction, which, as I will show further on, is usually expressed through interjections (‘Aw wow!!’ in the native’s expression in example (3)) or short utterances such as ‘Veramente??’ (example (2)) – almost the totality of the utterances containing a reaction-segment follow this pattern. On the other hand, when commenting on new information or asking for more details about it, speakers tend to use longer and complete sentences (‘I love Paris, never been to Rome though’ in example (3)) – this was true of 72% of the expressions with comment- and/or question-segments. Compared to the beginning of the utterance (‘Aw wow!!’), the second part (‘I love Paris, never been to Rome though’) is more grammatically structured, which is probably due to the fact that when the speaker comments on unexpected information, surprise has already faded away (François & Legallois, 2012). The speaker is not reacting instinctively anymore, as in the reaction-segment; rather, he is consciously analysing the news, which would explain the use of more complex and grammatically structured sentences. This can be summarised by Guillaume’s (1973) equation “Expression + Expressivity = 1”. The more intense the emotion is, the less grammatically structured the verbal expression needs to be. François and Legallois (2012, p. 214) consider ‘expression’ as making it understand, and ‘expressivity’ as making it evident. The more evident (i.e. intense) the emotion is, the less understandable (grammatically speaking) it needs to be. The entire corpus was analysed according to the order of the different segments within a surprise sequence. This way, it was possible to determine whether the expression of surprise follows a specific pattern in terms of reaction, comment and question, and if this pattern varies according to whether the language employed is the speaker’s first or second language.

132 Laura Ascone

Table 1.  Sequences in Italian natives’ utterances Sequence

Frequency

Reaction Reaction-comment Reaction-question Reaction-question-comment Reaction-comment-reaction Reaction-comment-question Question-reaction Comment

10 12  8  3  3  2  1  1

4.2.1 Italian native speakers’ reaction of surprise Table 1 shows how many times a specific sequence occurs in the Italian corpus. The results obtained confirm the hypothesis that surprise does not work like ordinary emotions. As already explained, in emotions like anger the reaction comes only after the interpretation of and the comments on the emotional stimulus (Gross & Thompson, 2007). Without interpreting the stimulus, it would be impossible to anyone to determine whether they should be angry or not. In contrast, when experiencing surprise, the stimulus interpretation comes after the psycho-physiological response. Almost in every Italian utterance – 38 out of 40 –, the reaction-segment occurs at the very beginning of the entire sentence (example (4)). (4) “Cosa?!?! :O E come hai fatto??” (“What?!?! :O And how did you do??”) (I: 4; T: talking with someone who broke his harm; S: reaction-question)

This specific order shows that new information is unexpected, which leads the speaker to comment on what he has been told in order to reach the pre-existing balance. Yet, this does not imply restoring balance at this point of the sequence; the speaker might need more time and information before accepting the unexpected event. The fact that these sequences are far more frequent than the reaction-­question-comment or reaction-comment-question sequences could be seen as evidence of the fact that surprise is a short instinctive reaction. As stated before, the more grammatically structured the sentence is, the more distant – in terms of time and feeling – it is from the few milliseconds during which the person was surprised (François & Legallois, 2012). Therefore, it is also hypothesised that this specific feature of surprise is reflected in the level of grammaticality of the reaction-segment; short verb-less sentences are expected to be used to convey real surprise (i.e. in reaction-segments).

The computer-mediated expression of surprise 133



Table 2.  Sequences in English natives’ utterances Sequence

Frequency

Reaction Reaction-comment Reaction-question Reaction-comment-question Reaction-comment-reaction Comment Comment-question

18  8  2  2  1  8  1

4.2.2 English native speakers’ expression of surprise A second control corpus of forty English natives’ utterances was analysed according to the order of the different segments in each sequence. Table 2 shows that the segments’ order of the English natives’ expressions follow a pattern similar to the Italian natives’ one. Yet, while the occurrences of reaction and reaction-comment remain the most frequent ones, the utterances composed of a comment-segment only occur only once in the Italian expressions and 8 times in the English ones (example (5)). (5) “Totally amazing!!” (I: 1; T: picture of a doctor and a newborn; S: comment)

Here, the speaker is commenting on a picture she has just seen. The adjective amazing and the two exclamation marks, which reproduce the typical surprise exclamative intonation, makes us understand that the person is surprised. Like in the Italian examples, the reaction-segment is at the beginning of the entire sentence in almost every utterance – 31 out of 40. The speaker may react with sounds (i.e. interjections), utterances (usually short verb-less sentences), or pauses to an unexpected event. In CMC, the speaker may use codes peculiar to chats – e.g. smileys – to reproduce the instinctive reaction and/or the pauses caused by the unexpected information. Comparing the Italian natives’ corpus to the English one, the former presents almost as many comment-segments as the latter – 21 in the 40 Italian utterances and 20 in the English ones –; on the other hand, the question-segment occurs 14 times in the Italian examples and only 5 in the English ones. Moreover, in the Italian utterances, the question-segment occurs only once at the beginning of the sentence. As far as the comment-segment is concerned, it occurs once at the beginning of the Italian utterances and 9 times at the beginning of the English natives’ expressions. The reduced use of the speakers’ comments or questions at the beginning of the sequence may be due to the fact that, when experiencing surprise, the

134 Laura Ascone

interpretation of the new information comes only after the instinctive reaction. To sum up, when expressing surprise verbally, the speaker tends to follow the cognitive pattern; s/he first expresses his/her instinctive reaction and, only then, s/he comments on and evaluates (Ekman & Friesen, 2003; Valetopoulos, 2013) what s/he has been told, and/or ask for more details about it. Despite the limited size of the corpus, these similarities led us to wonder whether the structures of the English expressions by the Italian natives followed the same pattern as the native speakers’, and whether they were closer to the Italian or to the English structures. We want to investigate whether, when speaking in their second language, Italian speakers tend to follow the structures of their mother tongue or those of the foreign language in which they are communicating. 4.2.3 English non-native speakers’ expression of surprise As we have seen, Italian and English native speakers follow quite a similar pattern; their utterances almost always present a reaction-segment at their beginning (38 out of 40 in the Italian expressions, and 31 out of 40 in the English ones), followed by a comment- and/or question-segment. In comparison to the native speakers’ utterances, the expressions used by the non-natives were far more disorganised. Eleven different sequences were identified, some of which appeared neither in the English nor in the Italian natives’ expressions. Table 3 shows the number of occurrences of each sequence. Yet, quite surprisingly, the most common sequences in the non-native speakers are frequently used by the English natives, while they never occur in the Italian utterances – i.e. (1) reaction-comment, (2) reaction, and (3) comment (example (6) and (7)). Table 3.  Sequences in English non-natives’ utterances Sequence

Frequency

Reaction Reaction-comment Reaction-question Comment Comment-reaction Comment-question Question Question-reaction Reaction-comment-question Reaction-comment-reaction Question-comment-reaction

 6 11  5  9  2  1  1  2  1  1  1



The computer-mediated expression of surprise 135

(6) “Wow!!! I wouldn’t be brave enough to move to the US” (I: 3; T: the interlocutor wants to move to the US; S: reaction-comment; E: 4-instructed environment) (7) “That’s unbelievable” (I: 3; T: video of a singer; S: comment; E: 4-bilingual)

In example (7), the speaker comments on the video he has just watched. Contrary to example (6), where the reaction is conveyed with the interjection and exclamation marks, in example (7) it is suggested by the adjective unbelievable, which underlines the gap between his knowledge and what he has watched. The tendency to use the sequence reaction-question and reaction-comment has not changed. However, in English non-natives’ utterances, the hypothesis that the reaction-segment occurs at the beginning of the sequence in almost every case was disproved. While only 9 English and 2 Italian expressions do not present the reaction-segment at their beginnings, out of the forty English utterances by non-native speakers, only 24 begin with a reaction. The verbal expression of surprise turned out to be far less natural and spontaneous when the Italian-native speakers chatted in English. This is due to the fact that when speaking in a foreign language, we need to focus our attention on both the content of the message and its form. When an emotion or an instinctive reaction is triggered, we may feel the necessity to express it to those around us immediately (Dewaele, 2010). In a bilingual context, we may realise that we are not proficient enough to formulate a sentence in our second language in a very short period. In a similar case, we may feel the necessity to express an emotion in the language we are more comfortable with. Speakers may switch to their mother tongue because they may perceive their first language as more emotional (Dewaele & Pavlenko, 2001– 2003; Bloom & Beckwith, 1989, in Harris, Gleason, & Aycicegi, 2006, p. 271). Yet, Harris et al. (2006) argue that a second language may become as emotional as the mother tongue even if it is learnt after puberty (LaBar & Phelps, 1998). According to them, in fact, what allows a foreign language to become emotional enough is not the age when it is learnt, but the context in which the speaker learnt it. It is for this reason that each English non-native utterance is provided with the description of the speaker’s language-learning background – i.e. instructed/natural environment. Even though the non-native speakers of English tend to follow the English pattern rather than the Italian one, evidence of the influence of the mother tongue could be identified. By favouring the comment-segment (example (8)) over the question-segment at the beginning of a sentence, the non-native speakers of English tend to follow the English pattern. However, the fact that the non-native speakers of English start their sentences with question-segments as well – which

136 Laura Ascone

is not the case for the English natives – suggests that the mother tongue still plays a part when communicating in a foreign language. It may be argued that only when speakers have to express an instinctive reaction in a foreign language – i.e. when they do not have the time to think about the form of the message – is it possible to see whether the influence of the speaker’s mother tongue is stronger than his/her attempt to follow the foreign-language pattern. However, here, the point is not to determine whether the performer is closer to his mother tongue or to the foreign language; the objective of this research is to see how this inter-language positioning influences the expression of such an instinctive reaction. (8) “I didn’t know kinesio taping were used for horses as well! That’s great!!” (I: 3; T: dissertation’ subject; S: comment-reaction; E: 4 – instructed environment)

Here, the entire utterance is constituted by the speaker’s comment – the surprise reaction is suggested by punctuation only. The expression ‘I didn’t know’ is used by the speaker to explain why what she was told surprised her. She is not expressing her surprise; rather, she is explaining her interlocutor why she was surprised. It is interesting to notice that, in the same utterance, the speaker comments on two different things: she first explains why she is surprised; then, she comments on what surprised her. It is as if the speaker focused, little by little, on the surprise process; she passes from the cause of her surprise to the surprise itself. However, as already discussed, the natural evolution of surprise would go exactly in the opposite direction; the speaker’s attention is usually focused first on the new piece of information (‘That’s great!!’) and, only then, on the cause of his/her surprise (‘I didn’t know’) in order to re-establish the pre-existing balance. In other words, the typical order of the expression of surprise is reversed. This change may be caused by the fact that the speaker is communicating in her second language. This leads her to focus on the form of the message in order to express herself as clearly as possible, which is detrimental to the instinctive expression of surprise. Some of the expressions present sequences that were never used by the native speakers. It is the case for the comment-reaction and question-comment-reaction sequences (example (9)). (9) “Do you need it even to come to Italy?! I didn’t know that :S” (I: 2; T: her interlocutor needed visa to go to Italy; S: question-comment-­ reaction; E: 4-instructed environment)

The typical sequence of surprise is completely reversed: the interpretation and the questions asked by the speaker precede her instinctive reaction – in this case, the instinctive reaction is reproduced by the smiley ‘:S’; it codes the speaker’s facial

The computer-mediated expression of surprise 137



expression, which is the real instinctive reaction (Pacori, 2010). We have the impression that surprise is so a spontaneous reaction that even only thinking about how to convey it alters it. Of course, it does not mean that, when talking in a foreign language, people experience surprise in a different and reversed way; if this seems to be the case, it is due to the interaction of two different languages only. Speaking in a foreign language is detrimental to fluency and, as a consequence, to the naturalness with which such an instinctive reaction is expressed. 4.3

Specificities of the reaction-segment

The consequence of this contrast between the two languages can be found not only in the order of the different segments but also within the reaction-segment. Attention was then focused on the expression of the surprise reaction in both native and non-native speakers. Despite its short duration, surprise may be expressed in several different ways. Differences were found both in the natives’ and non-natives’ expressions. The reaction-segments were then analysed according to the ways the performers expressed their instinctive reaction of surprise. The main differences concern the use of verbs and the specific use of smileys. 4.3.1 Italian native speakers’ expression of surprise As hypothesised, most of the reaction-segments are composed of verbless sentences (including interjections, false starts and repetitions) (Guillaume, 1973). Only one person used a verb to convey her reaction (example (10)). (10) “Nessuno vuole venirci?!” (“Nobody wants to come?!”) (I: 2; T: none of her interlocutor’s friends wants to go to Paris; S: reaction)

Here, the speaker’s reaction is conveyed through the punctuation, which reproduces the intonation of the utterance. Yet, we need to consider the fact this is an echo-question: the person has reacted by repeating what was said by her interlocutor. Therefore, the sentence might have a verb in it only because she has repeated, word by word, what she was told. As Table 4 shows, the most frequent structures are verb-less questions and verbless exclamatives (Rett, 2008) as in example (11). (11) “Che cosa?!?!? Ma la gente sta fuori?!?!” (“What?!?!? But people are crazy?!?!”) (I: 4; T: talking about baby prostitutes in Rome; S: reaction-comment)

138 Laura Ascone

Table 4.  Reaction-segment in Italian natives’ utterances Reaction-segment

Frequency

Verb-less question Verb-less exclamative Interjection Verb-less exclamative + Smiley Smiley + Verb-less exclamative Smiley + Assertion Verb-less question + Smiley Verb-question

14  8  4  1  1  1  3  1

Table 5.  Reaction-segment in English natives’ utterances Reaction-segment

Frequency

Interjection Verb-less exclamative Verb-less question

13  5  4

4.3.2 English native speakers’ expression of surprise In the English natives’ utterances, the variation of the reaction-segments is slightly reduced. Contrary to the Italian expressions, most of the reaction-segments are composed of interjections. Even though it is argued that verb-less sentences reflect surprise better than the grammatically structured ones (Guillaume, 1973), some of them were expected to present a verb since, in English, verb-less sentences are not as frequent as in Italian. 4.3.3 English non-native speakers’ expression of surprise The absence of verbs both in the English natives’ reaction-segments and in Italian, led to the hypothesis that non-native speakers of English express an instinctive reaction like surprise with a verbless sentence. However, this is not the case; two of the English non-natives’ reactions contain a verb. The use of verbs in the non-natives’ reaction-segments might be due not to the fact that verbless sentences are almost absent in English, but rather to the fact that, since they were talking in a foreign language, the Italian native speakers felt the need to focus on how to convey a message as clearly and accurately as possible in order to be better understood (rather than on the message itself) and, as a consequence, to use full and grammatically structured sentences (example (12)).

The computer-mediated expression of surprise 139



Table 6.  Reaction-segment in English non-natives’ utterances Reaction-segment

Frequency

Verb-less exclamative Verb-less assertion Verb-less question Interjection Verb-less exclamative + Smiley Verb-less question + Smiley Verb-less code-switching Smiley Verb proposition exclamation Verb code-switching

3 2 3 6 2 2 2 6 1 1

(12) “LOL Hugo studying French actually makes me laugh, *couldn’t guess that either” (I: 3; T: news about a common friend; S: reaction-comment; E: 4 – natural environment: three years)

4.4 Code-switching: A specific feature of non-native speakers Sometimes, the non-native speakers overcame the obstacles imposed by the weaker language in a specific way; in some cases, they reacted to the emotional stimulus in their dominant language and pursued the conversation in their weaker language again. The main debate concerning code-switching is whether the mental access to different languages is selective or non-selective; that is, whether speakers choose (more precisely, whether they can choose) to use one language rather than another one. De Groot and Nas (1991) argue that bilingualism is characterised by non-selectivity – i.e. speakers cannot completely separate the two languages. It was seen that, whatever the level of proficiency is, speakers are more likely to have non-selective access to the lexicon of the mother tongue when speaking in a foreign language than vice versa. Sometimes it is the speaker him/herself who decides to switch from one language to another (see Meuter, 2009) either because s/he may encounter some difficulties in expressing the same idea in the language s/he is speaking in, or because s/he may feel that the language in which s/he is speaking does not perfectly reflect what s/he wants to say, or, again, because s/he may want to be better understood. However, in the case of emotions and, even more, of surprise, code-switching is more likely to occur involuntarily. Since emotional arousal is something that is very short in time, the speaker may not have the time to think about the right way

140 Laura Ascone

to convey what and how s/he feels. According to Jarvis (2009), in fact, the more proficient the speaker is in a foreign language, the less s/he will feel the necessity to switch to his/her mother tongue in order to express his/her feelings. In the following example, the person is speaking in English with her interlocutor. Yet, at the very moment when she is given unexpected information, she switches to her mother tongue. Only when surprise has faded away does she switch back to English. (13) “Beccuuuuuu che mi combini?!?! Ahahah il video è fantastico, non ce la facevo a non ridere!! We know how he was, and it’s so weird to see him on a video rapping” (“Beccuuuuuu what are you doing?!?!? Ahahah the video is fantastic, I couldn’t help laughing!!) (I: 3; T: a friend’s news; S: reaction-comment; E: 4 – natural environment: three years)

The person instinctively reacts in her mother tongue; she expresses her surprise adding five ‘u’s to the name of the friend she is talking about, using exclamation and question marks, and laughing. In the second segment – “Il video è fantastico, non ce la facevo a non ridere!!” – the speaker comments on her own reaction. From the reduced number of the exclamation marks, it might be deduced that, in terms of intensity, surprise is much weaker. This is even more obvious in the last segment – “We know how he was, and it’s so weird to see him on a video rapping” –, where the speaker explains why she was surprised in English without using any kind of punctuation. What is common to the three utterances in which code-switching occur is the fact that the speakers reacted in Italian first, and then continued the conversation in English. The fact that the speakers switched to Italian only in the reaction-­ segment confirms Dewaele’s (2010) statement about bilinguals – which may be true for learners and late-bilinguals too –, according to which the instinctive reaction cannot always overcome the obstacles imposed by a foreign language. Therefore, non-native speakers may switch to their mother tongue to convey what they feel in whatever way allows them to express it as closely as possible to how they perceived it (Dewaele, 2010, p. 213). 4.5

Role of smileys

The utterances of eight Italian native speakers present smileys, which are generally used to reproduce the person’s facial expression and, therefore, the person’s emotions and feelings. The use of this code has developed with the emergence



The computer-mediated expression of surprise 141

and evolution of CMC and, more precisely, chats. In fact, in written texts neither smileys nor additional punctuation (e.g. ‘!!!!!’) are used, while in spoken language emphasis is given to facial expressions, tone and intonation. Furthermore, in the entire corpus two classes of smiley can be identified: on the one hand, those conveying emotions (e.g. :O), and, on the other hand, those giving the intonation to the sentence in which they occur (e.g. :)). 4.5.1 Italian native speakers’ expression of surprise The use of smileys characterises the Italian expressions, differently from the text in the English natives’ corpus. In the reaction-segment alone, smileys occur 5 times in the Italian utterances, while they never appear in the English natives’ ones (example (14)). (14) Ancora un altro esame??? :O” (“Another exam??? :O”) (I: 4; T: talking with someone who has an exam; S: reaction)

Here the smiley ‘:O’ conveys both intensity and negative valence. It reproduces the person’s facial expression and the silence caused by the new piece of information; the person is so negatively shocked that he is not able to say anything. When studying the oral expression of surprise, these silences can be perceived more easily. When the person we are talking with is not facing us, we are given the possibility to ‘perceive’ the other person’s silence through this specific smiley and/ or the expression ‘senza parole’ (‘speechless’). On the other hand, if the speaker’s reaction is a positive one, we may find the ‘*-*’ smiley like in example (15). The two asterisks reproduce the speaker’s eyes conveying her fascination with the new piece of information. (15) “Ma che figata!!!! Ho visto qualche foto..il paradiso!! *-*” (“How cool!!!! I’ve seen some pictures..paradise!! *-*”) (I: 2; T: talking about a place he has visited; S: reaction-comment-reaction) (16) “OMG!!! xD” (I: 4; T: she has seen a picture of a friend of hers; S: reaction)

Here, ‘xD’ reproduces the speaker’s eyes (‘x’) and mouth (‘D’), and it is quite a neutral expression. The speaker’s reaction is neither positive nor negative; the smiley simply reflects the unexpectedness and the disruption that the news creates between the person’s expectations and what actually happened.

142 Laura Ascone

4.5.2 English native speakers’ expression of surprise This way of conveying surprise does not appear in any of the English natives’ utterances; yet, an element similar to smileys occurs more than once: ‘lol’, which literally means ‘laughing out loud’. 4.5.3 English non-native speakers’ expression of surprise Compared to the Italian utterances, where the use of smileys is reduced, and to the English natives’ ones, where it is completely absent, in the English non-­natives’ expressions smileys are more frequent; they appear in 12 expressions. Rather than trying to steer the wave of emotions through the bottleneck of the weaker language, the emotions “spill over” and get verbalised in whatever language allows the speaker channel his/her feeling in a satisfactory manner.  (Dewaele, 2010, p. 213)

Here, Dewaele refers to bilinguals – i.e. people who have two mother tongues. Yet, this is even truer for learners of a foreign language or late-bilinguals. More than bilinguals, they may find it difficult to express something in their weaker language, which would lead them to switch to their mother tongue to express what they feel in the best possible way. Furthermore, the word ‘language’ should not be restricted to a language strictly speaking – e.g. Italian, English, French. It might be interpreted as any means of communication, including, for instance, smileys. It is not by chance that the structures favoured by the non-native speakers are those including smileys – e.g. ‘O.O’, which reproduces wide-opened eyes –, or expressions like ‘aaaaahh’ (example (17)). (17) “Who is that?? …aaaaahh O.O” (I: 3; T: picture of a friend of hers; S: question-reaction; E: 4-natural environment: three years)

The second language does not allow the speakers to fully express their surprise, which leads them to use ‘sounds’ like ‘aaaaahh’ and ‘visual’ expressions like smileys. The absence of this category in the English natives’ utterances and the reduced use in the Italian ones, leads us to put forward the following hypothesis: the use of these ‘musical’ and ‘visual’ expressions may be linked to the fact that the speakers are communicating in their second language, and not to the fact that they want to reproduce their facial expressions. In other words, when communicating in a foreign language, speakers may tend to use forms and codes – e.g. smileys or interjections – that allow them to express themselves more naturally without thinking of the form of the message. Therefore, smileys are not used to reproduce facial expressions only. It is not by chance that they are more frequent in the non-natives’ expressions than in the natives’ ones (Dewaele, 2010).

The computer-mediated expression of surprise 143



4.6 Surprise lexicon Whatever language is used, the way surprise is expressed reflects the disruption between the speaker’s knowledge and unexpected information. It also reveals whether the person is reacting positively or not, and how intense his/her reaction is. This final part reports on these three characteristic features of surprise: disruption (why the speaker is surprised and how he/she expresses this verbally), valence (what makes us understand whether the reaction is positive, negative or neutral) and intensity (how the speaker conveys the intensity of his/her surprise). 4.7

Surprise as a gap between expectations and reality

The gap between someone’s expectations and unexpected information is reflected in some specific words used in most of the examples. 4.7.1 Italian native speakers’ expression of surprise In Italian, we have adversative conjunctions like ‘ma’ and ‘però’, which mean but (example (18)). (18) “Ma daiii ti ha scritto lui?” (“Really? Did he write you?”) (I: 4; T: news about a friend of hers; S: reaction-question)

Here, ‘ma’ suggests an opposition between what the speaker knows about the person they are talking about and what she has just discovered. In fact, someone who does not know this person would not have had the same reaction. Among the Italian utterances, two other words reflect this disruption: ‘davvero’ and ‘veramente’. (19) “Ma che davvero???” (“But really???”) (I: 4; T: a friend has moved to London; S: reaction)

‘Davvero’ suggests that the piece of information goes against the speaker’s expectations. However, ‘davvero’ is not always linked to the gap between news and someone’s expectations. In example (19), the person asks if what she was told is true even if she is aware that it is. Asking for more information without waiting for the answer to his/her question (‘davvero???’) shows that the speaker does not question the semantic content of the information.

144 Laura Ascone

4.7.2 English native speakers’ expression of surprise In the English natives’ utterances, this disruption is reflected in words like ‘no way’ or ‘really’, which are far less frequent compared to ‘veramente’ and the other Italian expressions. (20) “Really? I love France! Wherabout in France are you?” (I: 1; T: talking with someone who lives in France; S: reaction-­comment-­ question)

In example (20), we are under the impression that the person does not believe what he has discovered, which leads him to ask for confirmation. He knows that the person he is talking with is Italian; in fact, he finds the fact that his interlocutor lives in France surprising. Yet, ‘really?’ is not a real question. He does not expect an answer from his interlocutor; in fact, he continues commenting and asking more questions about what he has been told. 4.7.3 English non-native speakers’ expression of surprise In the utterances by the English non-native speakers, the opposition between someone’s expectations and unexpected news is conveyed with different and more grammatically structured expressions: can’t believe, couldn’t guess, didn’t know, or would have never thought it possible (example (12)). In the English natives’ expressions there are no such clear and detailed expressions. If we compare example (12) with an English native’s utterance with the same sequence reaction-­ comment (example (21)), we can see that the latter looks more spontaneous than the English non-native’s utterance because of the grammatical structures of the two comment-segments. (21) “Oh that’s sooo cool I wish I could do that” (I: 1; T: video of someone playing the violin; S: reaction-comment)

While in example (21) the speaker expresses his ideas with simple and, therefore, more spontaneous sentences, in example (12) the structure of the comment-segment is far more complex (François and Legallois, 2012). ‘LOL’ (laughing out loud) already reproduces her reaction; yet, the speaker rephrases it (‘Hugo studying French actually makes me laugh’). This repetition is detrimental to the reaction spontaneity. The negative form of the modals could and would accentuates this gap between the person’s expectations and reality. Could features unreality, and its negative form expresses impossibility in the past, in opposition to the present. In example (12), the person did not consider the event possible, which, somehow, is the source of her surprise; her expectations – i.e. ‘this cannot happen’ – are not in line with what the other person has just told her – i.e. ‘this has happened’. At the

The computer-mediated expression of surprise 145



same time, the form ‘would + have –en’ expresses irrealis in the past. These two aspects reflect the incompatibility between what was considered possible/impossible, and what actually happened. 4.8 The encoding of valence in chats The use of specific words is relevant in the expression of valence; while intensity is mostly expressed through punctuation (‘!!!’) and additional vowels (‘aaaaah’), valence is conveyed through words, mostly adjectives and adverbs, or smileys, and never through punctuation and vowels. In this section, I will analyse how valence is expressed in the different groups of speakers. Tables 7, 8 and 9 show the different ways in which valence was expressed. There is no real difference between the Italian and the English expressions, except for one particular category that appears only in the Italian and English non-native speakers’ utterances: the context category. Here, the speakers do not use words conveying their positive, negative or neutral reactions; rather, it is the context that allows the interlocutor to understand how the speaker perceives the new information. This is made possible either by the fact that information is announced as being a positive or a negative one – e.g. I have some bad news for you (Maynard & Freese, 2012, p. 94), or by the fact that the stimulus is in contrast with what is Table 7.  How valence is expressed in Italian natives’ utterances Valence

Frequency

Word Smiley Context

26  5  3

Table 8.  How valence is expressed in English natives’ utterances Valence

Frequency

Word Smiley

33  3

Table 9.  How valence is expressed in English non-­natives’ utterances Valence

Frequency

Word Smiley Context

19  6  2

146 Laura Ascone

considered usual by the two interlocutors (see Macedo & Cardoso, 2001). In other words, the speaker giving surprising information expects his/her interlocutor to react either positively or negatively. The fact that the reaction’s valence may be perceived thanks to the context and not only thanks to the specific terms employed, suggests that the notion of surprise valence is controversial. As already mentioned, some researchers like Ortony et al. (1987) consider surprise as a reaction deprived of valence; rather, surprise refers to something unexpected and, once the person has evaluated the surprising event, it becomes either a positive or a negative surprise (p. 33) (see also Ekman & Friesen, 2003; Valetopoulos, 2013). Therefore, the category context comprises the surprise expressions that, if extracted from the contexts in which they occur, cannot be evaluated as positive or negative reactions (example (22)). (22) Are you sure?? (I: 3; T: talking about an exam; S: question; E: 4 – immersion class)

Announcing that their exam has been anticipated, the speaker’s interlocutor knows that the other person will react negatively. Yet, in another context, the same utterance might have a positive valence. Another difference between the way intensity and valence are conveyed is the place where they appear in the sentence. Intensity is almost always expressed in reaction-segments; people reveal whether they are very surprised or not in their instinctive reaction. When they comment on or wonder about new information, surprise has almost disappeared; as a consequence, comment- and question-­ segments hardly ever convey the reaction’s intensity. On the other hand, valence is likely to be conveyed in comment-segments. As stated before, it is only by interpreting the unexpected piece of information that we can say whether we are happy or not about the unexpected news. Furthermore, contrary to intensity, there is no link between valence and the level of intimacy between the two speakers. 4.9 The encoding of intensity in chats In oral exchanges, the spontaneity of surprise emerges in a more natural way than in written texts – e.g. through facial expressions, the pitch of the voice, etc. Computer-­mediated language reproduces this genuineness in a different way. In most of the utterances collected for this study, intensity is conveyed in a way that is not accepted in written productions. People often add vowels (‘wooow’, ‘whaaaaat’) or exclamation or question marks (‘davvero????’, ‘oh my god!!!’, ‘really?!’). Moreover, instead of using specific words, like superlatives, to express the intensity of their reaction, people intensify or weaken surprise through punctuation and smileys.

The computer-mediated expression of surprise 147



Table 10.  How intensity is expressed in Italian native’s utterances Intensity

Frequency

Punctuation Vowel variation Word Capital letters Smiley Neutral

32 11  4  1  1  3

4.9.1 Italian native speakers’ expression of surprise In 35 expressions, intensity is conveyed through punctuation, smileys and/or vowel variation (e.g. ‘woooow’). In 4 utterances only, intensity is conveyed using adverbs (‘veramente’) or adjectives (‘sconvolta’); moreover, in the 4 cases, intensity is expressed through punctuation and vowels as well. Intensity is never conveyed through words only. In the remaining examples, intensity is absent, which is probably a consequence of the lack of intimacy between the interlocutors. In fact, in two of the three neutral expressions, the two persons have just met, while in the other case the level of intimacy is 2 out of 4. 4.9.2 English native speakers’ expression of surprise The way intensity is conveyed by the Italian natives speaking in English is closer to how they express it in their mother tongue than to the way it is conveyed by the English native speakers. People tend to express intensity through punctuation rather than with specific words. In 6 expressions, it is conveyed through both punctuation and words. The remaining 11 examples are quite neutral; there is no clear link between the neutral valence and the level of intimacy of the two persons. Intensity is one of the most important semantic dimensions to determine surprise (Zammuner, 1998). The two control corpora present two very different ways of conveying this reaction’s intensity: contrary to the Italian speakers, who mostly Table 11.  How intensity is expressed in English non-natives’ utterances Valence

Frequency

Punctuation Vowel variation Word Neutral Smiley Capital letters Repetition of a word

18  5 10 11  2  2  1

148 Laura Ascone

express intensity with punctuation, the English natives tend to favour the use of specific words. The English non-natives’ utterances present the same characteristics as the Italian natives’ expressions. This similarity in the expression of intensity – which is one of the main features of the expression of surprise – confirms the influence of the mother tongue when expressing an instinctive reaction like surprise in a foreign language. 5. Conclusion Despite the simplicity and naturalness with which we experience emotions, it has always been difficult to find a definition able to capture all their various components. By retracing some of the basic findings of the literature on emotions, and by adopting different approaches to the study of these psycho-physiological reactions, this research has unveiled relevant divergences between surprise and other emotions. The differences in the way we experience them – stimulus-­interpretationreaction in emotions like fear and anger, and stimulus-reaction-­interpretation in surprise – as well as in their duration, leads us to consider surprise as the source of emotions. This interpretation of surprise has constituted the starting point of this study on the expression of surprise. Most of the past research conducted on this specific field examined the way surprise is expressed by monolinguals, collecting data in laboratories or from written productions. With this paper, the objective was to provide an analysis of how learners and late-bilinguals convey surprise in a foreign language when communicating in chats. Despite the limited dimension of the corpus used and the parameters chosen for this analysis – which, however, have allowed us to investigate the computer-­ mediated expression of surprise more in depth –, this study has brought interesting results to light, which might prompt further research on a larger corpus. The verbal expression of surprise tends to follow its cognitive process; most of the utterances presented the speaker’s instinctive reaction to the stimulus at the beginning of the sequence, followed by his/her comment on and/or questions about the unexpected information. This tendency was identified in both the native and non-native speakers. Yet, in the latter, the expression of surprise turned out to be less natural because of the fact speakers were communicating in a foreign language. The obstacles imposed by the weaker language were surmounted by the non-native in two ways: switching to their mother tongue or using smileys. The occurrence of these specific symbols has paved the way to the investigation of the codes peculiar to CMC, and of the way they are employed to convey surprise. Smileys, which never occur in the English natives’ utterances, are used by the Italian speakers to communicate the valence and intensity of their reaction. Contrary



The computer-mediated expression of surprise 149 .

to valence, which is expressed mainly through smileys, adjectives and adverbs, intensity is conveyed using punctuation and additional vowels (e.g. ‘aaaaah!!!’). This study reveals the influence that a mother tongue, the language-learning background (i.e. instructed/natural environment) and the kind of communication (in this case, CMC) have when expressing an instinctive reaction, such as surprise, in a foreign language. However, further research is needed in the expression of surprise in non-native speakers as well as in the way it is conveyed in CMC. It might be interesting to conduct a parallel study; a comparison with research on how English natives express surprise in Italian in chats and social networks might lead to more general and solid findings about the relationship between language and surprise, and between first and second language in the expression of instinctive reactions. Furthermore, this study reveals a different use of smileys in Italian and English. Analysing the expression of surprise by non-­ natives of different origins might be a way to examine the use of this specific code in diverse cultures. To conclude, analysing the correlation between the expression employed to convey surprise – i.e. reaction-segment – and the valence of the entire surprise expression, might reveal whether specific expressions are favoured to convey either a positive or a negative surprise, which might be crucial to the understanding of the notion of surprise valence.

References Aijmer, K. (2002). English discourse particles: Evidence from a corpus. Amsterdam: John Benjamin.  doi: 10.1075/scl.10 Arndt, H., & Janney, R. W. (1991). Verbal, prosodic, and kinesic emotive contrasts in speech. Journal of Pragmatics, 15(6), 521–549.  doi: 10.1016/0378-2166(91)90110-J Bloom, L., & Beckwith, R. (1989). Talking with feeling: Integrating affective and linguistic expression in early language development. Cognition & Emotion, 3(4), 313–342. doi: 10.1080/02699938908412711 Caffi, C., & Janney, R. W. (1994). Toward a pragmatics of emotive communication. Journal of Pragmatics, 22(3), 325–373.  doi: 10.1016/0378-2166(94)90115-5 Colomb, G. G., & Simutis, J. A. (1996). Visible conversation and academic inquiry: CMC in a culturally diverse classroom. In S. Herring (Ed.), Computer-mediated communication: Linguistic, social, and cross-cultural perspectives (pp. 203–222). Amsterdam. John Benjamins. doi: 10.1075/pbns.39.16gol Dewaele, J.-M., & Pavlenko, A. (2001–2003). Web questionnaire: Bilingualism and emotions. London: University of London. Dewaele, J.-M. (2010). Emotions in multiple languages. London: Palgrave MacMillan. doi: 10.1057/9780230289505 De Groot, A., & Nas, G. L. (1991). Lexical representation of cognates and noncognates in compound bilinguals. Journal of Memory and Language, 30(1), 90–123. doi: 10.1016/0749-596X(91)90012-9

150 Laura Ascone

Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6(3–4), 169–200. doi: 10.1080/02699939208411068 Ekman, P., & Friesen, W. (2003) [1975]. Unmasking the face. Cambridge, MA: Malor Books. Fehr, B., & Russell, J. A. (1984). Concept of emotion viewed from a prototype terspective. Journal of Experimental Psychology: General, 113(3), 464–486.  doi: 10.1037/0096-3445.113.3.464 François, J., & Legallois, D. (2012). Définition et illustration de la notion d’expressivité en linguistique. In N. Le Querler, F. Neveu, & E. Roussel (Eds.), Relations, connexions, dépendances: Hommage au Professeur Claude Guimier. Rennes: Presses Universitaires de Rennes. Grandjean, D., & Scherer, K. R. (2008). Unpacking the cognitive architecture of emotion processes. Emotion, 8(3), 341. Grinshpun, Y. (2008). Ô, entre langue(s), discours et graphie. Paris: Ophrys. Gross, J. J., & Thompson, R. A. (2007). Emotion regulation: Conceptual foundations. New York: Guilford Press. Guillaume, G. (1973). Leçons de linguistique de Gustave Guillaume, 1948–1949, série C: Grammaire particulière du français et grammaire générale (IV). Published by R. Valin. Québec/ Paris: Presses de l’Université Laval. Hakulinen, A., & Sorjonen, M.-L. (2012). Being equivocal: Affective responses left unspecified. In A. Perakyla & M.-L. Sorjonen (Eds.), Emotions in interaction (pp. 147–173). Oxford: Oxford University Press.  doi: 10.1093/acprof:oso/9780199730735.003.0007 Harris, C. L., Gleason, J. B., & Aycicegi, A. (2006). When is a first language more emotional? Psychophysiological evidence from bilingual speakers. In A. Pavlenko (Ed.), Bilingual minds: Emotional experience, expression and eepresentation (pp. 257–283). Clevedon: Multilingual Matters. Herring, S. C. (2001). Computer-mediated discourse. In D. Schiffrin, D. Tannen, & H. Hamilton (Eds.), The handbook of Discourse Analysis (pp. 612–634). Oxford: Blackwell. Herring, S. C. (2010). Computer-mediated conversation: Introduction and overview. Language@ Internet, 7(2). Jarvis, S. (2009). Lexical transfer. In A. Pavlenko (Ed.), The bilingual mental lexicon: Interdisciplinary approaches (pp. 99–124). Bristol: Multilingual Matters. Jucker, A. H., & Smith, S. W. (1998). And people just you know like ‘wow’: Discourse markers as negotiating strategies. In A. H. Jucker & Y. Ziv (Eds.), Discourse markers: Description and theory (pp. 171–201). Amsterdam: John Benjamins.  doi: 10.1075/pbns.57.10juc Ko, K. K. (1996). Structural characteristics of computer-mediated language: A comparative analysis of InterChange discourse. Electronic Journal of Communication/La Revue Électronique de Communication, 6(3). Kövecses, Z. (2003). Metaphor and emotion: Language, culture, and body in human feeling. Cambridge: Cambridge University Press. Krashen, S. D. (1981). The “Fundamental Pedagogical Principle” in second language teaching. Studia Linguistica, 35(1–2), 50–70.  doi: 10.1111/j.1467-9582.1981.tb00701.x LaBar, K. S., & Phelps, E. A. (1998). Arousal-mediated memory consolidation: Role of the medial temporal lobe in humans. Psychological Science, 9(6), 490–493. doi: 10.1111/1467-9280.00090 Macedo, L., & Cardoso, A. (2001). Creativity and surprise. In G. Wiggins (Ed.), Proceedings of the AISB’01 symposium on creativity in arts and science (pp. 84–92). York, UK: The Society for the Study of Artificial Intelligence and Simulation Behaviour.



The computer-mediated expression of surprise 151

Macedo, L., Reisenzein, R., & Cardoso, A. (2004). Modeling forms of surprise in artificial agents: Empirical and theoretical study of surprise functions. In K. Forbus, D. Gentner, & T. Regier (Eds.), Proceedings of the 26th annual conference of the Cognitive Science Society (pp. 873–878). Mahwah, NJ: Lawrence Erlbaum Associates. Marsella, S. C., & Gratch, J. (2009). EMA: A process model of appraisal dynamics. Cognitive Systems Research, 10(1), 70–90.  doi: 10.1016/j.cogsys.2008.03.005 Maynard, D. W., & Freese, J. (2012). Good news, bad news, and affect: practical and temporal ‘emotion work’in everyday life. In A. Peräkylä & M.-L. Sorjonen (Eds.), Emotion in interaction (pp. 92–112). Cambridge: Cambridge University Press. doi: 10.1093/acprof:oso/9780199730735.003.0005 Maynor, N. (1994). The language of electronic mail: Written speech. In G. D. Little & M. Montgomery (Eds.), Centennial usage studies (pp. 48–54). Tuscaloosa, AL: University of Alabama Press. Meuter, R. (2009). Neurolinguistic contributions to understanding the bilingual mental lexicon. In A. Pavlenko (Ed.), The bilingual mental lexicon: Interdisciplinary approaches (pp. 1– 25). Bristol: Multilingual Matters. Murray, D. E. (1985). Composition as conversation: The computer terminal as medium of communication. In L. Odell & D. Goswami (Eds.), Writing in nonacademic settings (pp. 203– 228). New York: Guilford. Ortony, A., Clore, G. L., & Foss, M. A. (1987). The referential structure of the affective lexicon. Cognitive Science, 11, 341–364.  doi: 10.1207/s15516709cog1103_4 Pacori, M. (2010). Come interpretare i messaggi del corpo. Bari: Giunti Editore. Pavlenko, A. (2008). Emotion and emotion-laden words in the bilingual lexicon. Bilingualism: Language and Cognition, 11(2), 147–164. Plantin, C. (2011). Les Bonnes raisons des émotions: Principes et méthode pour l’étude du discours émotionné. Bern: Peter Lang. Reeve, J. (2012). Psychologie de la motivation et des émotions. Brussels: De Boeck. Rett, J. (2008). A degree account of exclamatives. In T. Friedman & S. Ito (Eds.), Proceedings of SALT XVIII (pp. 601–618). Ithaca, NY: Cornell University. Rusinek, S. (2004). Les émotions: Du normal au pathologique. Paris: Dunod. Scherer, K. R. (1984). On the nature and function of emotion: A component process approach. In K. R. Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 293–317). Hillsdale, NJ: Erlbaum. Scherer, K. R. (2001). Appraisal considered as a process of multilevel sequential checking. In K.  R. Scherer, A. Schorr, & T. Johnstone (Eds.), Appraisal processes in emotion: Theory, methods, research (pp. 92–120). Oxford/New York: Oxford University Press. Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52, 1061–1086.  doi: 10.1037/0022-3514.52.6.1061 Sorianello, P. (2012). A prosodic account of Italian exclamative sentences: A gating test. In Q. Ma, H. Ding, & D. Hirst (Eds.), The proceedings of the 6th international conference on speech prosody, Shanghai (China), vol. 1 (pp. 298–301). Shanghai: Tongji University Press. Valetopoulos, F. (2013). Traduire la peur et la surprise: Une étude contrastive en grec moderne et en français. In H. Chuquet, R. Nita, & F. Valetopoulos (Eds.), Des sentiments au point de vue: Études de linguistique contrastive (pp. 95–116). Rennes: Presses Universitaires de Rennes.

152 Laura Ascone

Wierzbicka, A. (1999). Emotions across languages and cultures: Diversity and universals. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511521256 Wilkinson, S., & Kitzinger, C. (2006). Surprise as an interactional achievement: Reaction tokens in conversation. Social Psychology Quarterly, 69(2), 150–182. doi: 10.1177/019027250606900203 Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151–175.  doi: 10.1037/0003-066X.35.2.151 Zammuner, V. L. (1998). Concepts of emotion: “Emotionness”, and dimensional tatings of Italian emotion words. Cognition & Emotion, 12(2), 243–272.  doi: 10.1080/026999398379745 Zanuttini, R., & Portner, P. (2003). Exclamative clauses: At the syntax-semantics interface. Language, 79(1), 39–81.

Surprise routines in scientific writing A study of French social science articles* Agnès Tutin

Université Grenoble Alpes

Emotions are almost absent from scientific articles except surprise, which plays a specific role in this genre. Surprise markers such as contrairement à nos attentes (‘contrary to our expectations’) or ce résultat relativement surprenant (‘this somewhat surprising result’) are used in the framework of a scientific prediction model, implicit or explicitly formulated. A corpus linguistic study of adjectival and verbal markers allows us to determine several trends: (a) contrary to other genres such as novels or newspaper articles, surprise is not polar, that is to say it is neither positive nor negative, but stylistically “neutral”, (b) surprise is far more source-oriented than experiencer-oriented, (c) surprise generally involves the reader as a witness, and contributes, with other rhetorical markers, to “interlocutive” dialogism, (d) surprise often occurs in a prefabricated discursive scenario including several steps: (i) prediction model, (ii) observed facts, (iii) expression of surprise (or absence of surprise), (iv) explanation of surprising facts. Finally, we can question the status of surprise as an affect in scientific writing. It seems to be more a state of consciousness associated with the observation of complex facts. In any case, it appears to be a complex state, with rich conceptual content. Keywords: scientific writing, emotions, surprise, stance

* Special thanks to my colleagues Francis Grossmann, Iva Novakova, Julie Sorba, for their stimulating comments on a first version of this paper, as well as Anne Jugnet, in the framework of the “Symposium on Describing and Expressing Surprise”, organized by Agnès Celle and Laure Lansari. I would also like to thank warmly Laura Hartwell and Lucy Garnier for their thorough revision of the English. doi 10.1075/bct.92.06tut © 2017 John Benjamins Publishing Company

154 Agnès Tutin

1. Introduction This paper aims at identifying the linguistic properties of the semantic field of surprise in scientific writing, especially in French research articles in the social sciences. Studying affects in scientific texts might seem a little… surprising since scientific writing is often considered as a neutral genre where the subjective dimension is rarely present, although recent studies have shown that scientific authors are often more visible in their writing than is frequently believed (e.g. Fløttum, Dahl, & Kinn, 2006; Tutin & Grossmann, 2014). This subjective dimension frequently occurs in attitude and stance markers such as evaluative lexicon (e.g. Swales & Burke, 2003; Tutin, 2010) or opinion markers (e.g. Hyland, 2002). However, the emotional dimension, strictly related to feelings, is almost completely absent from this genre with the exception of surprise, which is quite widely used in contexts such as the following: (1)1 Ce résultat relativement surprenant s’explique, à nos yeux, par une définition… ‘This somewhat surprising result can be explained, in our view, with a definition…’ (2) Sans surprise, ces résultats nous ont permis de valider l’hypothèse posée comme générale. ‘Unsurprisingly, these results have allowed us to validate our general hypothesis.’

Surprise markers involve lexical expressions meaning “an unexpected or astonishing event, fact, etc.” (Oxford English Dictionary), whether they are multiword expressions or not, and regardless of their part of speech. The presence of this affect in scientific articles could be related to the fact that surprise is non-polar, i.e. neither a positive nor a negative affect but a neutral one, and is therefore not a prototypical emotion (Ortony, Clore, & Collin, 1999). In this particular genre, the expression of surprise is almost always a reaction to scientific facts (results, statements, etc.) and the experiencer of the affect is generally the scientific author(s) or the scientific community. Actually, as we will see, surprise markers seem to be used to establish a strong reader-writer interaction, similar to the use of statement verbs as evidential markers (when confirming the validity of scientific facts) (see

1. Translations are provided to allow readers who do not speak French to better understand the examples. For this reason, these translations follow the French syntax as far as possible. However, due to the differences between French and English, this is not always possible and the reader should bear in mind that the features under discussion are always those present in the original examples rather than the translations.



Surprise routines in scientific writing 155

Grossmann & Tutin, 2010a) or hedges and epistemic modality in order to enable interpersonal negotiation (e.g. Hyland, 1998). In this article, we aim to demonstrate that the expression of surprise is fully routinized in this specific genre, not only where phraseology and specific prefabricated expressions are concerned, but also in the use of stereotyped rhetorical functions regarding scientific evidence. We also wish to explore the extent to which the expression of surprise is similar or different here than in other written genres. We begin with an overview of the notion of surprise in scientific articles and its relation to rhetorical strategy. We will then present a corpus study of adjectival and verbal surprise markers derived from a sub-corpus of the Scientext project (which includes 344 research articles across 10 disciplines) before analyzing syntactic and semantic properties. These elements will then be used in the final section devoted to enunciative, i.e. formal linguistic properties involving the reader-writer interactions, and discursive and textual properties. 2. Scientific writing and surprise markers Scientific writing is an important part of scientific activity, especially in the field of social sciences and humanities, and studies focusing on the genre of published research articles reveal interesting facts about the epistemology of disciplines. Far from being neutral, scientists produce strongly argumentative texts where authorial position can be expressed in the evaluation of scientific objects and opinions, as well as in positions taken towards peers (Grossmann, Tutin, & Garcia Da Silva, 2009; Tutin & Grossmann, 2014). However, the expression of emotion is quite rare in this genre, with the exception of surprise, which often appears in routinized expressions such as de façon surprenante (‘surprisingly’), les résultats sont étonnants (‘the results are surprising’), contre toute attente (‘against all expectations’), and il est surprenant que (‘it is surprising that’). Emotions are unlikely to be central to scientific texts, contrary to other genres such as readers’ letters or editorials in newspapers (cf. Plantin, 2011). As scientific discourse is not built on emotions, we can therefore assume that surprise markers are intentionally used as part of a rhetorical strategy. Following Plantin (2011) and Celle and Lansari (2014), after Caffi and Janney (1994), we can contend that surprise does not fall within the scope of “emotional communication”, where the emotion is spontaneously expressed, but rather within the scope of “emotive communication” with the “intentional strategic signalling of affective information in speech and writing” (Caffi & Janney, 1994, p. 348). As mentioned by Plantin (2011, p. 140), “intentional communication and

156 Agnès Tutin

the strategic use of emotions contrast with the incursion of emotions in language activity… Controlled emotion contrasts with experienced emotion.”2 According to Boch, Grossmann, and Rinck (2007) and Grossmann and Wirth (2007), surprise markers belong to markers of non-congruence and are used in the framework of an argumentative device related to expectations: “The writer creates expectations in the reader, awakening curiosity by showing interest in the issue, by not readily providing an answer to the questions…” (Boch et al., 2007, p. 112, our translation). Expectations can be explicitly formulated in a “prediction model” (we assume that… our hypothesis is… our method should find…), especially in the introductory section of a research article, as the following example illustrates: (3) La première hypothèse postule qu’un état émotionnel suffisamment intense devrait produire une quantité substantielle de pensées intrusives… [psychology-407-introduction] ‘The first hypothesis is that a sufficiently intense emotional state is expected to produce a substantial amount of intrusive thoughts.’ More frequently, though, expectations are implicitly assumed and are related either to the “doxa” (e.g. “scientific texts are devoid of subjectivity”) or to peershared scientific knowledge (e.g. “the emotion of surprise has a specific status in the field of emotions, as it is non polar”). In scientific writing, as in other genres (see Celle & Lansari, 2014; Kern & Grutchus, 2014), surprise is generally formulated when observed facts contradict expectations. By studying surprise markers in scientific writing, we also wish to explore the extent to which surprise markers are similar to those observed in other genres such as literature or newspapers. Since we are mainly concerned by the issue of surprise, we will focus on non-congruence markers although congruence markers are probably involved in the same kind of rhetorical process as surprise. We will look at the extent to which there is a semantic and syntactic specificity to scientific writing. Though surprise is not a polar affect itself, is polarity involved in the surrounding context of surprise markers? And in this genre, can we consider, as claimed by Kövecses (2003, p. 33), that “surprise is not a socially very complex phenomenon, and consequently, that there is not a great amount of conceptual content to be associated with it”?

2. Our translation.

Surprise routines in scientific writing 157



3. Corpus study of the semantic field of surprise in scientific articles 3.1

Methodology

In this study, we focus on expressed emotions relying on explicit surprise markers in a corpus of 344 French research articles in 10 disciplines within the social sciences and humanities,3 drawn from French peer-reviewed journals (3.2 million words). This subcorpus is part of the extended Scientext Corpus, a free corpus available online developed in the framework of an ANR4 project. The corpus has been tagged using a syntactic parser, and allows queries on specific lemmas, parts of speech, or syntactic functions. For example, it is possible to extract all the nouns syntactically related to a surprise adjective (frappant ‘striking’, surprenant ‘striking’ or étonnant ‘astonishing’), with an attributive or a predicative relation, as shown in Figure 1 and in Figure 2. Surprise markers were extracted from texts with queries using available lexical resources of emotion for French: (a) the lexicon for affective annotation (Augustyn et al., 2008) and (b) the Emotaix Lexicon (Piolat & Bannour, 2009) developed for the study of affective discourse in psychology. These resources already contain lexical classes of emotions such as ‘surprise’, ‘joy’, and ‘sadness’. Words included in the category of ‘surprise’ were selected and merged. The list of surprise words includes 51 adjectives, 17 nouns, 24 verbs, and 6 adverbials5 (see Appendix). Although this lexical list may not be exhaustive, we believe that it is representative of single words belonging to the semantic field of surprise.

Figure 1.  A syntactic query (nouns syntactically related to 3 adjectives of surprise)

3. The disciplines in question are anthropology, economics, geography, history, linguistics, psychology, educational sciences, political sciences, information sciences, and sociology. 4. Agence Nationale de Recherche – the French National Research Agency, which provides funding for research projects on a selective basis. The website of the project is: http://scientext. msh-alpes.fr 5. The list of adverbials will need to be completed.

158 Agnès Tutin

Figure 2.  Surprise adjectives syntactically related to nouns (extract of the concordances)

Drawing upon these resources, relevant concordances of our corpus were disambiguated, since surprise lexicon can be ambiguous. For example, frapper (‘to strike’) is quite frequent as a verb of surprise. However, literal meanings can also be encountered, as in the following example: (4) La misère frappe violemment les populations les plus fragiles. [sed-the-480-body] ‘Misery violently strikes the most vulnerable populations.’

Furthermore, we only selected surprise contexts associated with the writer/reader of the scientific text,6 and consequently surprise expressions associated with external experiencers were not taken into account. For example, we excluded example (5) because surprise is not associated to the reader/writer but to one of the survey respondents in a study of anthropology. Conversely, example (6) was selected, since surprise is associated with the reader/writer of the article, even though the experiencer is not explicitly present. (5) Lorsque je précisais à mes enquêtés être issue d’une “Faculté des Sciences du Sport”, certains furent étonnés, d’autres, énervés …  [ant-art-415-body] ‘When I explained to my interviewees being from a “Faculty of Sport Sciences”, some were surprised, others angry …’ (6) Ces deux extraits présentent un parallélisme frappant.  ‘These two excerpts show a striking parallelism’

6. Which are statistically the most frequent cases.

[lin-the-646-body]

Surprise routines in scientific writing 159



Surprise words all have their semantic specificity but, like most emotion words, they all involve two semantic roles, at least implicitly: an experiencer and a source (or a cause). Some lexical units are experiencer-oriented, for example s’étonner (‘to be surprised’), while others are source-oriented like surprenant (‘surprising’) or étonner (‘to astonish’). Nouns (e.g. surprise, étonnement ‘surprise, astonishment’) express the concept with a neutral orientation. As we will see below, in scientific writing, experiencers are often implicit while sources are always explicitly present. 3.2

Lexical markers of surprise

Our corpus extraction and selection revealed the presence of 325 occurrences of surprise markers, mostly adjectives and verbs (226 adjectives, 8 nouns, 75 verbs, and 16 adverbials). Our corpus of more than 3.2 million words contained relatively few references to surprise. Nouns and adverbials were the most infrequent; therefore we decided to focus on the two most productive parts of speech, adjectives and verbs, by analyzing their semantic and syntactic features. The syntactic and semantic analyses will then be used in the enunciative and discursive analysis in Section 4. 3.2.1 Surprise adjectives As mentioned above, surprise adjectives were the most frequent part of speech found in our corpus (226 occurrences7). All adjectives were source-oriented (résultats surprenants ‘surprising results’, il est étonnant que … ‘it is astonishing that …’), which is to say that all of them were related to sources and not to the experiencer. Nevertheless, it should be noted that the words surpris (‘surprised’) and étonné (‘astonished’), which are both experiencer-oriented, were analyzed as past-­participle verbal forms and not as adjectives (see 3.2.3). However, we will see below that surpris and étonné were anyhow infrequent in our corpus, probably due to the impersonal style of scientific writing. Four adjectives (surprenant ‘surprising’, frappant ‘striking’, étonnant ‘astonishing’, inattendu ‘unexpected’) constituted 70% of this category, dominating the distribution as can be seen in Figure 3. Several adjectives that are frequent in general language (merveilleux ‘marvellous’, incroyable ‘incredible’, stupéfait ‘amazed’) were almost completely absent, along with very intense adjectives, adjectives including a polar dimension, or adjectives belonging to a colloquial register. Scientific writing therefore uses the most neutral and the least polar surprise adjectives.

7. They include -ant forms erroneously analyzed as verbs.

160 Agnès Tutin

Saisissant ‘startling’ Bizarre ‘bizarre’

; ;

Curieux ‘puzzling’

; 

Etrange ‘strange’

; 

Inattendu ‘unexpected’

; 

Etonnant ‘astonishing’

; 

Frappant ‘striking’

; 

Surprenant ‘surprising’

; 

Figure 3.  The most frequent adjectives of the semantic field of surprise

As mentioned above, there was almost no explicit mark of the experiencer, even in a large co-text, for example in the surrounding sentences. Implicitly, when unexpected facts are observed, the experience of surprise is shared by both writer and reader, and even the discourse community (see Section 4.1). Concerning syntactic constructions, we can observe that they were quite varied (see distribution in Table 1). They can be attributive, as in the following example: (7) Il ne serait pas difficile de débusquer d’autres exemples d’analyses surprenantes ou contradictoires, … [linguistics-202-body] ‘It would not be difficult to track down other examples of surprising or contradictory analyses’

Predicative constructions were also quite productive (see (8)) and, very interestingly, we noticed a large proportion of impersonal constructions (il est Adj de ‘it is Adj to’ or il est Adj que ‘it is Adj that’) where the clause introduced can be considered as the source of surprise (e.g. (7)). This shows that these adjectives are often associated with facts or events, more than with simple objects. (8) Ce constat est d’autant plus surprenant que la société gabonaise valorise la solidarité familiale et parentale dans l’ensemble des interactions quotidiennes. [sociology-83-body] ‘This finding is all the more surprising given that the Gabonese society values family and parental solidarity throughout daily interactions.’ Table 1.  Syntactic distribution of surprise adjectives Syntactic function

Attributive

Predicative

Impersonal construction

Other

44.0%

32.0%

21.8%

2.2%



Surprise routines in scientific writing 161

(9) Quel que soit le type d’impact attribué aux évolutions démographiques, il est frappant de constater l’importance de ce facteur dans le débat allemand. [geography-348-body] ‘Regardless of the impact attributed to demographic changes, it is striking to note the importance of this factor in the German debate’

In one-fifth of occurrences, surprise adjectives were included in negative clauses. Almost half of the cases in impersonal clauses were found with que, as in the following example. (10) Il n’est donc pas surprenant que l’approche de Vygotsky ait une forte influence en Russie (en dépit du fait que le travail … [economics-275-body] ‘It is therefore not surprising that Vygotsky’s approach has a strong influence in Russia (despite the fact that work …’

In about 10% of cases, hedging could be observed with epistemic modal verbs such as sembler or pouvoir or with adverbs such as plutôt ‘rather’. This reinforces the impression that surprise is quite weak and that this feeling can be “negotiated” with the reader, as is often the case with hedging in scientific writing (see Hyland, 1998). Moreover, several adverbials related to the apparent, but not proven, character of surprise were observed (e.g. a priori, apparemment, à première vue, ‘at first sight, apparently, on first glance’) (see (11)), but this surprise did not resist a thorough examination of the facts. The authors went on to show that they were able to provide a logical explanation of these facts, as we will see below (4.2). (11) Un tel résultat pourrait a priori paraître extrêmement curieux: on s’attendrait à ce que les processus à longue mémoire permettent des prévisions bonnes à court terme …  [economics-36-body] ‘Such a result could seem extremely odd at first sight: one would expect that long memory processes would provide good short-term forecasts …’

Finally, we noticed several verbless sentences including surprise adjectives, as in the following examples: (12) Plus surprenant encore, les résultats engrangés au fil de ces multiples efforts de recherche sont substantiellement cohérents.  [educational science-482-body] ‘Even more surprising, the results achieved through these multiple research efforts are substantially consistent.’ (13) Fait curieux, qui conforte et éclaire notre propos: le texte n’a pas de co-texte … [linguistics-38-body] ‘An odd fact, which reinforces and clarifies our argument, is that the text has no co-text …’

162 Agnès Tutin

The use of verbless sentences in the semantic field of surprise has already been observed by Novakova and Sorba (2014) with the noun stupéfaction ‘amazement’. This increases the element of surprise, as the emotion is mentioned before the cause. We will see a similar effect below (4.2.2) with the use of pseudo-cleft sentences. On a semantic level, we analyzed the different kinds of sources related to surprise adjectives. In attributive and predicative constructions, different semantic types of nouns emerged: – Quality nouns such as aspect ‘aspect’, caractère ‘character’, contraste ‘contrast’ dealing with the properties of observed objects; – Results pertaining to observed objects: constat ‘statement’, élément ‘element’, exemple ‘example’, résultats ‘results’, phénomène ‘phenomenon’; – Logical relations: effet ‘effect’, conséquence, ‘consequence’, lien ‘link’; It should be highlighted here that no polar nouns, whether positive or negative, were found and that no trace of polarity could be found in the co-text of surprise adjectives. Many of these nouns have a large scope and can be considered as “signalling nouns” (Flowerdew, 2003), whose meaning is contextual with an endophoric function. In the following example, the word oubli (‘oversight’) summarizes a complex fact previously mentioned in the text. Moreover, many of these anaphoric nouns are introduced with the help of demonstrative determiners, a very common discursive device in scientific writing (Boch & Rinck, to appear). (14) Si l’on admet qu’un statut plus élevé réduit les coûts de production (toutes choses égales par ailleurs), Podolny devrait en toute rigueur appliquer la loi des coûts et conclure que la hausse du statut tend à faire baisser le prix de vente. Or, il ne le fait pas. Cet “oubli” est étonnant puisque dans l’enquête empirique de son article il applique bien cette loi … [sociology-552-body] ‘If we assume that a higher status reduces production costs (all else being equal), Podolny should theoretically apply the law of costs and conclude that higher status tends to lower the selling price. But he did not. This “oversight” is surprising since, in the empirical investigation of his article, he applies this law …’

We also noted several collocations and preferred associations between nouns and adjectives (although beyond the scope of this study) such as contraste frappant ‘striking contrast’, effet inattendu ‘unexpected effect’, and résultat surprenant ‘surprising result’. These collocations can be explained by the semantic properties of each adjective and contribute here to the routinized flavor of surprise markers. Sources in impersonal constructions are infinitive clauses and que-clauses. Infinitive clauses mainly include verbs with an evidential function (voir ‘see’,



Surprise routines in scientific writing 163

constater ‘note’, and observer ‘observe’ (see (11)), while que-clauses (often negative as mentioned above) introduce scientific facts. (15) Tout d’abord il est assez frappant de constater que les cadres évacuent largement leur fonction évaluative: la production d’une fiche d’évaluation et l’attribution d’une note …  [information science-547-body] ‘First of all, it is quite striking to note that executives largely set aside their evaluative function: producing a scorecard and a rating …’ (16) Dès lors, il n’ est guère étonnant que le “culte” d’Angela Davis s’étiole après 1973 [history-513-body] ‘Therefore, it is hardly surprising that the “cult” of Angela Davis waned after 1973’

In summary, surprise adjectives used in these scientific articles are generally stylistically neutral and moderately intense. They are source-oriented and the experiencers (writer and reader) are generally implicit. Surprise is often toned down with hedging (and is negated in 20% of cases), and adjectives are used to qualify scientific facts and qualities. Polar words are absent in both sources and the co-text. 3.2.2 Surprise verbs Occurrences of surprise verbs were far less numerous than surprise adjectives (74 vs 226), although verbal forms included passive past participles, e.g. je suis surpris ‘I am surprised’ or étonné ‘astonished’. All the verbs belong to Ruwet’s second class of verbs (Ruwet, 1994; Novakova, Goossens, & Grossmann, 2013), i.e. verbs which have the experiencer as the direct object and the cause as the grammatical subject. But with passive and pronominal alternations, these verbs can become experiencer-oriented. Three verbs (and their passive or pronominal alternations) were most present in our corpus: étonner/s’étonner/être étonné ‘amaze, astonish’ (30 occ.), surprendre/être surprise/se surprendre ‘surprise’ (27 occ.) and frapper/être frappé ‘strike’ (17 occ.). As is the case for adjectives, the most intensive and polar verbs (e.g. sidérer ‘stupefy’, méduser ‘baffle’, and estomaquer, ‘flabbergast’) were absent, but contrary to adjectives, half of the verbs were experience-oriented (36/74), as in the following example: (17) On pourrait donc être surpris que seulement deux familles du Brabant se soient établies en Flandre après 1450, alors qu’avant cette date … [history-184-body] ‘One might therefore be surprised that only two families of Brabant were located in Flanders after 1450, whereas before that date …’

164 Agnès Tutin

The other half is source-oriented, as in: (18) La première chose qui frappe, dans l’analyse de cette élection est la difficulté à retrouver les clivages habituels en politique…  [geography-550-body] ‘The first thing that strikes one in the analysis of this election, is the difficulty to find the usual political divisions …’

Among the source-oriented verbs, we found only four occurrences of the experiencer (nous ‘we’). A third of the verbs appeared with a negation, which was a greater proportion than for adjectives. Interestingly, though, similar to the occurrences with adjectives, affirmative clauses were often toned down with epistemic modals (on peut s’étonner ‘we may be surprised’) or adverbs (sans doute ’probably’). ‘Pure’ surprise without negation or hedging represented only half of the occurrences. We also noted (see also Novakova et al. (2013) in novels and newspapers), frequent pseudo-­cleft constructions, especially with frapper, which thus seems more typical of the semantic field than the genre: (19) Si une telle prodigalité demeure somme toute exceptionnelle et limitée essentiellement au XIIIe siècle, ce qui frappe plus encore est le souci des hommes d’approvisionner les indigents en victuailles, …  [history-63-body] ‘While such extravagance remains altogether exceptional and limited primarily to the thirteenth century, what is more striking is men’s desire to supply the needy in victuals, …’

This construction could be compared to verbless clauses with surprise adjectives, where it contributes first to placing emphasis on the expression of surprise and second, to mentioning the source of the surprise. Several other recurrent verbal expressions were observed such as on ne s’étonnera pas… ‘one will not be surprised that’… X ne laisse pas de surprendre… ‘X is nonetheless surprising’, and aussi ne serait-on pas surpris… ‘one will therefore not be surprised…’. These prefabricated expressions contribute to the formulaic style of mentions of surprise in scientific articles. If we now look at the semantic roles of surprise verbs, among experiencers (35/74 occ.), we find mainly occurrences with personal pronouns: 24 on ‘one’, 6 nous ‘we’ (4 are subject ‘we’, 2 are direct object ‘us’), only two je ‘I’ and two NP (l’observateur ‘the observer’, un regard distancié ‘a distanced view’). The use of these pronouns shows that experiencers are not highlighted (we will return to this point in 4.1). Regarding the semantic role of the source, contrary to other genres (e.g. dialogues in Celle and Lansari (2014), novels and newspapers (Novakova et al., 2013)), we did not encounter examples such as nous sommes surpris ‘we are

Surprise routines in scientific writing 165



surprised’ without any mention of the source. In the genre of scientific writing, sources are obligatory arguments while experiencers are optional arguments. Looking at the syntactic type of the source revealed a large number of clauses (21 occ./74 que-clauses, si-clauses, and infinitive clauses). Pronominal sources (ce, cela), which have an endophoric function, were more frequent than with adjectives (12 occ.). Nominal sources of surprise verbs were quite similar to what was observed in the case of adjectives: many of them could be considered as “signalling nouns”, but this feature was even more pronounced with surprise verbs, with frequent generic nouns such as chose ‘thing’ and fait ‘fact’. For surprise verbs, Novakova et al. (2013) noticed a duplication of syntactic actants of cause in some syntactic contexts such as pseudo-cleft sentences. In scientific writing, we often notice a duplication of discursive source arguments (as observed in example (19)): the signalling noun or the pronoun is the syntactic actant, but the real semantic actant has to be found in the larger context. In short, our observations regarding verbs confirmed the trends regarding adjectives. Verbal surprise markers are stylistically neutral, not very intense, and place more emphasis on sources than on experiencers – the former are obligatory, contrary to the latter. As is the case for adjectives, sources are facts or signalling nouns or pronouns related to facts. Since experiencers are discrete and hedging frequent, the expression of surprise is definitely not a sign of strong subjectivity in scientific writing. 4. Enunciative and discursive features Having examined the formal properties of surprise markers, we will now turn to the use of these elements in scientific texts. First of all, we will address the topic of the experiencer to identify who experiences surprise in scientific texts. Then we will look at some discursive properties of surprise in scientific writing. 4.1

Enunciative properties

Who exactly experiences surprise in scientific writing? We saw previously that experiencers are often implicit with surprise markers: they are completely absent with source-oriented adjectives, and only present in half the cases of surprise verbs. When present, they are represented using mostly on ‘one’ or nous ‘we’, and rarely je ‘I’. But nous and on are polyfunctional and can have several referential values in scientific writing (Fløttum et al., 2006; Tutin, 2010; Hartwell & Jacques, 2014). They can refer:

166 Agnès Tutin

a. to a single or plural author (e.g. nous pensons, ‘we think’…), b. to the author and the reader (e.g. comme nous pouvons le voir dans cette figure, ‘as we can see in this figure’), c. to the author and the scientific community (e.g. on considère souvent que la surprise n’est pas une émotion polaire… ‘one often considers that surprise is not a polar emotion…’), d. seldom with nous, more frequently with on, to a generic human being (e.g. en général, nous pensons tous/on pense tous que la surprise est une émotion passagère… ‘generally, we all feel that surprise is a brief emotion’). Close examination of on and nous (we shall leave aside je for which we have only two occurrences) shows that several referential values can be found with surprise verbs, although disambiguation is quite tricky and requires thorough examination of the textual context. In some examples, the experiencer refers strictly to the author. In this case, surprise is often related to the scientific narration and is a step in the scientific study. The tense here can be the passé composé as in the following example: (20) Nous l’avons dit plus haut, nous avons été frappés par le fait de nous retrouver devant quelque chose de très complexe  [educational science-574-body] ‘As we have said before, we were struck by the fact that we faced something very complex:

Far more frequently, nous et on refer to a larger community, including the author and the reader but also the scientific community. Surprise expressed by the author engages the reader as a witness and is related to the expectations of the discourse community. This is especially true when modals or hedging are used, such as in the following example: (21) On pourrait donc être surpris que seulement deux familles du Brabant se soient établies en Flandre après 1450, alors qu’avant cette date…[history-184-body] ‘One might therefore be surprised that only two families of Brabant were located in Flanders after 1450, whereas before that date …’

By including the reader and the discourse community in the observation of unexpected facts, the author establishes complicity with the reader: they share the same references and the same expectations. The author’s surprise is not personal: it is a shared feeling – or may simply be a statement of non-congruence with regard to expectations – which would be experienced by any observer in the same community. While scientific writing has an inherent interdiscursive dimension, the expression of surprise – like other types of markers (for example, statement markers with evidential functions comme nous pouvons le voir dans la figure 1 ‘as

Surprise routines in scientific writing 167



we can see on Figure 1’, cf. Grossmann & Tutin 2010a, 2010b) – is clearly an aspect of “interlocutive” dialogism. 4.2

Discursive properties of surprise in scientific writing

In this section, we will address two aspects of discursive properties: the thematic role of surprise arguments and the stereotypical discursive scenario of surprise in our corpus. 4.2.1 Thematic observations As previously discussed, surprise experiencers are very often shadows of more predominant sources. We wished to look more precisely at the thematic positions of the surprise-experiencer-source trio in our corpus by examining the first element occurring in discourse. Concerning adjectives, all source-oriented, we observed several interesting facts. While most occurrences seemed to follow the source-surprise order (un fait frappant ‘a striking fact’; ces résultats sont frappants ‘these results are striking’), in some cases the reverse order was also observed. Several surprise adjectives (frappant, étonnant, curieux, ‘striking’, ‘astonishing, ‘curious’) occurred in a prenominal position, which appears as a marked position compared to the more ordinary postnominal position in French: (22) Il est possible d’aller chercher dans les arcanes du droit administratif des éléments de compréhension de cette étonnante asymétrie  [political science-380-body] ‘It is possible to hunt through the labyrinths of administrative law for the elements necessary for understanding this surprising lack of symmetry’

Surprise was also mentioned before the source in some verbless sentences, as observed above (cf. examples (12) and (13)) or in impersonal constructions where infinitive or que-clauses were obviously located after the adjectives. With verb surprise markers, while in most cases we found sources or experiencers in the first position, we also noticed that a large number of experiencer-­ oriented verbs had an implicit grammatical subject, as in the following example: (23) Comment ne pas être frappé par la diversité et l’irréductibilité des sens octroyés à cette notion dont, au demeurant, l’usage … [educational science-383-body] ‘How is it possible not to be struck by the diversity and irreducibility of meanings ascribed to this notion, the use of which, incidentally, …’

As with pseudo-cleft sentences (see example (19)), this reinforces the impression that the emphasis is placed on the surprise predicate rather than on the experiencer.

168 Agnès Tutin

To sum up, in our corpus, the observation of thematic position shows that emphasis is placed on sources. Interestingly, though, several marked constructions seem to be used to emphasize the surprise predicate, in order to create a dynamic scientific narrative. 4.2.2 Scenario of surprise If we now look at the discursive scenarios involving surprise in our corpus, some regularities can be identified. When attempting to analyze the argumentative construction of this emotion (see Micheli, 2010), several topics can be highlighted, as observed by Novakova and Sorba (2013) with the verb stupéfier ‘amaze’ in the journalistic genre. In scientific writing, surprise clearly occurs in the framework of a scientific prediction model, although this model is more frequently implicit than explicit. The surprise scenario (or the absence of surprise) is thus generally associated with a number of discursive sequences, some of which are optionally realized. The chronological order of these sequences of the surprise scenario does not necessarily correspond to the sequential order in the text, as described below: 1. Sequence 1: Scientific expectations As stated earlier in Section 1, expectations can be expressed explicitly, as in on s’attendrait à ce que les processus à longue mémoire permettent des prévisions bonnes à court terme … (‘one would expect that long memory processes would provide good short-term forecasts…’). Given the size of the corpus, we could not examine all the occurrences exhaustively, but implicit expectations, be they common sense or shared knowledge of the discourse community, seem far more frequent in this genre than explicit expectations. 2. Sequence 2: Observations of facts The facts observed constitute the source of the predicate of surprise, in its widest sense. As mentioned earlier, the source role immediately associated with the surprise predicate is often an endophoric element or a signalling noun whose antecedent has to be found in the textual environment. The fact is generally associated with an observation or a statement. 3. Sequence 3: Congruence or non-congruence with expectations Surprise (or the absence of surprise) is related to the extent of observed facts being congruent with expectations.

Surprise routines in scientific writing 169



4. Sequence 4: Explanation of surprise Scenarios of emotions should also take into account post-emotion effects (Plantin, 2011). We observed that surprise was often toned down in scientific writing or considered as apparent (e.g. surprenant à première vue ‘surprising at first sight’). In scientific writing, surprise is often followed by a specific sequence of explanation: surprising facts can actually be explained (and thus do not remain that surprising). Either the prediction model has to be revised or the facts are actually more complex than assumed, as in the following example: (24) BENSAÏD et de PALMA [1995] concluent à une FACT transmission supérieure à «un pour un» de la politique monétaire Ce résultat surprenant tient, SOURCE and SURPRISE pour une part, au fait que le coût marginal des EXPLANATION ressources est, dans leur article, assimilé au taux de la politique monétaire. ‘BENSAÏD and Palma [1995] conclude that there is a transmission greater than “one for one” of monetary policy. This surprising result is due, in part, to the fact that the marginal cost of funds is, in their article, likened to the rate of monetary policy….’

This simplified scenario remains representative of those found in our corpus, although the rhetorical strategies of surprise need to be studied further. 5. Conclusion While surprise is fairly present in the lexicon and discourse strategies of research articles of French social sciences, they do not have a very emotional flavor in this genre, for several reasons: – Surprise is more source-oriented than experiencer-oriented. The human dimension is set aside in favor of the objective dimension. In other contexts, such as literature or dialogue, the focus is the experience of surprise by a person. In scientific writing, the objects of surprise, in this case, scientific facts and scientific objects, take center stage, while experiencers are either absent or diluted in the discourse community.

170 Agnès Tutin

– The lexical environment is not polar. While surprise in itself is not good or bad, the context is often polar in literature or newspapers; in scientific writing, there is simply surprise, corresponding to the non-congruence of predicted facts. This seems to be a specificity of this genre compared to other ones (Kern & Grutchus, 2014; Novakova & Sorba, 2014). – Surprise is not a strong feeling, if it is one. Surprise is often apparent (or attenuated) and is almost always accompanied by an explanation. It is used mainly to establish complicity and to strengthen the scientific narrative, as highlighted by Boch et al. (2007), as evidence or congruence markers are also used. Some stylistic effects such as verbless clauses or pseudocleft sentences seem to further support this analysis. – Contrary to what is commonly accepted as a specific property for this feeling, surprise is not in this genre a passing affect. It is more a routinized process associated with a discursive scenario in the sociolect of the scientific community. Finally, we question the status of surprise as an affect in scientific writing. It seems to be more a state of consciousness associated with the observation of complex facts expressed in stereotypical rhetorical scenarios. In any case, it appears to be a complex state, with rich conceptual content. Finally, these findings confirm that while a kind of subjectivity may be present in scientific writings through stance or opinion markers, the affective component is definitely not a part of them. Furthermore, they reinforce the particular status of surprise in the field of affects/ emotions.

References Augustyn, M., Ben Hamou, S., Bloquet, G., Goossens, V., Loiseau, M., & Rinck, F. (2008). Constitution de ressources pédagogiques numériques: Le lexique des affects. In M. Loiseau, M. Abouzaïd, L. Buson, C. Cavalla, A. Djauron, C. Dugua, et al. (Eds.), Autour de la langue et du langage: Perspective pluridisciplinaire (pp. 407–414). Grenoble: Presses Universitaires de Grenoble. Boch, F., Grossmann, F., & Rinck, F. (2007). Conformément à nos attentes: Les marqueurs de convergence/divergence dans l’article de linguistique. Revue Française de Linguistique Appliquée, 2(12), 109–122. URL : www.cairn.info/revue-francaise-de-linguistique-appliquee2007-2-page-109.htm. Boch, F., & Rinck, F. (2016). Anaphores démonstratives dans les écrits d’étudiants de Master. Comparaison avec les pratiques expertes. Linx, 72, 111–130. Caffi, C., & Janney, R. W. (1994). Toward a pragmatics of emotive communication. Journal of Pragmatics, 22(3), 325–373.  doi: 10.1016/0378-2166(94)90115-5



Surprise routines in scientific writing 171

Celle, A., & Lansari, L. (2014). ‘I’m surprised’ / ‘Are you surprised?’ Surprise as an argumentation tool in verbal interaction. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discours / Emotions in discourse (pp. 267–277). Bern: Peter Lang. Fløttum, K., Dahl, T., & Kinn, T. (2006). Academic voices across languages and disciplines. Amsterdam: John Benjamins.   doi: 10.1075/pbns.148 Flowerdew, J. (2003). Signalling nouns in discourse. English for Specific Purposes, 22(4), 329– 346.  doi: 10.1016/S0889-4906(02)00017-0 Grossmann, F., & Tutin, A. (2010a). Evidential Markers in French Scientific Writing: the Case of the French Verb voir. In E. Smirnova & G. Diewald (Eds.), Evidentiality in European languages (pp. 167–197). Berlin/New York: Mouton de Gruyter. Grossmann, F., & Tutin, A. (2010b). Les marqueurs verbaux de constat: Un lieu de dialogisme dans l’écrit scientifique. Actes du colloque Dialogisme: langue, discours 2010, 8–10 septembre 2010. Montpellier, Praxiling. http://www.univ-montp3.fr/praxiling/IMG/pdf_GrossmannTutin.pdf. Grossmann, F., Tutin, A., & Garcia Da Silva, P. (2009). Filiation et transfert d’objets scientifiques dans les écrits de recherche. Pratiques, 143/144, 187–202.  doi: 10.4000/pratiques.1447 Grossmann, F., & Wirth, F. (2007). Marking evidentiality in scientific papers: The case of expectation markers. In K. Fløttum (Ed.), Language and Discipline Perspectives on Academic Discourse (pp. 202–218). Newcastle: Cambridge Scholars. Hartwell, L., & Jacques, M.-P. (2014). Authorial presence in French and English: ‘Pronoun + verb’ patterns in biology and medicine research articles. Discours, 15. Hyland, K. (1998). Hedging in scientific research articles. Amtserdam/Philadelphia: John Benjamins.   doi: 10.1075/pbns.54 Hyland, K. (2002). Authority and invisibility: authorial identity in academic writing. Journal of Pragmatics, 34, 1091–1112.  doi: 10.1016/S0378-2166(02)00035-8 Kern, B., & Grutchus, A. (2014). Surprise vs étonnement: Comportement discursif et perspectives contrastives. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discours / Emotions in discourse (pp. 187–198). Bern: Peter Lang. Kövecses, Z. (2003). Metaphor and emotion: Language, culture, and body in human feeling. Cambridge, MA: Cambridge University Press. Micheli, R. (2010). L’émotion argumentée: L’abolition de la peine de mort dans le débat parlementaire français. Paris: Éd. du Cerf. Novakova, I., Goossens, V., & Grossmann, F. (2013). Interactions entre profil discursif et structures actancielles: l’exemple des verbes de surprise et de respect. Langue Française, 4, 31–46.  doi: 10.3917/lf.180.0031 Novakova, I., & Sorba, J. (2013). Stupéfier et jalouser dans les séquences textuelles journalistiques: Quel profil discursif pour quelle stratégie argumentative? Le discours et la langue, 4(1), 203–220. Novakova, I., & Sorba, J. (2014). L’émotion dans le discours. À la recherche du profil discursif de stupeur et de jalousie. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discours / Emotions in discourse (pp. 161–173). Bern: Peter Lang. Ortony, A., Clore, G., & Collin, A. (1999). The cognitive structure of emotions. Cambridge: Cambridge University Press. Piolat, A., & Bannour, R. (2009). EMOTAIX: Un scénario de tropes pour l’identification automatisée du lexique émotionnel et affectif. Année psychologique, 109(4), 655–698. doi: 10.4074/S0003503309004047

172 Agnès Tutin

Plantin C. (2011). Les bonnes raisons des émotions: Principes et méthode pour l’étude du discours émotionné. Bern: Lang. Ruwet, N. (1994). Être ou ne pas être un verbe de sentiment. Langue française, 103, 45–55. doi: 10.3406/lfr.1994.5725 Swales, J., & Burke, A. (2003). “It’s really fascinating work”: Differences in evaluative adjectives across academic registers. In P. Leistyna & C. F. Meyer (Eds.), Corpus analysis: Language structure and language use (pp. 1–18). Amsterdam: Rodopi. Tutin, A. (2010). Dans cet article, nous souhaitons montrer que…: Lexique verbal et positionnement de l’auteur dans les articles en sciences humaines. Lidil. Revue de linguistique et de didactique des langues, 41, 15–40. Tutin, A., & Grossmann, F. (Eds). (2014). L’écrit scientifique: Du lexique au discours. Rennes: Presses universitaires de Rennes.

Appendix List of surprise markers Adjectives ahuri, abasourdi, abasourdissant, ahuri, ahurissant, baba, bizarre, cloué, coi, confondant, confondu, curieux, déconcertant, déconcerté, ébahi, ébaubi, éberlué, ébouriffant, ébouriffé, épatant, épaté, époustouflant, époustouflé, estomaqué, étonnant, étonné, étrange, foudroyant, frappant, frappé, inattendu, incroyable, interdit, interloqué, invraisemblable, médusé, merveilleux, pantois, prodigieux, renversant, renversé, saisissant, sidérant, sidéré, soufflé, stupéfait, stupéfiant, suffocant, suffoqué, surprenant, surpris Nouns abasourdissement, ahurissement, ébahissement, éberluement, ébouriffement, effarement, épatement, étonnement, saisissement, sidération, soubresaut, stupéfaction, stupeur, surprise, sursaut, tressaillement, tressautement Verbs ahurir, abasourdir, ébahir, bluffer, époustoufler, étonner, interloquer, clouer, sidérer, surprendre, sursauter, tressaillir, éberluer, tressauter, ébouriffer, épater, estomaquer, frapper, méduser, renverser, scier, sécher, souffler, stupéfier Adverbials de façon surprenante, de façon étonnante, de manière surprenante, de manière étonnante, incroyablement, imprévisiblement, étonamment, contre toute attente

Surprise in the GRID* Cristina Soriano,a Johnny R. J. Fontaineb and Klaus R. Scherera a

b

Swiss Center for Affective Sciences, University of Geneva / Ghent University

This paper analyses surprise in the framework of the GRID paradigm as part of a research project on the meaning of emotion words across languages and cultures. Based on psychological component theories of emotion, an online instrument was designed to rate the meaning of 24 emotion terms on 142 features of emotion. Data was collected for 23 languages in 27 countries. The mean rates within and across languages offer a semantic profile of the emotion terms. Results are presented on the meaning of “surprise” across languages and in English and French. The data also indicate four dimensions are necessary to describe the emotion space: valence, power, arousal and novelty. Novelty captures variation in suddenness and expectedness and differentiates surprise from other emotions, as well as types of emotion within a family, revealing itself as a relevant aspect of emotion universally encoded in our affective lexicons. Keywords: surprise, GRID, cross-cultural semantic profile, English, French, novelty dimension

1. Introduction Emotions are adaptive mechanisms for a more efficient interaction with the environment. They constitute congruent patterns of cognitive appraisals, physiological changes, and behaviors that allow the organism to detect relevant contextual cues (for survival or socialization) and respond in a successful way. The most important and recurrent affective experiences of a community tend to become * This work was supported by the ERC Advanced Grant 230331-PROPEREMO, awarded in the European Community’s 7th Framework Program to Klaus Scherer, and by the Swiss Center for Affective Sciences, financed by the Swiss National Science Foundation (51NF40-104897) and hosted by the University of Geneva. doi 10.1075/bct.92.07sor © 2017 John Benjamins Publishing Company

174 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

lexicalized in their language for the members to be able to refer to them. In the realm of personality psychology, this is known as the sedimentation hypothesis (e.g., Cattell, 1943, 1945; Saucier & Goldberg, 1996). Applied to emotion, it means that the meaning of emotion words can be understood as the “sedimented” average of all lived or witnessed experiences that were labeled in that way. In other words, the meaning of an emotion term like anger, joy or fear is a condensed and prototypical (in Rosch’s sense – e.g., Rosch, 1975; Rosch & Mervis, 1975) account of that kind of experience. This is why the study of language, and of the emotion lexicon in particular, can provide us with valuable information about human emotionality. In this paper we present a research paradigm designed to investigate the meaning of emotion words across languages and cultures. It involves a questionnaire with which native speakers of different languages report on the meaning of emotion words by rating them on a number of features using a numerical scale. It is called the “GRID study” because the affective semantic space is captured by “a grid” of columns and rows, where the rows are occupied by emotion words, the columns by emotion features, and the cells by the rating given by each participant to each feature for each word to indicate the pertinence of each feature for the emotion designated by the respective word. The GRID is useful to investigate the representation of emotion concepts in different languages as well as cross-culturally. One particularly interesting domain of application is the investigation of surprise, where the GRID can shed some light on two debated issues: whether surprise is an emotion, and (if so) whether it is a universal one. The status of surprise as an emotion has been questioned on at least three grounds. First, most emotions tend to have what psychology refers to as “valence”: in other words, they are typically either positive or negative. For some (e.g., Barrett, 2006; Barrett, Mesquita, Ochsner & Gross, 2007; Ortony, Clore & Collins 1988; Russell, 2003), this is an essential or definitional feature of emotions. The valence of surprise appears to be neutral, though, because the eliciting events could be either positive or negative, and the experience of surprise in itself can be intrinsically pleasant or unpleasant (e.g., Reisenzein & Meyer, 2009; but see Noordewier & Breugelmans, 2013, for an account of surprise as negative). A second reason to question the nature of surprise as an emotion is its short duration, shorter than other experiences one may want to call “emotion”. Similarly, the word “surprise” cannot refer to a dispositional state or to a mood, although this is possible for many other emotions (“she is angry” vs “she is an angry person”). Finally, a third reason to question whether surprise is an emotion is its resemblance to cognitive operations (not necessarily emotional) like novelty detection and attention orientation that should in principle apply to all emotions. If surprise



Surprise in the GRID 175

is best understood as a component of emotions, then it cannot be an emotion itself. However, none of these arguments is sufficient to disqualify surprise as an example of the category emotion. To begin with, valence may not be an essential feature in the definition of emotion, but even if it were, the apparent neutral valence of surprise would not be a reason to reject surprise as an emotion. Saying that “surprise has neutral valence” can be taken to mean two things. It may mean that the experiences of surprise have neither a positive, nor a negative feel to them. Alternatively the expression may be used to mean that the concept of surprise has neither a positive, nor a negative value. In the first case, “neutral valence” – if it actually happens – would not be a problem from a theoretical point of view, because ‘neutral’ is a value on the valence dimension occupying an equidistant point in the continuum between good and bad. As compellingly argued by Noordewier and Breugelmans (2013, p. 1327), “if valence is seen as a dimensional property on which emotions take a certain position there is no logical reason why a neutral position would be problematic”. In the second case, the neutral valence of the surprise concept is not a challenge to the status of surprise as an emotion either, because it could simply reflect that, in experience, the valence of surprise events is open-ended. In other words, the concept as a whole could be neutral if the actual experiences of surprise have as many chances of being positive as they do of being negative in real life. The second reason to question the nature of surprise as an emotion was its comparative short duration with respect to other emotions. However, this cannot be by itself a disqualifying argument because duration is a gradual feature and there are no fixed boundaries to demarcate the extension of an emotional episode. At most, short duration would make surprise a less prototypical kind of emotion as far as timing is concerned. A final argument against the status of surprise as an emotion was its resemblance to cognitive operations. However, although related, surprise is not the same thing as novelty detection. Novelty detection is a basic cognitive operation, automatic and unconscious, consisting in the detection of new information while we routinely scan the environment (Brosch, 2009). It is necessary for the emergence of emotion, but it does not always lead to emotional episodes. The new information detected may also be unexpected if it clashes with the cognitive schemas that we use to represent the world (e.g., Meyer, Reisenzein & Schützwohl, 1997). New and unexpected events with an abrupt onset and relevance to our goals typically lead to the full “syndrome” of experiences (physiological, expressive, motivational, etc.) that we commonly refer to as a surprise episode (Reisenzein, 2000; cf. e.g., Scherer, 1984, and Frijda, 1987, for componential views of emotion). In sum, surprise is more than mere novelty or unexpectedness detection and they should not be equated or confused.

176 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

These and similar arguments ultimately suggest that the question as to whether or not surprise qualifies as an emotion depends on one’s definition of emotion, for which there is no consensus in the affective sciences. Following one of the most widespread approaches in psychology, the componential view (e.g., Averill, 1980; Lazarus, 1991; Reisenzein, 2000; Scherer, 1984), and more specifically Scherer’s Component Process Model (Scherer, 2009), in the GRID we define emotion as an episodic process elicited by goal-relevant events that consists in the patterned synchronization of the five basic human subsystems, also called emotion components (i.e. appraisal, bodily reaction, expression, action tendencies, and feeling) in order to prepare the person for adaptive action (e.g., Scherer, 2005). Appraisal refers to the unconscious or conscious evaluations we routinely carry out about the events we perceive in the environment (e.g., is this expected, is it good, was it intentional, can I cope with it, etc.). Bodily reaction refers to the physiological changes or symptoms of the emotion (e.g., increased heart rate, sweat, increased body temperature, etc.). Expressions have to do with the way the emotion is manifested in the face (e.g., smile, tears, wide-open eyes, etc.), voice (e.g., speaking louder, speaking faster, going silent, etc.) and gestures (e.g., abrupt movements). Action tendencies refer to the way one feels compelled to behave (e.g., wanting to do damage, wanting to run away, wanting to sing and dance, etc.). Finally, feeling is the emerging sensation one can consciously perceive (e.g., feeling good, feeling strong, feeling calm, etc.). Under this account, emotions are characterized by distinct componential profiles. In this sense, surprise would be an emotion, because it seems to involve the aforementioned components: it is triggered by specific appraisals like novelty and unexpectedness (e.g., Meyer, Reisenzein & Schützwohl, 1997), it is accompanied by physiological changes like sweating (Sokolov, Spinks, Näätänen & Lyytinen, 2002, in Reisenzein & Meyer, 2009, p. 386), it involves an action tendency to stop what one is doing to pay attention to the new stimulus (Meyer, Niepel, Rudolph & Schützwohl, 1991), and it is associated to expressive behaviors like eyebrow rising, arguably to facilitate seeing (Darwin, 1872). Not all surprise experiences may display the full componential profile (cf. Reisenzein, 2000), but the point at stake in determining whether surprise is an emotion is whether the experiences respond to this profile on average. The GRID study is valuable in this sense because, according to the sedimentation hypothesis, if the average surprise experience involves a number of systematic changes in all components, the meaning of the word “surprise” should also exhibit a systematic componential profile. In addition to its status as an emotion, a second aspect frequently debated concerning surprise is its cross-­cultural stability. Is this a universal emotion? Surprise is often included in the list of so called basic emotions (e.g., Ekman, Friesen & Ellsworth, 1982; Izard, 1977; Plutchik, 1980), claimed to be universal.



Surprise in the GRID 177

Basic emotions are also supposed to have universally recognizable facial expressions (Ekman, 1989, 1992; Ekman & Friesen, 1971; Ekman, Sorenson & Friesen, 1969; Izard, 1971). The question is, thus, whether the prototypical “surprise face” (raised eyebrows, wide-open eyes, open mouth) is recognized in all languages as “surprise”. At a practical level, this empirical question largely depends on whether a word similar to the English surprise can be found in those languages. Earlier studies have only established that the indigenous words can refer to the same facial expression, but it is uncertain whether the words actually mean the same.1 The distinction is important, because we can only speak of universality in the second case. If sufficient overlap exists between the meanings of the “surprise” terms, it would constitute compelling evidence in favor of the universality of surprise as an emotional experience. In addition to these two questions, the GRID can contribute to a hotly debated issue in psychology: whether emotions can be described in terms of a small set of universally valid dimensions, and if so, which dimensions those are. Most psycholinguistic research on emotion words and emotional experiences is rooted in this approach to emotion conceptualization. The number and type of dimensions hypothesized to organize the emotion space has varied considerably throughout time, with current theories still proposing a number of different solutions (for a review see Fontaine, 2013). Two main trends can be found: two-­dimensional models and three-dimensional models. The two-dimensional models rely on the dimensions of valence (Watson & Tellegen, 1985), arousal (Thayer, 1989), or both of them (Larsen & Diener, 1992; Russell, 1980) – the latter being the most widespread solution. Valence, also referred to as pleasantness, evaluation, hedonic tone, pleasure, and (in other disciplines) axiology or polarity, is the dimension that opposes good/positive to bad/negative and, therefore, differentiates emotions like joy and love from others like sadness and fear. Arousal, also known as activation, is the dimension that opposes aroused/excited/active to relaxed/calm/ inactive; on this dimension, emotions like anger and fear are opposed to emotions like sadness and contentment. The three-dimensional models include these same two dimensions plus a third one: power. Power, also referred to as potency and dominance, is the dimension opposing strong/powerful to weak/powerless; emotions like anger and pride are opposed to emotions like shame and sadness on this dimension. The three-dimensional model was first developed in psycholinguistic studies of semantic connotation (Osgood, May & Miron, 1975; Osgood, Suci & 1. Two words can refer to the same thing without meaning the same. For example, the Spanish word “hermana” means “sister”, while the English word “sibling” means “brother or sister”. The meanings are different, but both words could be used to successfully identify the same person in the world.

178 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

Tannenbaum, 1957), but has also been identified in the study of emotion words using similarity sorting and similarity rating tasks (e.g., Fontaine, Poortinga, Setiadi & Markam, 2002; Shaver, Schwartz, Kirson & O’Connor, 1987; Shaver, Wu & Schwartz, 1992). The latter methods have provided two-dimensional solutions as well (e.g., Gehm & Scherer, 1988; Russell, Lewicka & Niit, 1989), but the output in those case is likely to be due to scarcity of stimuli and/or data, since similarity studies working with a varied enough sample of emotion terms and a large sample of participants recovered three dimensions, not two (Fontaine, 2013, p. 37). In addition to these two types of models, our more recent work on the denotative meaning of emotion words across languages and cultures using the GRID paradigm (Fontaine, Scherer, Roesch & Ellsworth, 2007; Fontaine, Scherer & Soriano, 2013a) has identified a fourth dimension, novelty, related to variation in terms of suddenness and expectedness. On this dimension, emotions like surprise, and to a lesser extent fear, are differentiated from emotions like guilt and pride. The fourth dimension in the affective semantic space suggests that novelty is also relevant in the representation of actual affective experiences (for psycholinguistic evidence on this respect see Fontaine, Veirman & Groenvynck, 2013). In what follows we will first present the GRID paradigm in some more detail (Section 2). Then, the meaning of “surprise” across languages will be analyzed to determine whether an average cross-cultural surprise concept can be identified across all samples (Section 3). The feature-based and dimensional profile of this cross-­cultural prototype will be described next (Section 4). We will then illustrate the capacity of the GRID to capture differences (and not only similarities) by zooming in on two languages, English and French, and describing the differences in meaning between their respective “surprise” words (Section 5). The relevance of the novelty dimension for the characterization of surprise as well as other emotions will be discussed in the following section (Section 6). We will conclude with an overview of our findings, lessons learned, and a proposal on how to address the remaining open questions. 2. The GRID project The GRID project (http://www.affective-sciences.org/grid) is an international interdisciplinary research collaboration to investigate the meaning of emotion words as perceived by speakers in different languages and cultures. The GRID is interdisciplinary in two main ways. First, it brings together a research goal squarely rooted in lexical semantics with a theoretical framework stemming from emotion psychology. More specifically, componential theories of emotion are used to operationalize the features against which to evaluate the meaning of emotion

Surprise in the GRID 179



words. This disciplinary collaboration is possible because emotion concepts are assumed to overlap (if not completely, at least largely) with the decontextualized meaning of the lexical units that name them. Secondly, the GRID is interdisciplinary in committing to a methodology common to both domains: self-report. Self-report and surveys constitute an inescapable methodology for much psychological research, especially when the measurement of external objective variables is not possible. In linguistics, self-­ report (more commonly referred to as introspection) is frowned upon if it means taking the intuitions of a single researcher as valid evidence about the features of a language. However, self-report remains a valid source of evidence when it is many native speakers, rather than only one, who introspectively reflect about their language by means of a survey. This approach is routinely used in categorical acceptability judgments and can also be used for gradual responses provided that an appropriately large number of informants is used, as was done in the GRID. The GRID instrument contains 142 questions about possible emotion features pertaining to the 5 components of emotion (appraisal, bodily changes, expression, action tendencies and feeling), as well as questions on regulation strategies. Moreover, two other general questions concerning the frequency and social acceptability of the emotional experiences are included (a copy of the instrument is available in Fontaine, Scherer & Soriano, 2013a, Appendix 2). The study has been conducted in 23 languages and 27 countries, composing a cross-cultural database of 34 data samples (cf. Soriano, Fontaine, Scherer et al., 2013 for additional details). In some cases several languages were collected in one country (e.g., Basque and Spanish in Spain), several countries were sampled for a single language (e.g., English in the USA, UK and Singapore), or several regions were sampled within the same country (e.g., Italian in the North and South of Italy). The entire database included 4948 participants (1434 males and 3514 females), with a mean age of 23.29 (mean SD 5.31). Twenty four words were investigated (Table 1), selected to cover the main areas in affective space (Fontaine, Scherer & Soriano, 2013b, pp. 88–89). The original version of the questionnaire and its emotion words was in English and they were translated into each of the additional languages by the respective Table 1.  Emotion words investigated in the GRID project (English version) Anger Anxiety Being hurt Compassion

Contempt Contentment Despair Disappointment

Disgust Fear Guilt Happiness

Hate Interest Irritation Jealousy

Joy Love Pleasure Pride

Note: A translation of these words into the remaining GRID languages can be found at http://www.affective-sciences.org/words

Sadness Shame Stress Surprise

180 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

collaborators through a process of translation and back-translation (e.g., van de Vijver & Leung, 1997). An average of twenty-five participants responded to the questionnaire for each of the emotion words in each language, evaluating how likely each of the features was on a scale from 1 (extremely unlikely) to 9 (extremely likely) (with the middle point of the scale, 5, meaning neither likely, nor unlikely). For each emotion word in each sample the data consisted of a matrix with N rows (where N corresponds to the number of participants that evaluated the meaning of that emotion word) and 142 columns representing the 142 features rated. Two more features, pertaining to frequency and social acceptability, were not included in the analyses because they did not describe the emotion process itself. The mean score obtained by each term for each feature constitutes the respective word’s feature-based semantic profile. As predicted by componential theories of emotion and the sedimentation hypothesis, componential profiles were successfully identified in the overall analysis of the emotion vocabularies of over 20 languages and cultures (Fontaine & Scherer, 2013), indicating that the meaning of emotion words is indeed componential, and lending support to the claim that emotional experiences are componential as well. Principal Component Analysis (PCA) was also conducted on the cross-­ cultural database to identify the underlying dimensional structure of the affective space covered by the GRID words (see Fontaine, Scherer & Soriano 2013b, pp.  94–97). For these analyses no assumption was made about the translation equivalence of the 24 emotion terms, and each term in each sample was treated as a separate observation (i.e. the matrix included 816 rows representing the 24 emotion terms in the 34 samples). The PCA revealed four underlying factors, namely (in order of importance): Valence, Power, Arousal and Novelty2 (Figure  1). In the GRID, features from all components contribute to the emergence of the four dimensions. For example, on the positive pole of Valence we find appraisals (e.g., the eliciting event is pleasant), bodily reactions (e.g., relaxed muscles), expressions (e.g., smiling), action tendencies (e.g., wanting to sing and dance) and feelings (e.g., feeling good). All components are represented on the negative pole of Valence too, with appraisals (e.g., the eliciting event has negative consequences), bodily changes (e.g., stomach troubles), expression (e.g., frowning), action tendencies (e.g., wanting to break contact with others) and feelings (e.g., feeling negative). The overall scores of a term on the four dimensions constitute the word’s dimensional semantic profile, which informs us of its nature on the continuum between positive and negative, powerful and weak, aroused and relaxed, as well as unexpected vs expected. 2. The name of the dimensions identified by the GRID is capitalized to distinguish them from dimensions obtained through other methods.

Surprise in the GRID 181



Figure 1.  Plot of the average factor scores for the 24 emotion words on Valence and Power (a), Valence and Arousal (b) and Valence and Novelty (c) 2 (a) anger

1.5

hate

irritation

contempt

1

pride

interest

jealousy 0.5

POWER

0

happiness

stress –0.5

being hurt disappointment

anxiety despair fear

–1 –1.5 –2

–1.5

–1

compassion

love

guilt

shame

sadness

–2

joy

contentment pleasure surprise

disgust

–0.5

0

0.5

1

1.5

2

VALENCE 1.5 (b)

anxiety 1

irritation

0.5

surprise guilt

jealousy hate

0

love

stress shame

anger

AROUSAL

fear

happiness

joy pleasure

despair disgust

–0.5

–1

–2

compassion

contempt disappointment sadness

–1.5

–2

–1.5

–1

–0.5

pride

interest

being hurt

0 VALENCE

0.5

1

contentment

1.5

2

182 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

4 (c)

surprise

3

NOVELTY

2

fear

1

disgust disappointment being hurt despair sadness anger anxiety irritation

0

contempt hate

joy happiness interest

pleasure

compassion

pride

jealousy

-1

contentment

stress shame love

guilt -2

-2

-1.5

-1

-0.5

0 VALENCE

0.5

1

1.5

2

Note: Reproduced from Fontaine, J. R. J. & Scherer, K. R., The global meaning structure of the emotion domain: Investigating the complementarity of multiple perspectives on meaning. In J. R. J. Fontaine, K. R. Scherer & C. Soriano (Eds.), Components of emotional meaning: A sourcebook, p. 115–116, Figures 7.1–7.3 © 2013. Reproduced by permission of Oxford University Press [http://ukcatalogue.oup. com/product/9780199592746.do]

3. A cross-cultural surprise concept? In the previous section we saw how the GRID revealed a componential as well as dimensional structure in the emotion vocabulary of over 20 languages and cultures around the world. In this section we will focus on one of those emotion concepts, surprise. As suggested in the introduction, if surprise is an emotion, and a universal one, the experience should exhibit a stable profile of emotion component features across cultures. The way we investigate this stability in the GRID is by looking at language as “fossilized experience” and investigating the semantic profile of the word “surprise” across languages in search for a stable pattern. The specific “surprise” lexemes for which data was collected in the framework of the GRID study are represented in Table 2.

Surprise in the GRID 183



Table 2.  “Surprise” lexemes in the GRID languages Language

Term

Afrikaans Arabic Burmese Basque Chinese (Taiwan) Chinese (Beijing) Czech Ducth English Estonian Finnish French German Greek Hebrew Italian Japanese Polish Romansh Russian Slovak Spanish Turkish Ukrainian

Verrassing ‫المفاجأة‬ Harridura 驚訝 惊奇 překvapení Verrassing Surprise Üllatus Yllättyminen Surprise Überraschung Έκπληξη ‫הפתעה‬ Sorpresa 驚き Zdziwienie surpresa Удивление prekvapenie sorpresa Şaşkınlık Подив

The indigenous surprise concepts were highly homogeneous, with an average interrater rating reliability (Cronbach’s alpha) of .87 (see Soriano, Fontaine, Scherer et al., 2013 for more details). This means that a group of (on average) 24 native speakers in each language had no difficulty in describing the concept behind their “surprise” word coherently with one another. In order to measure more precisely the degree of similarity of the “surprise” terms across languages, we computed an average surprise profile across all 34 samples and computed the Pearson correlations between this average profile and the 34 sample specific profiles. These profile correlations were high overall, ranging from .59 to .95 with an average of .88, and were all significant at .001 level. In 18 samples the profile correlation was higher than .90, in 10 samples the profile correlation was between .80 and .90, in four samples it was between .70 and .80. In one sample, Ukraine, the profile correlation was almost .70 (r = .69). Only in one sample, the Burmese, the profile correlation was as low as .59. These results

184 Cristina Soriano, Johnny R. J. Fontaine and Klaus R. Scherer

indicate that, with the possible exception of Burmese, all indigenous “surprise” concepts were very similar to the mean cross-cultural profile. In addition, the cross-cultural surprise profile was distinct enough to avoid confusion with any other emotion concept. This was attested by two multiple discriminant analyses (MDA), one on the basis of the 142 emotion features and another on the basis of the four overall factors. Surprise could be correctly identified out of 24 terms based on both its dimensional profile and its featural profile in 100% of the cases, making it the most distinct emotion concept (highest classification accuracy) in the whole database (Fontaine & Scherer, 2013, p. 121). These and the previous results attest to the existence of a general “surprise” concept shared by all these languages. In the following section we will investigate its most characteristic features. 4. The meaning of surprise across languages The meaning of the general, cross-culturally valid surprise concept (henceforth surprise) can be described at two levels: dimensional and featural. In other words, one can calculate the cross-cultural average score of surprise on the four affective dimensions (dimensional profile), and one can look at the features that cross-culturally best differentiate surprise from the remaining emotion concepts (feature-based semantic profile).

The surprise dimensional semantic profile Valence, Power, Arousal and Novelty are not directly measured in the GRID as a priori semantic variables. Instead, they emerge empirically from the semantic profile of the words as meaning dimensions or factors underlying the semantic space. Since the factor scores are not direct measures of Valence, Arousal, Power and Novelty, a good way to interpret whether an emotion concept is, for example, more or less positive or negative, is to look at the size of the factor score. Factor scores are standardized scores, with a mean of 0 and a standard deviation of 1. In the GRID structure, a factor score thus indicates how many standard deviations an emotion term deviates from the average of all emotion terms across all 34 samples. We interpret the size of the factor scores in the same way as a Cohen’s d (Cohen, 1992). Small size effects (with absolute factor scores between .20 and .50) can be interpreted to indicate that the concept is slightly related to that dimension (for example, that the emotion is either slightly positive or negative). Medium effects (with absolute factor scores between .50 and .80) mean the word is characterized by the dimension (for example, that the word is positive, or negative). Finally, large size effects (with average factor scores of .80 and above) indicate that



Surprise in the GRID 185

the dimension in question is highly marked for that concept (for example, that the emotion is very positive, or very negative). For surprise, the average coordinates on the four dimensions across the 34 samples are .18 for Valence, .17 for Power, .55 for Arousal, and 3.03 for Novelty. Less than small effects (≥ .20) were found for Valence and Power, indicating that surprise is neutral on these two dimensions – in other words, surprise is neither positive nor negative, and neither powerful nor weak. A medium size effect (≥ .50 & NewsMark; AppBase(src : user; pol : polAppUser; negation: [true|false]), where(AppBase) corresponds to the appreciation pattern; pol={positive,negative} is the polarity of the appreciation; src is the source of the appreciation, i.e. the person expressing the appreciation, who is not necessarily the speaker. The NewsMark (“which implies information that is treated as both



Surprise and human-agent interactions 209

new and as surprising or interesting” (Svennevig, 2004)) correponds to the term “ah bon” (translated as “really” in English). Campano, Langlet, Glas, Clavel, and Pelachaud (2015) provides an evaluation grounded in a perceptive study of both alignment at the lexical level with OR and alignment at the polarity level, i.e. sharing appreciation. Concerning sharing appreciation, evaluation results suggest that user’s engagement is improved. Concerning ORs, the statistical tests following our evaluation could not support the idea that other-repetitions by the ECA enhance user’s engagement. First, it could suggest that other-repetition is a subtle phenomenon whose impact on user’s engagement is difficult to isolate and hence to evaluate. Second, contrary to expectations, it could also indicate that other-repetition does not enhance user’s engagement. As user’s engagement seems to be improved by sharing appreciation, this could lead to the interesting conclusion that different alignment phenomena (here, other-repetition and exchanging appreciations) do not contribute to user’s engagement in the same extent. 4. Conclusion We have presented the different roles that surprise can play in human-agent interaction – from user’s surprise detection to agent’s surprise generation based on interaction strategies – and the underlying computational methods. However, the literature focusing on surprise in these computational models is rather sparse. Some systems detect surprise among other emotions without providing in-depth analysis of the specificity of the linguistic and acoustic cues that are the most relevant to detect surprise. Besides, they rarely provide analyses of the special features of surprise in the theoretical framework they use. The assessment is better for existing interaction models between human and ECA and generation model where surprise is better studied both in order to generate facial expressions and prosody associated with surprise and in order to determine when it is relevant to trigger surprise.

References Ahmad, K., Budin, G., Wien, U., Devitt, A., Glucksberg, S., Heyer, G., Leipzig, U., Musacchio, M. T., Pazienza, M. T., Rogers, M., Vogel, C., & Wilks, Y. (2011). Affective computing and sentiment analysis: Emotion, metaphor and terminology. New York: Springer Science+ Business Media.  doi: 10.1007/978-94-007-1757-2

210 Chloé Clavel

Anderson, K., André, E., Baur, T., Bernardini, S., Chollet, M., Chryssafidou, E., Damian, I., Ennis, C., Egges, A., Gebhard, P., Jones, H., Ochs, M., Pelachaud, C., Porayska-­Pomsta, K., Rizzo, P., & Sabouret, N. (2013). The TARDIS framework: Intelligent virtual agents for social coaching in job interviews. In D. Reidsma, H. Katayose, & A. Nijholt (Eds.), Advances in computer entertainment (pp. 476-491). New York: Springer. doi: 10.1007/978-3-319-03161-3_35 Aubergé, V., Audibert, N., & Rilliard, A. (2004). Acoustic morphology of expressive speech: What about contours? Speech prosody 2004 international conference. Nara, Japan. Audibert, N., Vincent, D., Aubergé, V., & Rosec, O. (2006). Expressive speech synthesis: Evaluation of a voice quality centered coder on the different acoustic dimensions. Proceedings of speech prosody. Baccianella, S., Esuli, A., & Sebastiani, F. (2010). Sentiwordnet 3.0: An enhanced lexical resource for sentiment analysis and opinion mining. Language resources and evaluation conference (LREC). Valletta, Malta. Bertrand, R., Blache, P., Espesser, R., Ferré, G., Meunier, C., Priego-Valverde, B., & Rauzy, S. (2008). Le CID-Corpus of interactional data: Annotation et exploitation multimodale de parole conversationnelle. Traitement Automatique des Langues, 49(3), 1–30. Bloom, K., Garg, N., & Argamon, S. (2007). Extracting appraisal expressions. In C. Sidner (Ed.), Proceedings of the HLT-NAACL (pp. 308–315). Morristown: Association for Computational Linguistics. Bohus, D., & Horvitz, E. (2014). Managing human-robot engagement with forecasts and um hesitations. In Proceedings of the 16th international conference on multimodal interaction (pp. 2–9). New York: ACM Press. Callejas, Z., Ravenet, B., Ochs, M., & Pelachaud, C. (2014). A computational model of social attitudes for a virtual recruiter. In Proceedings of the 2014 international conference on autonomous agents and multi-agent systems (pp. 93–100). International Foundation for Autonomous Agents and Multiagent Systems. Campano, S., Durand, J., & Clavel, C. (2014a). Comparative analysis of verbal alignment in human-­human and human-agent interactions. In Proceedings of the ninth international conference on language resources and evaluation (LREC-2014), Reykjavik, Iceland, May 26–31, 2014 (pp. 4415–4422). European Language Resources Association (ELRA). Campano, S., Glas, N., Langlet, C., Clavel, C., & Pelachaud, C. (2014b). Alignement par production d’hétéro-répétitions chez un aca. In Workshop Affect, Compagnon Artificiel, Interaction. Campano, S., Langlet, C., Glas, N., Clavel, C., & Pelachaud, C. (2015). An ECA expressing appreciation. In Workshop on Engagement in Human Computer Interaction, ACII. Cassell, J. (2000). Embodied conversational agents. Cambridge, MA: MIT Press. Celle, A., & Lansari, L. (2014). ‘I’m surprised’ / ‘Are you surprised?’: Surprise as an argumentation tool in verbal interaction. In P. Blumenthal, I. Novakova, & D. Siepmann (Eds.), Les émotions dans le discour / Emotions in discourse (pp. 267–277). Bern: Peter Lang. Clavel, C., Adda, G., Cailliau, F., Garnier-Rizet, M., Cavet, A., Chapuis, G., Courcinous, S., Danesi, C., Daquo, A.-L., Deldossi, M., Guillemin-Lanne, S., Seizou, M., & Suignard, P. (2013b). Spontaneous speech and opinion detection: Mining call-centre transcripts. Language Resources and Evaluation, 47, 1089–1125.  doi: 10.1007/s10579-013-9224-5 Clavel, C., Pelachaud, C., & Ochs, M. (2013a). User’s sentiment analysis in face-to-face human-­ agent interactions – prospects. In Workshop on affective social signal computing, satellite of Interspeech. Association for Computational Linguistics.



Surprise and human-agent interactions 211

Clavel, C., & Richard, G. (2013). Recognition of acoustic emotion. In C. Pelachaud (Ed.), Emotion-­oriented systems (pp. 139–167). London: John Wiley & Sons. doi: 10.1002/9781118601938.ch5 Clavel, C., Vasilescu, I., Devillers, L., Richard, G., & Ehrette, T. (2008). Fear-type emotions recognition for future audio-based surveillance systems. Speech Communication, 50, 487– 503.  doi: 10.1016/j.specom.2008.03.012 Devillers, L., & Vidrascu, L. (2006). Real-life emotions detection with lexical and paralinguistic cues on human-human call center dialogs. In Interspeech 2006. Ekman, P. (1999). Basic emotions. In T. Dalgleish & T. Power (Eds.), The handbook of cognition and emotion (pp. 45–60). New York: John Wiley & Sons. Ekman, P., & Friesen, W. V. (2003). Unmasking the face: A guide to recognizing emotions from facial clues. Los Altos, CA: Ishk. Fant, G. (1960). Acoustic theory of speech production. The Hague: Mouton. Fontaine, J. R., Scherer, K. R., Roesch, E. B., & Ellsworth, P. C. (2007). The world of emotions is not two-dimensional. Psychological Science, 18(12), 1050–1057.   doi: 10.1111/j.1467-9280.2007.02024.x Ishizuka, M. (2012). Textual affect sensing and affective communication. In IEEE 11th international conference on cognitive informatics and cognitive computing (pp. 2–3). Kyoto. Izard, C. E. (1971). The face of emotion. New York: Appleton-Century-Crofts. Karpouzis, K., Andre, E., & Batliner, A. (2010). Emotion-aware natural interaction. Advances in Human-Computer Interaction. Kumar, R., Rosé, C. P., & Litman, D. J. (2006). Identification of confusion and surprise in spoken dialog using prosodic features. In Interspeech 2006. Langlet, C., & Clavel, C. (2014a). Modélisation des questions de l’agent pour l’analyse des affects, jugements et appréciations de l’utilisateur dans les interactions humain-agent. In Conference on traitement automatique du langage naturel, TALN. Langlet, C., & Clavel, C. (2014b). Modelling user’s attitudinal reactions to the agent utterances: Focus on the verbal content. In 5th international workshop on corpora for research on emotion, sentiment & social signals (ES3 2014), Reykjavik, Iceland. Martin, J. R., & White, P. R. (2005). The language of evaluation: Appraisal in English. Basingstoke/ New York: Macmillan. McKeown, G., Valstar, M., Cowie, R., Pantic, M., & Schroder, M. (2011). The SEMAINE database: Annotated multimodal records of emotionally colored conversations between a person and a limited agent. IEEE Transactions on Affective Computing, 3(1), 5–17. doi: 10.1109/T-AFFC.2011.20 Munezero, M., Montero, C.S., Sutinen, E., & Pajunen, J. (2014). Are they different?: Affect, feeling, emotion, sentiment, and opinion detection in text. IEEE Transactions on Affective Computing, 5(2), 101–111.   doi: 10.1109/TAFFC.2014.2317187 Neviarouskaya, A., Prendinger, H., & Ishizuka, M. (2010a). User study on AffectIM, an avatar-­ based Instant Messaging system employing rule-based affect sensing from text. International Journal of Human-Computer Studies, 68(7), 432–450. doi: 10.1016/j.ijhcs.2010.02.003 Neviarouskaya, A., Prendinger, H., & Ishizuka, M. (2010b). Recognition of affect, judgment, and appreciation in text. In Proceedings of the 23rd international conference on computational linguistics (pp. 806–814). Association for Computational Linguistics.

212 Chloé Clavel

Niewiadomski, R., Obaid, M., Bevacqua, E., Looser, J., Anh, L. Q., & Pelachaud, C. (2011). Cross-media agent platform. In Proceedings of the 16th international conference on 3D web technology (pp. 11–19). ACM.  doi: 10.1145/2010425.2010428 Osgood, C., Mai, W. H., & Miron, M. (1975). Cross-cultural universals of affective meaning. Urbana: University of Illinois Press. Osherenko, A., & André, E. (2009). Differentiated semantic analysis in lexical affect sensing. In Conference on affective computing and intelligent interaction (ACII) and workshops (pp. 1–6). Păiş, A. L., Moga, S. A., & Buiu, C. (2010). Emotions and robot artists: State-of-the-art and research challenges. Petroleum-Gas University of Ploiesti Bulletin, Mathematics-Informatics-­ Physics Series, 62(1), 26–40. Pang, B., Lee, L., & Vaithyanathan, S. (2002). Thumbs up?: Sentiment classification using machine learning techniques. In ACL-02 conference on empirical methods in natural language processing-10 (pp. 79–86).  doi: 10.3115/1118693.1118704 Pennebaker, J. W., Francis, M. E., & Booth, R. J. (2001). Linguistic inquiry and word count LIWC2001. Hillsdale, NJ: Erlbaum. Perikos, I., & Hatzilygeroudis, I. (2013). Recognizing emotion presence in natural language sentences. Communications in Computer and Information Science, 384, 30–39. doi: 10.1007/978-3-642-41016-1_4 Picard, R. (1997). Affective computing. Cambridge, MA: MIT Press. doi: 10.1037/e526112012-054 Pickering, M. J., & Garrod, S. (2004). Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27(2), 169–190. Plutchik, R. (1984). Emotions: A general psychoevolutionary theory. In K. R. Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 197–219). Hillsdale, NJ: Lawrence Erlbaum. Poggi, I. (2007). Mind, hands, face and body: A goal and belief view of multimodal communication. Berlin: Weidler. Reisenzein, R., Hudlicka, E., Dastani, M., Gratch, J., Hindriks, K. V., Lorini, E., & Meyer, J.-J. C. (2013). Computational modeling of emotion: Toward improving the inter- and intradisciplinary exchange. IEEE Transactions on Affective Computing, 4(3), 246–266. doi: 10.1109/T-AFFC.2013.14 Rosis, F., Pelachaud, C., Poggi, I., Carofiglio, V., & Carolis, B. D. (2003). From greta’s mind to her face: modelling the dynamics of affective states in a conversational embodied agent. International Journal of Human-Computer Studies, 59(1), 81–118. doi: 10.1016/S1071-5819(03)00020-X Scherer, K. (2003). Vocal communication of emotion : a review of research paradigms. Speech Communication, 40(1–2), 227–256.  doi: 10.1016/S0167-6393(02)00084-5 Scherer, K. R. (2005). What are emotions? And how can they be measured? Social Science Information, 44(4), 695–729.  doi: 10.1177/0539018405058216 Schuller, B., Valster, M., Eyben, F., Cowie, R., & Pantic, M. (2012). Avec 2012: the continuous audio/visual emotion challenge. In Proceedings of the 14th ACM international conference on Multimodal interaction (pp. 449–456). ACM. Smith, C., Crook, N., Dobnik, S., & Charlton, D. (2011). Interaction strategies for an affective conversational agent. Presence: Teleoperators and Virtual Environments, 20(5), 395–411. doi: 10.1162/PRES_a_00063 Svennevig, J. (2004). Other-repetition as display of hearing, understanding and emotional stance. Discourse Studies, 6(4), 489–516.  doi: 10.1177/1461445604046591



Surprise and human-agent interactions 213

Taboada, M., & Grieve, J. (2004). Analyzing appraisal automatically. In Proceedings of AAAI spring symposium on exploring attitude and affect in text (AAAI technical report SS04#07), Stanford University, CA (pp. 158–161). AAAI Press. Vinciarelli, A., Pantic, M., Heylen, D., Pelachaud, C., Poggi, I., D’Errico, F., & Schröder, M. (2012). Bridging the gap between social animal and unsocial machine: A survey of social signal processing. IEEE Transactions on Affective Computing, 3(1), 69–87. doi: 10.1109/T-AFFC.2011.27 Wrede, B., & Shriberg, E. (2003). Relationship between dialogue acts and hot spots in meetings. In IEEE Workshop on automatic speech recognition and understanding, 2003. ASRU’03 (pp. 180–185). IEEE.  doi: 10.1109/ASRU.2003.1318425 Yildirim, S., Lee, C. M., Lee, S., Potamianos, A., & Narayanan, S. (2005). Detecting politeness and frustration state of a child in a conversational computer game. In Interspeech (pp. 2209-2212).

Expressing and describing surprise Agnès Celle, Anne Jugnet, Laure Lansari and Emilie L’Hôte Université Paris Diderot

This paper re-examines the well-established distinction between expression and description of emotion as regards surprise. First, the authors show that the expression of surprise does not involve the use of surprise lexemes, but rather mirative utterances and specific syntactic constructions (while the description of surprise involves surprise lexemes). Second, the investigation of the corpus data indicates that surprise lexemes – namely the noun surprise and the adjective surprised – differ from other emotion lexemes in that they do not systematically refer to the emotional state of an experiencer. The noun surprise can have such a state reading but can also evaluate a source seen as a salient unexpected referent. As for the adjective surprised, it may serve various argumentative strategies. These uses reflect the epistemic rather than emotional nature of surprise, which sets it apart from other emotions. Keywords: adjective surprised, argumentation, emotion, epistemic, English, expression, mirativity, noun surprise

1. Introduction Surprise is one of the six primary emotions identified by Ekman (2003). Not all psychologists agree on whether surprise should be deemed an emotion, for surprise does not involve a valenced reaction.1 In linguistics, surprise is explored from different perspectives, not all of them being concerned with emotion as such. For instance, the verb surprise both in English and French is classified as a psychological verb by Levin (1993), Grafmiller (2013) and Mathieu (2000) on the

1. Ortony et al. (1988, 13) define emotions as “valenced reactions to events, agents or objects, with their particular nature being determined by the way in which the eliciting situation is construed.” For an in-depth discussion of the status of surprise as an emotion, see Soriano et al. (this volume). doi 10.1075/bct.92.09cel © 2017 John Benjamins Publishing Company

216 Agnès Celle et al.

basis of a syntactic feature, i.e. expressing the experiencer as the object.2 Recent accounts of the emotion lexicon have established correlations between the syntactic behavior of affect nouns and verbs and semantic dimensions such as intensity, aspect, causativity and manifestation (Novakova, Goossens, and Melnikova 2012). And yet, the goal of these studies is not to ascertain whether the surprise lexeme realized as a verb, a noun or an adjective actually expresses emotion. Indeed, they all focus on the descriptive function of surprise lexemes but leave aside the expressive contribution of surprise words. The aim of the present paper is to define the status of surprise with respect to emotion from a linguistic viewpoint. It is based on a sample of spoken dialogue data extracted from the American Series In Treatment, a TV drama about a psychologist conducting weekly sessions with patients. In this series, verbal interaction is part of a therapy in which patients discuss personal issues and interpersonal conflicts with their therapist in order to uncover problems and repair relationships. The patient-therapist relationship becomes more complex in the course of the therapy, as the therapist’s self-doubt jeopardizes his position as a resolution expert. On the basis of this data, and in accordance with Plantin (2011, 94; 114–115) and Michaelis (2001),3 we uphold the view that emotion in general and surprise in particular cannot be reduced to a response to a stimulus contra Hochschild (1979). It is argued that surprise entails an epistemic judgment whereby a specific event is connected to the speaker’s set of expectations. Drawing upon both the distinction between emotive communication and emotional communication originally made in philosophy by Marty (1976) and the parallel distinction between expression and communication originally made in linguistics by Bally (1926, 148), we argue that the most salient characteristic of surprise is to disconnect these two functions. Focusing on surprise lexemes, that is, on emotive rather than emotional communication, we then show that the noun surprise is not a prototypical psychological noun. Taking a closer look at the discourse use of the adjective surprised, we finally show that this adjective enters rich pragmatic scenarios. Even if surprise lexemes do not express emotion, their use in discourse suggests that surprise is emotionally meaningful. 2. Goddard (this volume) also analyses the English verb surprise but he adopts a different approach: using the Natural Semantic Metalanguage methodology, he proposes a specific template for this verb. 3. Michaelis (2001, 1039) argues that surprise entails a non-canonicity judgment and rightly points out that a physical startle may not be evidence of surprise: “For example, I might not have predicted a hallway encounter with a colleague, but I would not necessarily find that encounter surprising (even if my colleague startled me).”



Expressing and describing surprise 217

2. The expression of surprise In an illuminating paper on how to develop a more unified pragmatics of emotive communication, Caffi and Janney (1994) stress that the foundations of such a project can be reconstructed from earlier contributions inter alia by Marty (1976) and Bally (1926) in the philosophy of language and in linguistics respectively. In an attempt to clarify the relation between affect and language, these scholars distinguish between two communication modes. Marty (1976) distinguishes between emotional communication and emotive communication. Emotional communication is defined as “a type of spontaneous, unintentional leakage or bursting out of emotion in speech” (Caffi and Janney 1994, 328). Emotive communication, on the other hand, is “the intentional, strategic signaling of affective information in speech and writing […] in order to influence partners’ interpretations of situations and reach different goals” (Caffi and Janney 1994, 328). In the wake of Bergson,4 Bally draws the line between expression and communication, the former being concerned with affect (mode vécu) and the latter with intellectual communication (mode pur). These categories, later taken up in philosophy of language and pragmatics, notably in Searle’s (1969) theory of speech acts,5 are highly relevant for philosophers, psychologists as well as linguists. Such categories provide linguists of different theoretical persuasions with a fruitful and powerful analytical tool for research on emotion.6 While emotions are certainly signaled by specific words in the lexicon, it is not clear whether emotion lexemes themselves actually express emotions, or whether they simply describe them. As pointed out by Legallois and François (2012, 209),

4. Bergson’s intellectual legacy in Bally’s stylistics is pointed out by Legallois and François (2012, 198). These authors also highlight Bally’s pioneering research on the relation between syntax and expressivity long before the role of emotion in discourse started being explored in cognitive linguistics as well as in construction grammar. Given the social nature of affective language in Bally’s research, Caffi and Janney (1994, 335) even regard it as a first landmark in modern pragmatics. 5. Searle (1969) distinguishes between assertive and expressive speech acts. In his theory, assertives aim to describe a state of affairs and may be judged true or false; expressives, on the other hand, express speaker attitude based on sincerity conditions of the speech act. 6. Croft (2009, 412) argues that recognition of the speaker’s intention by the addressee is a prerequisite for any successful linguistic communicative act, a concern that is not addressed in speech act theory: “the identification of construal as an essential part of symbolic meaning, and relating it to cognitive psychological processes, is a major contribution of cognitive semantics. But it must be situated in the larger model of language, communication and joint action in order to understand why it exists and how it is used by speaker and listener.”

218 Agnès Celle et al.

“expressivity is shown, it is not signified.”7 Typically, interjections seem to be devoted to the expression of emotions (oh, hooray, wow). But according to Kövecses (2000, 2–3), adjectives like angry, happy, sad and depressed “describe the emotion they signify” - while I love you may be used both to describe and express emotion. Angry, happy, sad and depressed are used both to describe and express an emotion. The adjective surprised, however, can only describe surprise. Crucially, these two communication modes remain distinct in the case of surprise, which appears to be a characteristic of that emotion.8 While in section 3, the function of surprise lexemes in the description of surprise is analysed on the basis of our corpus data, we start this study with a brief outline of two major means of expressing surprise: (i) mirativity – a category that does not exist in the morphosyntax of English, (ii) the use of specific syntactic constructions. 2.1

Mirativity

Surprise is encoded in a number of languages through mirative markers. The concept of mirativity was first introduced by DeLancey (1997) and further described in DeLancey (2001, 369–370): “The term ‘mirativity’ refers to the linguistic marking of an utterance as conveying information which is new or unexpected to the speaker”. DeLancey claims that mirativity is a category in its own right.9 He illustrates the opposition between old first-hand information (1) and new information (2) in Lhasa Tibetan as follows: (1) nga-r dngul to=tsam yod I-LOC  money some exist ‘I have some money.’ (e.g. I brought some with me) (2) nga-r dngul to=tsam ‘dug I-LOC  money some exist ‘I have some money!’ (quite to my surprise) 

(DeLancey 1997, 44)

The existential copula is direct evidential yod in the case of old information (1), and mirative ‘dug in the case of unexpected new information (2).

7. “L’expressivité se montre, elle ne se signifie pas” (Legallois and François 2012, 209). Our translation. 8. In her analysis of the computer-mediated expression of surprise, Ascone (this volume) reaches the same conclusion: in her data, surprise is never expressed by surprise lexemes. 9. Peterson (this volume) upholds a slightly different view. He claims that mirativity is linked to evidentiality.

Expressing and describing surprise 219



It is not the purpose of the present paper to determine whether mirativity should be analysed as a separate category or in relation to evidentiality. Let us note, however, that several scholars have argued for a unified treatment of information depending on speaker attitudes – rather than only on the source of information. By positing an epistemic scale, i.e. a continuum ranging from the realis domain to the irrealis domain, Akatsuka (1985) argues that surprise is related to the irrealis domain and suggests the following paraphrase for capturing the cognitive assimilation process of newly-learned information: “I didn’t know this until this moment”. More recently, Rett and Murray (2013, 457) have also accounted for the affinity between indirect evidence and speaker surprise by highlighting a common epistemic component. In Cheyenne, among other languages, the same morpheme may be interpreted either as a reportative, i.e. indirect evidential, or as a mirative evidential: (3) a. É-x-hoo’k˙ohó-neho. 3-REM.PST-rain-NAR.SG.INAN ‘It rained long ago, it is told.’ b. É-hoo’k˙ohó-neho! 3-rain-NAR.SG.INAN ‘It’s raining!’ 

(Rett and Murray 2013, 457)

The indirect evidential neho (3) is also a marker of mirativity (3), a fact that is widely represented across languages. Rett and Murray (2013, 470) argue that unlike direct evidentials in (1), indirect evidentials (3) and mirative markers in (2) and (3) both “relate the at-issue proposition p to some salient, contextually-­ valued set of expectations E”. Although mirativity does not exist as a morphosyntactic category in English, DeLancey (2001, 377–378) as well as Rett and Murray (2013) claim that it is marked by intonation. It is indeed possible in English to distinguish between a mirative intonational contour and the intonational contour of an ordinary statement.10 A parallel may thus be drawn between the pair in (1) and (2) and that in (4), the difference between (a) and (b) being purely intonational: (4) a. John arrived on time. b. (Wow,) John arrived on time! 

(Rett and Murray 2013, 455)

10. DeLancey (2001, 377–378) and Rett and Murray (2013, 455–458) point out that the mirative intonational contour is often found in complimentary comments in English.

220 Agnès Celle et al.

According to Rett and Murray (2013, 467), following Rett (2011),11 the mirative construction in (4b), conflates the assertion that p and the expression of speaker surprise: “Because the recency restriction is satisfied, the illocutionary content of the speech act is to assert that p and to express that p was not previously in the speaker’s expectation set”. Rett and Murray (2013, 455) put forward four distinguishing features of the expression of surprise. First, the expression of surprise is “undeniable in discourse” (# No, you are not surprised). Only the content can be denied (No, John didn’t arrive on time). Second, it is “unembeddable under negation and other sentential operators”. John didn’t arrive on time! cannot express that the speaker is not surprised that John arrived on time; Mary said John arrived on time! can only express the speaker’s surprise that Mary said John arrived on time, not the subject’s surprise that John was on time. Third, the expression of surprise is always speaker-oriented – it has an exclamation intonation and not a question intonation. Fourth, it is subject to the recency restriction. The first criterion, i.e. the deniability test, enables Rett (2011, 437) to highlight an important difference between the assertion of surprise and its expression, namely that it is possible to confirm or deny an assertion, whereas the expression of surprise is not debatable: (5) A: I am surprised that I won the contest. B: Yes, you seem shocked. (6) A: Wow, I won the contest. B: #Yes, you seem shocked.

(Rett 2011, 437)

In Rett’s words (2011, 436–437) “while the speaker’s surprise in (5) can be agreed with, a sign that it was part of the assertion, the speaker’s expression of unfulfilled expectation in (6) cannot.” Some of the criteria proposed by Rett to define the expressive dimension of mirative utterances are identical to those used by Potts to define the term expressive (2005, 156). Potts draws a parallel between expressives and conventional implicatures. According to him, it is not clear whether they are best conceived as entailments. Potts argues that expressives have the properties of lexicality, entailment, speaker orientation and independence.12 On this view, the deniability test 11. Rett (2011, 413) defines sentence exclamations as both assertive and expressive: “The utterance of a sentence exclamation counts as an assertion of the denoted proposition p […] and an expression that p violates the speaker’s expectation”. 12. Potts gives the following definitions for these properties. Lexicality: “expressive meaning is part of the lexical meaning of certain expressions, a semantic quality of words and phrases” (Löbner 2002, 32). Entailment: “the aspects of meaning under discussion, in particular, the semantic information displayed by expressive, can have consequences for the notion of logical

Expressing and describing surprise 221



reveals that it is difficult to determine whether expressives are usual entailments. “Expressives are, in this sense, performative. But expressives, unlike more familiar entailments, are also not challengeable by a hearer”. For instance, “a sincere utterance of damn cannot be challenged or turn out to be false” (Potts 2005, 157). Similarly in (7), the property bastard attributed by Speaker A (Ann) cannot be challenged by Speaker B (Kyle) in the following example: (7) Ann: That bastard Conner was promoted. Kyle: Conner is not a bastard.

(Potts 2005, 157)

Although the pairing of the subject and the predicative complement is denied by negative not, Potts claims that “the contribution of the epithet that bastard does become part of the common ground. Kyle refuses to accept the characterization, but this does not in any way mitigate Ann’s use of the epithet.” In Potts’s view, expressives express properties of the speaker’s emotional states – hence outright denial of their content by a hearer makes little sense. The distinction made by Rett (2011) and Rett and Murray (2013) following Potts (2005) is an essential one. Not only do these authors provide formal arguments to disentangle assertion and expression of surprise, but their results also allow us to better understand why the adjectival phrase “be surprised” may enter pragmatic scenarios (see section 4). Assertion as such (I am surprised) may be denied or debated in dialogue, allowing for surprise to be (self-)attributed. In contrast, the expression of surprise is indexical and not accessible to the hearer, who may only deny the content: in (6), No, you didn’t win the contest would be a felicitous reply by B, as opposed to #No, you are not surprised. However, we maintain that Rett’s argument should be qualified. Even in (5), B’s reply is necessarily a modalised assertion (yes, you seem shocked) because it is not possible for B to produce an assertion about A’s inner state (#yes, you are surprised). The infelicity of (6) should be reconsidered, given that (6’) would be more felicitous than (6): (6’) A: Wow, I won the contest. B: You sound shocked / surprised.

Continuation with an evidential utterance in which speaker B evaluates whether s/he rightly perceived A’s expressive statement is perfectly possible, as long as the

validity” (Kaplan 1999, 13). Speaker orientation: “another characteristic distinguishing expressive meaning from propositional meaning is that it is valid only for the utterer, at the time and place of utterance. This limitation it shares with, for instance, a smile, a frown, a gesture of impatience […]” (Cruse 1986, 272). Independence: “Expressive meaning carried by a lexical item in a statement plays no role in determining its truth-conditions” (Cruse 1986, 272).

222 Agnès Celle et al.

utterance is not an outright confirmation or a blunt denial of a proposition. This pattern is represented in (8): (8) Frances: I kept telling myself, “don’t worry, Frances, it’s like riding a bicycle.” Paul: And was it? Frances: He took off my shirt and touched me, and I thought I was gonna lose it. I mean, he loved my breasts. Paul: You sound surprised. Frances: I’m not 25 or synthetic, if you know what I mean. And he just kept kissing them and touching them. Well, I forgot how sensitive they are. Are you shocked?  (In Treatment)

On the basis of its intonational contour in the film and despite the absence of an exclamation mark in the script, he loved my breasts is taken to be a sentence exclamation, i.e. a mirative utterance. While it is true that this utterance cannot be agreed with, which is a sign that it differs from a straightforward assertion, the evidential utterance that follows functions as a request for A’s confirmation of B’s construal of A’s expressive statement. Thus although the proposition is not stable enough to be confirmed or denied, the identification of the emotion that is associated with it is mediated by the auditory evidential. In Sections 3 and 4, we focus on the noun surprise and on the adjectival phrase be surprised, i.e. on lexemes which describe, rather than express surprise. However, it may turn out to be difficult to draw a dividing line between expressives and assertives. For instance, surprise lexemes may be found in exclamative constructions, which, according to Rett (2011; 2008), are expressive, as in (9): (9) What a surprise! How surprising!

Our corpus data suggest that in such cases, surprise lexemes can be used ironically. Along Rett’s (2011, 416; 2008, 603) line of reasoning, such uses correspond to insincere speech acts. We contend, however, that ironic expressives are not only a matter of insincerity, irrespective of possible cultural differences. From a psychological point of view, ironic expressives illustrate that expressives do not necessarily coincide with emotional communication. Contra Potts (2005), we argue that expressives do not necessarily reflect the speaker’s emotional state. Following Legallois and François (2012, 210–211), we consider expressivity to be able to serve not only an emotional function, but also a mimetic one – as in onomatopoeia – and an ethical one – typically in self-presentation and enactment constructions such as be + like. In the latter case, expressives contribute vividness to argumentation, allowing the speaker to display an emotion that s/he does not necessarily feel. Our hypothesis is that the ethical function of expressives is exploited in emotive communication. From a linguistic point of view, we show in

Expressing and describing surprise 223



Section 3 that this use might be better accounted for in terms of modal/evaluative judgment than insincerity. In addition, we argue that the noun surprise used in verbless utterances and exclamations does not behave like a prototypical psych noun. Predicative and metonymic uses of the noun surprise lead us to offer a reinterpretation of its semantic network altogether. 2.2

Syntactic constructions

Syntactic constructions are the second major means of expressing surprise. Krawczak and Glynn (this volume) offer an analysis of three surprise-encoding constructions in English: [what + the + NP], [what + a + NP] and [to + my + NP]. To our knowledge, two types of construction have been described in the literature. Firstly, Mad Magazine sentences (Akmajian 1984; Lambrecht 1990) and Incredulity Response Constructions (Fillmore, Kay and O’Connor 1988, 511) are a type of exclamatives that border on interrogative sentences: “MMs are used by speakers to express surprise, disbelief, skepticism, scorn and so on, at some situation or event” (Akmajian 1984, 2). Secondly, the “What’s X doing Y?” (WXDY) construction is defined by Kay and Fillmore (1999, 21) in this way: “In the WXDY construction we have an overall scene (frame) in which a judgment of incongruity is attributed by a judge (prag) to some state of affairs”. Such constructions are not represented in the corpus data chosen for this study. Therefore, we will simply mention their contribution to the expression of surprise in connection with mirativity. The WXDY construction is illustrated in the following example: (10) Waiter, what’s this fly doing in my soup? 

(Kay and Fillmore 1999, 4)

Such a question is not a request for information about the activity of the fly, but a request for an explanation of its incongruous presence in the soup. As shown by Kay and Fillmore (1999, 30), interrogative what is not referential here. Kay and Fillmore call this use of WXDY a construction because like a “how come” question, it “directly encodes, in addition to a request or demand for an explanation, the pragmatic force of attributing what we call INCONGRUITY to the scene or proposition for which the explanation is required” (Kay and Fillmore 1999, 4). “How come” questions are represented in our corpus data. The proposition that is questioned may not be repeated when it is mentioned in the preceding context: (11) Patient: I’m not ready to talk about that yet. Paul: How come? 

(In Treatment, III 18)

As a question, this construction differs from mirative utterances in that it expresses surprise without asserting it openly. Furthermore, this expression of surprise

224 Agnès Celle et al.

is not indexical, but semantically inherent to the construction. Unlike mirative utterances, WXDY constructions allow for the sense of incongruity to be embedded, as in a conversation between members of a criminal defense team in (12): (12) How are we going to deal with our client’s confusing account of what the photographs were doing in his brief case?  (Kay and Fillmore 1999, 5)

As shown by Kay and Fillmore (1999, 5; 21–22), the source of the incongruity judgment is unclear in embedded contexts: “This means that the WXDY construction does not tell the interpreter how to identify the source of the judgment of incongruity but, in effect, contains an instruction that this is something one should be able to figure out for oneself ”. The judgment is “conventionally associated” with the construction, which accounts for its unrestricted use with respect to time and source. By contrast, a mirative utterance is subject to the recency restriction (cf. Rett) and can only be interpreted as such in indexical relation to the speaker: (13) Waiter, there’s a fly in my soup!

The other construction type that expresses surprise takes the form of an incredulous quotation of a context proposition in interrogative or exclamative utterances: 13

(14) Speaker A: I hear that John may wear a tuxedo to the ball… Speaker B: Him wear a tuxedo? He doesn’t even own a clean shirt.  (Lambrecht 1990, 216)

This use of the infinitive was first described by Akmajian (1984) and compared with the imperative. Lambrecht (1990) claims that this type of structure made up of a NP and a predicate is related not to the imperative, but to topic-comment constructions. He argues that the quotation can only be analysed correctly from a constructionist perspective, by looking at the preceding context proposition and at the follow-up proposition in which the speaker’s judgment is expressed. The context sentence forms a complex topic, whereas the follow-up proposition is a comment about that topic, more specifically a rejection of its validity. Lambrecht (1990, 222) further argues that the IRC is part of a widely attested construction which he calls the Unlinked-Topic Construction, from which it differs in only one respect: “in the IRC one of the topic constituents (the NP) designates a discourse entity”. 13. Lambrecht (1990, 220) distinguishes between the context proposition, i.e. the proposition that precedes in the immediate context, and the context sentence in which the argument and the predicate of the context proposition are quoted and called into question. We use the same terminology for the sake of clarity. However, the term “context sentence” is not well-chosen since it designates a non sentential utterance.



Expressing and describing surprise 225

The context sentence is indeed an incredulous reaction to a discourse event, deconstructing as it does the very pairing of the argument and the predicate previously made by Speaker A. This construction expresses surprise, but unlike mirative utterances it challenges the validity of p. In addition, emotion is expressed in stages: firstly, the context sentence abstracts away the predication from any temporal situation and subjective source, sounding purely emotional. It expresses surprise and incredulity, which is reflected in its non-sentential syntax.14 Secondly, a syntactically more elaborate comment introduces Speaker B’s point of view. It is only in that comment that Speaker B’s emotive judgment appears, overtly rejecting the validity of the context proposition. These constructions express surprise without asserting that p. The expression of surprise may either be inherent to the construction (WXDY construction) or arise from the pragmatic use of a topic-comment construction. In English, it appears that the emotional reaction of surprise is expressed by intonation in mirative utterances and by non sentential elements in constructions. Surprise lexemes are combined neither with mirative utterances nor with constructions. This leads us to focusing more specifically on the noun surprise and its uses in context. 3. The noun surprise: An atypical psych noun The noun surprise has been characterized as a ‘psych’ noun, i.e. a noun that can describe the emotional state of an experiencer (Barque, Fabregas and Marín 2012 and references therein). Our aim is to determine whether this noun is indeed best characterized as a lexeme describing an emotion, i.e. whether this noun shares the properties of core members of the class (such as love, hate or disappointment). We use a corpus-based, cognitive approach to analyze the lexical semantics of the noun surprise – such an approach allows us to account for the relations between the different senses of this noun (see Kövecses (this volume) for a different methodological approach to surprise lexemes). In various languages, psych nouns can be polysemous, in that they can either denote a state or the source (or stimulus) of the state. Disappointment, embarrassment, preoccupation or obsession (inter alia) can denote a feeling or the state of affairs that triggers this feeling (Barque, Fábregas, and Marín 2012; Tutin 2009). 14. Our analysis of the emotional contribution of this construction differs from Lambrecht’s. Lambrecht regards the follow-up judgment as emotional. We argue that the expression of the speaker’s point of view in the follow-up proposition is emotive rather than emotional. By contrast, no point of view is associated with the context sentence, which conveys a spontaneous emotional reaction.

226 Agnès Celle et al.

(15) illustrates the feeling reading of disappointment (that feeling of disappointment presupposes disappointment is what you are feeling), while (16) illustrates the source reading (a disappointment is understood as a source of disappointment): (15) That feeling of disappointment comes from you. (16) Was I that big a disappointment to you? 

(In treatment, III.14) (In treatment, I.5)

Our aim is to determine whether surprise illustrates this regular polysemy, i.e. whether it belongs to the subclass of polysemous (psych) nouns that can both denote a state and have ‘derived’ meanings and refer to the objects or sources of an affect. Our analysis is based on the attested uses of the noun surprise in the script of the series In Treatment. When evidence was lacking in our In Treatment corpus (henceforth IT), we present examples taken from a preliminary study (Jugnet and L’Hôte, to appear) done on the occurrences of the noun in the spoken section of the Corpus of Contemporary American English (henceforth COCA). The occurrences of surprise (N) were encoded on the basis of Plantin’s definition of emotional utterances. Plantin (2011) argues that an emotional utterance combines a series of four parameters: an emotion term, an experiencer, a situation and an attributor. We thus analysed the occurrences of the noun surprise, specifying whether the noun had a state or a source (/object) interpretation (emotion term), whether the experiencer was mentioned or identifiable in context (experiencer), whether the source or stimulus was mentioned (situation), and which attributor could be identified. We discovered that though the opposition between stimulus and state corresponds to an ontologically clear distinction, it is often difficult to make a clear distinction between the state reading and the source reading of the noun in context. It has been argued that psych nouns tend to be mass nouns when they denote a state while they tend to be used as count nouns when they denote a stimulus. Tutin (2009, 68), however, qualifies this generalization, arguing that independent parameters – such as the modification by a qualifying adjective or a type reading, may lead to the use of psych nouns as count nouns. This mass/count criterion can be combined with a second generalization: when the noun describes the source of a state, it cannot occur with a complement describing (specifying) this source: (17) The early arrival of their son [= source] was a surprise to them [=experiencer]. (18) A couple welcomed the surprisingly early arrival of their new son en-route to hospital vs.*A couple welcomed the surprise of the early arrival of their new son [noun surprise denoting the source + of-complement specifying it].

Expressing and describing surprise 227



But as soon as we look at language in use – as is the case in this study for instance, it is often much less obvious whether a state or a source reading is to be preferred, in spite of these two theoretical criteria. 3.1

The state/source polysemy – ambiguities

The difficulty of disambiguating the state/source readings is partly due to the fact that the noun surprise is very often used predicatively, or appears in set phrases. This use, as is argued below, is usually correlated with a comment or epistemic judgment on a given state of affairs (rather than with the expression of an emotion). Though the prototypical function of noun phrases is to refer to objects and to be the arguments of a predicate, some nouns can also be used predicatively, to describe or to comment on a referential argument. Predicative uses of the noun surprise are prominent in our corpus data: in the eleven occurrences of surprise in IT, the noun does not refer to a given participant in a situation, but is used to characterize a situation that was either previously mentioned or mentioned in the following clause. Various types of predicative uses can be distinguished. Most often the noun is part of the predicate, and is introduced by the copula be, as in (19), (20) and (21): (19) Of course they are that’s not really a big surprise.

(IT II,1)

(20) But they didn’t help, which is not really a surprise because the health center is total shit. (IT II,2) (21) It’s no surprise that I can’t find a man.

(IT II,31)

The noun also appears in verbless sentences – without any determiner, modifier or complement. In these cases we consider that the noun is used predicatively, as we assume that verbless sentences correspond to a syntactic structure which involves a verbless predicate – whether adjectival, nominal, pronominal, adverbial or prepositional – associated with (assertive, exclamative, interrogative or imperative) modality (cf. Lefeuvre 1999 inter alia). The modal character of verbless sentences is reflected in the presence of parameters usually found in exclamative utterances, such as repetitions as in (22), or the ‘interjection’/morpheme fucking in (23). (22) Surprise, surprise. (23) Surprise, sur-fucking-prise.

(IT I,4) (IT II,27)

The noun also occurs in the set phrase by surprise, as in (24), which we classify as an instance of predicative use. In this case surprise does not refer to a participant’s

228 Agnès Celle et al.

state of mind in a given situation, but is part of a modifying prepositional phrase, easily paraphrased by an adverbial phrase such as unawares. (24) So your feelings kind of caught you by surprise.

(IT II, 6)

The predicative uses of a noun make it difficult to determine whether a state or a source reading is to be preferred, as the noun is not used to ‘refer’ to a stative eventuality or an object, but rather to comment on a topic. We have nonetheless chosen to determine whether the noun is closer to a state or a source reading, as predicates (whether verbal, adjectival or nominal) can also be polysemous. We thus assume that the distinction between state and source readings still holds in predicative contexts. We now examine in more detail the three types of occurrences of the noun surprise in IT, focusing on the state/source ambiguity. 3.2

Source readings

The analysis of specific occurrences of surprise leads us to conclude that its source reading is relatively more frequent than the state reading.15 When the noun is part of the verb phrase, in extraposed constructions, we give it a source reading: as illustrated in (25), (26) and (27), surprise can be replaced with a cause for surprise in these three contexts. (25) [cf. (19)] Of course they are that’s not really a cause for a big surprise / a reason to be much surprised. (26) [cf. (20)] But they didn’t help, which is not really a cause for surprise because the health center is total shit. (27) [cf. (21)] It’s no cause for surprise that I can’t find a man.

In these cases, surprise is used as a count noun, as is often the case when it has a source meaning. It is noteworthy that negation receives a particular interpretation in such contexts: as the noun is used to comment on a given event, what is negated is not the occurrence of the event but rather its characterization as a cause for surprise. This event can be characterized as not being a (prototypical) member of the class of surprising events (according to the speaker’s mental representations), as in it is not a surprise…, or it can be characterized as lacking all features of a

15. Among the eleven occurrences of the noun surprise in IT, eight have a source reading. This corroborates the findings of another study of 200 occurrences of the noun surprise in the COCA, where 165 occurrences of the nouns were analyzed as involving a source reading, Jugnet and L’Hôte (to appear).



Expressing and describing surprise 229

surprising event (according to the speaker’s standards), as in It is no surprise that I can’t find a man. In verbless utterances such as (22) above, it is particularly difficult to favour one interpretation over the other. Our first criterion, i.e. the use of the noun as a mass noun, may justify its interpretation as a state noun. But it is difficult to perform our second test (using a more explicit noun phrase, such as “#Feeling of surprise, feeling of surprise” or “#Cause for surprise, cause for surprise”). It could be argued that verbless sentences such as “surprise, surprise” are used to describe the addressee’s expectations, i.e. that a state reading is to be preferred. But this type of verbless sentences can also be used ironically (as in (28)), in which case a possible paraphrase for such utterances would be: “(It is) no cause for surprise”. (28) I have a theory about that pizza. – Surprise, surprise. – Would you like to hear it? – Do I have a choice?  (IT, I,28)

In another context, i.e. the prepositional phrase by surprise in (24), the noun surprise can be argued to have a state reading. Our first criterion is that it is used as a mass noun (the singular noun is introduced by the zero determiner rather than the indefinite determiner a). Second, though it is difficult to use our second criterion, i.e. the replacement with a more explicit noun phrase (#your feelings caught you by a state of surprise / #your feelings caught you by a cause for surprise), the possible paraphrases of this set phrase clearly involve the description of a state, as in (29): (29) Your feelings caught you unawares / in a state of surprise.

Though the interpretation of the noun in the set phrase by surprise is not straightforward, we therefore argue for a state interpretation. What is not obvious is whether the state of the subject described by such a phrase is an emotional state or rather an epistemic state. We consider the difficulties of disambiguating the reading of surprise in many contexts significant, as they point to this noun differing from usual psych nouns (such as sorrow or disappointment), as well as from other nouns related to the notion of surprise. If the noun wonder seems to be as ambiguous as its near-­ synonym surprise, nouns like amazement, astonishment, bewilderment or stupefaction are on the other hand non-ambiguous simple state nouns, denoting the (episodic) state of a subject. The fact that they are not ambiguous may be related to their denoting an ‘extreme’ state, or a high degree of surprise: the gradability they involve favours a stative reading (rather than reference to an entity). In any case, the ambiguity of surprise is identified as a crucial property of the noun: at the very essence of surprise (N) lies the fact that the state and source readings cannot

230 Agnès Celle et al.

be analysed as two clearly distinct facets in its semantics, because the two interpretations are too closely connected. Such a position is also justified by the fact that some features of one reading can be connected to the other interpretation. For instance in (30), the noun surprise gets a source interpretation – as shown by the use of the indefinite determiner a and the possible paraphrase That’s not really a cause for surprise. (30) ‘Well, of course the Navy are going to absolve himself ’. ‘Of course they are that’s not really a big surprise’.

(IT II,1)

However, the modifier big cannot modify the source, as it cannot modify events or specific pieces of information. It is rather used to describe the intensity of the state of surprise: as shown in Morzycki (2009), size adjectives can get a degree interpretation when they modify gradable nouns. So surprise (N), even when receiving a source reading, can be associated with degree modifiers that usually modify state nouns. Having established that the two facets of meaning are intrinsically related, we now examine more precisely the type of relation that holds between them, which allows us to revise the semantic network of the noun. 3.3

The state reading as a metonymic extension

Previous descriptions of polysemous state nouns rely on the often implicit assumption that the source reading is derived from the state noun (Novakova, Goossens, and Melnikova 2012; Tutin 2009). More specifically, it is often hypothesized that the source reading is a metonymic extension of the state noun (Barque, Fábregas, and Marín 2012). Our analysis of attested uses of surprise (N) from a cognitive perspective leads us to question the validity of such a hypothesis. Based on the definition of conceptual metonymy and its attested properties in language in use, we argue that surprise (N) cannot be considered as a prototypical state noun. Cognitive linguistics considers metonymy as a conceptual phenomenon with a linguistic expression: “Metonymy is a conceptual projection whereby one experiential domain (the target) is partially understood in terms of another experiential domain (the source) included in the same common experiential domain” (Barcelona 2003, 4). Adopting a cognitive approach to the semantic network of surprise (N) involves the identification of a prototype and of a series of senses that are extensions of this prototype: “A prototype is a typical instance of a category, and other elements are assimilated to the category on the basis of their perceived resemblance to the prototype; there are degrees of membership based on degrees of similarity” (Langacker 1987, 371). In his seminal work on the issue, Langacker



Expressing and describing surprise 231

only allows for metaphorical extensions. But more recent publications – including Langacker 2004 – have come to emphasize the relevance of metonymic extensions in semantic networks (Seto 2003; Blank 2003; Radden and Kövecses 1999). Cognitive semantics provides us with several arguments in favour of a semantic network for the noun surprise in which the source reading is considered as the prototypical reading, while the state reading becomes a metonymic extension of the latter. These arguments rely (i) on etymology, (ii) on cognitive patterns involved in metonymic relations, (iii) on the distinction between predicative and referential use, and (iv) on the relevance of set expressions and linguistic patterns in the analysis of metonymic extensions. In this study, these arguments are only used to determine how the source reading and the state reading of surprise are connected in the semantic network of the noun – not to prove whether or not there is a metonymic extension within the network in the first place, which is one of the founding assumptions of our work. Let us start with the etymological argument (i). Diachronically speaking, the source reading of surprise is older than its state reading in English. In the late 14th century, occurrences of the noun are found to mean “unexpected attack or capture”. Similarly, in 1590, surprise occurs with the meaning of “something unexpected”. It is only in 1600 that attested use of the state reading can be found (Hoad 1993). While recent work in metaphor analysis (Pragglejaz Group 2007) points out that the basic (and thus non-metaphorical) reading of a word is not necessarily the oldest, we argue that in the case of the analysis of metonymic relations within a semantic network, the diachronic argument is relevant. In their work on the development of a comprehensive and integrated cognitivist view of metonymy (ii), Radden and Kövecses (1999, 39) show that consequences of an event are more frequently identified as metonymies of its causes than the other way around. This corroborates our hypothesis that in the case of surprise, it is the state reading – the state of surprise that is triggered by a surprising event – that is a metonymic extension of the source reading – the surprising event which causes a subsequent state of surprise. The distinction between referential and predicative uses of a noun is the third argument we put forward in favour of our hypothesis (iii). Indeed, in his study of the role of mappings and domains in the understanding of metonymy, Ruiz de Mendoza Ibanez (2003, 114) points out that a metonymy is rarely used predicatively, but tends to be used referentially. These findings fit within our analyses, which have established that a majority of predicative uses of the noun surprise rely on the source reading, while the state reading rarely occurs in such cases. Thus it makes sense to consider the source reading as prototypical, and the state reading as a metonymic extension of the latter.

232 Agnès Celle et al.

Finally, in his corpus-based study of the occurrences of the noun eye, Hilpert (2006, 145) argues that all the metonymic extensions identified for the semantic network of the word occur as part of recurrent linguistic patterns (iv). Similarly, in the case of surprise, we have noted that set expressions and recurring linguistic patterns (e.g. by surprise) account for a large proportion of the occurrences of the state reading of the noun, thereby pointing to the latter following the usual pattern of metonymic extensions rather than prototypical readings for nouns. Based on the combination between our preliminary COCA study, the analysis of our IT corpus data and the afore-mentioned series of arguments, we thus propose a gradient of meaning between the prototypical sense of surprise (i.e. its source reading) and its metonymic extension (i.e. its state reading). This allows us to take into account the attested ambiguities of the noun in the new representation of its semantic network, primarily because of inherent ambiguities in the definition of domains in cognitive linguistics: according to Barcelona (2003, 8), they include “all the entrenched knowledge that a speaker has about an area of experience”. This means that just like prototypes, domains are culturally dependent and may even vary from speaker to speaker. Just as the distinction between metaphor and metonymy can sometimes be blurred by these fuzzy boundaries between domains, we argue that the distinction between the prototypical sense of a word and its metonymic extension is not always clear-cut, for similar reasons. This is how we account for the fact that surprise (N) collocates with adjectives usually associated with states as well as with adjectives usually associated with events. All previously analysed occurrences of surprise are thus placed on a gradient of meaning between the source reading of the noun (i.e. its prototypical sense) and its state reading (i.e. the metonymic extension of the prototype). Cognitive semantics proposes a specific location on the source-state gradient, thereby allowing for ambiguities to be fully accounted for. Several arguments have led us to claim that the prototypical sense (or ‘primary reading’) of the noun surprise is its source reading, its state reading being a metonymic extension. In this respect, surprise differs from most other psych nouns like sorrow, worry, concern, obsession or disappointment, which primarily denote states and not sources. This conclusion leads us to wonder whether surprise can be analyzed as a true psych noun. 3.4

What kind of state?

We thus question the very inclusion of surprise in the class of psych nouns in English, as it differs from ‘prototypical’ (polysemous) state nouns, which tend to follow the patterns described by Barque et al. (2012): they can denote states or sources, but are not particularly ambiguous, as in (31).



Expressing and describing surprise 233

(31) She gets overwhelmed. She needs to vent. So after she vents and she puts the stress, the disappointment and the worry of her own life on your shoulders, who do you turn to?  (IT II,17)

The noun is usually not used to describe the feelings or emotions of an experiencer, but rather an epistemic judgment on the likelihood of a given state of affairs. As illustrated above (examples (19) to (21)), surprise can be found in the extraposed structure (i.e. the pattern it is (not) a surprise that p). The function of such a structure is not to describe the feelings of an experiencer, but rather to introduce a judgment as to the (un)expected character of a situation. It could thus be argued that the structure allows a speaker to express an epistemic judgment (based on a cognitive state, a given situation is considered (un)likely), rather than an emotional stance towards a state of affairs. The structure it’s no surprise that p can thus be paraphrased as According to my knowledge (set of expectations), that p is very likely/obvious/must be the case. The interpretation of verbless sentences (examples (22) and (23)) is similar to that of the extraposed structure: the speaker judges a given situation likely (or unlikely), and does not express his/her emotion. Even in contexts in which the noun can have a state reading, as by surprise in (24), the noun is not used to express the cognitive or epistemic state of a subject (as shown by the paraphrase unawares: such an adverb characterizes a subject’s stance on a topic, not their own cognitive state). Still, surprise shares some properties with typical psych nouns. Like other state nouns, surprise can appear in the prepositional phrase in + his/her (or any description of an experiencer) + psych noun. In this phrase, the noun denotes a feeling with a causal value – or, in Van de Velde’s (1998) terms, the feeling is conceived as an (abstract) location of a subject. (32) a. In his fear, he reached out for support. / In his despair he cursed those who had taken her away from him. b. In his surprise, he could not help staring, suddenly sure that the fellow Klingon would recognize and expose him.

Surprise can also be the argument of experience or feel: (33) The despair/fear/surprise that he experienced…

But contra Blumenthal (2009) for instance, we argue that the lexeme surprise is not a central (or prototypical) psych noun, based on a series of parameters. We also reject the hypothesis that surprise is better characterized as an epistemic state noun. If this hypothesis were validated, surprise (and wonder) would be similar to nouns such as belief, knowledge or assumption, which can also be ambiguous and denote the state or the (propositional) content of an epistemic

234 Agnès Celle et al.

stance. Such nouns, which are a subclass of “conceptual shells” (Schmid 2000), can have sentential complements, and can appear in an identificational structure involving a sentential constituent (the clause describing a propositional content), as in (34) and (35): (34) Still, the idea that we have little access to the workings of our minds is a dangerous one. (35) The idea that she mentioned is that we have little access to the workings of our minds is a dangerous one.

There is no attested use of the noun surprise in such contexts in IT or in COCA. Moreover, if surprise denoted an epistemic stance or its content, it could be modified by the adjectives true or false (cf. a false belief/assumption/hypothesis), which is obviously not the case. Thus surprise cannot be considered a prototypical epistemic state noun either. Overall, surprise cannot be considered as a prototypical psych noun, but cannot be classified as an epistemic state noun either. In other words, the noun resists categorization. The specificity of surprise lexemes is confirmed by our analysis of the adjective surprised. Section 4 investigates the discourse use of the adjective surprised in association with first and second persons in the present tense. 4. The adjective surprised and complex pragmatic scenarios This corpus study confirms that surprise lexemes do not express emotion. The adjective surprised enters various pragmatic scenarios which all point to a disconnection between the assertion and the expression of surprise. Two main cases can be distinguished in the patient/therapist interactions under scrutiny in our IT corpus. First, the adjective appears in questions asked or inferential statements made by patients (you’re surprised, are you surprised?) in an attempt to expose the therapist’s surprise – more often than not the therapist responds by denying his surprise in order to regain control over the situation. Secondly, the adjective is used in assertions of surprise (I’m surprised). Such assertions do not correspond to an expression of surprise but constitute a strategy to minimize emotions such as anger or disappointment, or correspond to a way of rationalizing surprise.

Expressing and describing surprise 235



4.1

Experiencing and controlling one’s surprise

We first focus on interactions where the therapist is clearly surprised by one of his patients’ declarations and yet refuses to acknowledge his surprise, thus illustrating what Plantin (2011, 150) calls “dissonance” between emotive and emotional communication.16 Such instances show that attributing surprise may be viewed as face-threatening and, as a corollary, that displaying surprise may amount to a problematic loss of control. As described in Goffman’s work on the issue, hiding or denying one’s surprise is thus a face-saving attempt to regain control (Goffman 1986). In his work on the figurative language of emotions, Kövecses (2000) argues that the metaphors used to express emotions all rely on the same cognitive scenario involving five stages: Cause → Emotion → Control → Loss of Control → Behavioral Response

Interestingly, Kövecses (this volume) shows that surprise differs from other emotions as far as control and loss of control are concerned. In the case of anger and fear, there exist many metaphors describing the experiencer’s attempt at regaining control over the situation. In the case of surprise, metaphors signalling attempts to resist the emotion are conspicuously absent. Kövecses thus concludes that loss of control is the most fundamental quality of surprise, as shown in metaphors such as “catch someone off-guard”. Kövecses does not assume that these findings are relevant for the other two types of emotion language he defines (i.e. descriptive language and non-metaphorical expressive language), but in our IT corpus it is the control stage that is particularly relevant. The need for control on the part of the therapist is highly understandable in the context of his sessions with his patients: surprise is inherently epistemic and emerges when there is a discrepancy between one’s expectations and reality, thus exposing the speaker’s “unprepared mind” (Aikhenvald 2006, 195). For the therapist, being caught off-guard by his patients would point to his possible lack of insight or “epistemic failure” (Kövecses, this volume) on the matter discussed , as exemplified in (36): (36) Laura: Because he’s right. Andrew’s right. I am being unfaithful to him. Paul: You mean what happened last night at the club? Laura: Not just last night. I’ve been unfaithful to him for a long time. Paul: Why haven’t we talked about this before? Laura: I think we have. It’s been here all along. You mean to say you’ve never noticed it? You’re surprised. 16. “dissonance entre l’émotif et l’émotionnel”. Our translation.

236 Agnès Celle et al.

Paul: No, I’m not surprised. I just…I don’t follow what you mean. Laura: This isn’t… the reaction I was hoping for. Not at all (IT).

Laura has been in love with her therapist, Paul, for a long time and has decided to tell him. Seeing Paul’s face, Laura realizes he is surprised by her tentative revelation. Her inferential statement “you’re surprised” may be regarded as an indirect reproach, hence Paul’s denial (No, I’m not surprised). His denial is all the more interesting as it is contradicted by his facial reaction throughout the dialogue and by linguistic evidence, namely the syntactic disfluency I just… I don’t follow…. Similarly, in (37), Paul didn’t expect Alex to come and Alex infers from Paul’s face that he is surprised. However, Paul is reluctant to admit his surprise: (37) Alex: Paul: Alex: Paul: Alex:

Surprised? I didn’t think, uh… I didn’t think you’d come back again. I’m back. Good. Come on in. Would you like some coffee? I’ve just made a pot. Yeah, coffee’s good.  (IT)

In both examples, Paul denies or eludes his surprise by resorting to rational arguments – he points to his lack of understanding in (36) and refers to prior expectations in (37) – but his surprise remains visible at the syntactic level. In (38), as Paul almost admits to being surprised, syntactic disfluency in his response is associated with the clear acknowledgment that surprise is not an adequate reaction on his part (see I shouldn’t be): (38) Alex: OK, so my squadron commander calls me back up to his office. This time all the brass were there and they want to know if I’m interested in active duty. If I am, I have to go through a modified basic training at Pensacola, a battery of tests, physical, psychological… just to make sure I’ve recovered. So I said, “Bring it on, baby! I’m ready for anything!” Are you surprised? Paul:  Well, a little, I guess. I, um… I shouldn’t be, but I suppose it was always a possibility. Alex: But you think it’s too fast, right? Paul: Do you think it’s too fast? Alex: Doc, you have no idea how good it feels to be useful again. I think that’s been the real problem. I’ve been sitting around feeling sorry for myself. I’m a doer, not a talker, so all of this, that’s over. Those days are over.

In all these utterances, the speaker experiencing surprise never spontaneously asserts his surprise by saying I am surprised. This shows that the expression of

Expressing and describing surprise 237



surprise is not conveyed by lexemes but by syntactic phenomena, as evidenced by the recurrence of questions and cases of disfluency. Conversely, we argue that saying I’m surprised does not correspond to the expression of an emotional reaction of surprise either. 4.2

Asserting one’s surprise

4.2.1 Minimizing one’s anger and disappointment Our contention is that, in many cases, saying I’m surprised does not constitute an expression of surprise but rather a strategy to minimize other emotions associated with negative valence such as anger and disappointment. Resorting to a more neutral emotion17 is thus a way to save one’s face and/or avoid hurting the interlocutor’s feelings. Like Paul’s reluctance to admit surprise examined in previous examples, this strategy of minimization is an attempt to control one’s emotions. Contrary to (36) and (37) above, no hesitation may be observed in the utterance supposedly expressing surprise in the following dialogue between Paul and Gina, his supervisor: (39) Gina: You’ve created this cold version of me that’s limiting and castrating. Because when you’re arguing with me you want to see yourself as battling the forces of repression. And if the truth is that I’m an emotional person who has to rein herself in, it doesn’t fit with what you need, now does it? […] I didn’t sleep with Charlie because… that’s not who I wanted to be. I loved David and I loved the life we had and who I was with him. I loved myself with David. David’s Gina. And you know what? When I was standing over… his grave at the funeral… I thought… if I was magically reborn… I would choose him all over again. For all his faults… (Sniffs)… he allowed me to be who I wanted to be. I’m surprised you didn’t see it, how crazy I was about him. He was everything. Paul: I never heard you say it. Gina: Oh, shit. (Sniffs) Everything has to be said? And now I have to sit here and listen to your crappy, near-sighted, self-involved theories about my life.

Gina’s “I’m surprised you didn’t see it, how crazy I was about him” is not spontaneously triggered by what Paul has just said; rather, it comes as a well-prepared conclusion to her own angry speech. Gina relies on the surprise lexicon to re17. See Ortony et al. (1999, 32), who point out that “surprise can arise in the absence of a valenced reaction”.

238 Agnès Celle et al.

proach Paul with his misconception of who she is. It is an indirect way to tell him: you should have seen it. The syntactic structure in the complementation of surprised (right dislocation) explicitly points to the referent of the pronoun it (i.e. how crazy I was about him), thereby suggesting a lack of trust in Paul’s own analysis. Surprised does not denote emotional communication but emotive communication here. What Gina is truly experiencing is more akin to anger and disappointment  – and more particularly person-related disappointment.18 The question is, then, why she should say I’m surprised instead of stating I’m disappointed/angry/sad…, which are defined as negatively valenced adjectives in the literature on emotions (Ortony, Clore, and Collins 1988). Let us compare (39) with (40): (40) Paul: Have you spoken to Dr Kornreich about terminating this pregnancy that you’ve wanted for so long? Amy: Yes. I don’t think, um… I’ll be able to love another baby. I’m angry at him already. I feel like it’s forcing itself into the middle of things. Ever since I became pregnant I felt the hormones raging, driving me insane, and I know I’ll have postpartum, that’s what happened last time. Paul: Are you angry? It feels like you’re angry tonight. Amy: I’d say that’s a fairly accurate assessment. Paul: You wanna tell me why? Amy: Actually, I’m angry at myself for taking your advice, even though I knew better. I will say that it’s not the first time I’ve left here disappointed or angry. Paul: If I was you, I’d look into that. Amy: Of course, I’m angry at everything these days. More or less all the time. This week was particularly…

In this sequence, there is no denial, since pregnant Amy readily admits to experiencing negative emotions, mostly anger. In 3 occurrences out of 4, she even provides her therapist with the source of her anger (see the complements introduced by at). The therapist’s suggestion (If I was you, I’d look into that) shows that negative emotions are inherently problematic and are proof of a malfunction. Asserting person-related disappointment or anger in (39) would have been a direct reproach on Gina’s part, and the sign of her problematic relationship with Paul. In terms of valence, I’m angry/disappointed that you didn’t see it denotes negative 18. See Van Dijk and Zeelenberg (2002) on the distinction between outcome-related disappointment and person-related disappointment. These authors contend that anger is always person-related.



Expressing and describing surprise 239

valence, whereas I’m surprised presents itself as a neutral epistemic judgment and amounts to I don’t understand: it merely states the discrepancy between one’s prior expectations and reality, without any negative evaluation. In other words, I’m surprised is part of Gina’s control strategy to sound as professional as possible and to avoid an open conflict with Paul. But after his justification (I never heard you say it), she finally gives way to her anger, thus illustrating the shift from control to loss of control defined in Kövesces’s scenario. The same kind of control strategy is at stake in the exchange between Paul and his patient Mia in (41). Mia, who has always been a difficult patient, accuses Paul of sleeping with another patient. Paul tries to remain calm and refuses to acknowledge that he is angry with her for interfering with his private life: (41) Mia: Paul, do you have any idea how many therapists are successfully sued for sleeping with their patients? Paul: What are you trying to say, Mia? Mia: How could you? Was she that young and pretty? Or did she make you feel that big and important? Take it. It’s yours. Do you want me to pick that up? It’s all yours. I’ve just made an ass of myself, haven’t I? I bet you’re angry. Paul: Maybe I am. Mostly, I’m… I’m surprised. Mia:  No, you’re furious. I can see it all over your face. But… you just can’t express it because you’re not supposed to. You want me to go?

The hesitation in Paul’s verbal response shows that he is buying time to find an appropriate label for his reaction. The choice of surprise, a neutral emotion which is not associated with a negative valence, is nonetheless contradicted by his facial reaction. Mia is no fool and is fully aware of Paul’s control and face-saving strategy. She points out that admitting his anger would be unprofessional on his part (cf. because you’re not supposed to). It would clearly reveal that Mia has managed to destabilize him – which is exactly what she has been aiming at since the beginning of her therapy. It would be tempting to conclude from the examples examined so far that the assertion of surprise never corresponds to an expression of surprise but is systematically used to hide or deny negative emotions. Our data suggest however that the “dissonance” between emotive and emotional communication is more complex than that. 4.2.2 Rationalising surprise In (42), it is not the therapist who says I’m surprised but a patient – Amy, who had a miscarriage during a previous session with Paul.

240 Agnès Celle et al.

(42) Amy: Where’s the stain? I saw it when we left here. You must’ve scrubbed the couch like a lunatic. Paul: Actually, it, uh… it came off quite easily. Amy: Oh. Paul: Hm. Amy: I’m surprised, I… actually thought you’d have been down on all fours scrubbing away, and your wife standing over you shouting instructions and… you saying, “It’s not coming out, I don’t know what to do, “my next patient’ll be here any minute.” I didn’t think it would come out that easily, that’s all. Paul: Did you want it to still be there? Amy: Yes, I was very attached to it. Where’s your sense of humour? C’mon. Paul: Well… Guess I, uh… I don’t find it so… so funny. I think you’re disappointed. Amy: Why are you making a big deal out of it? Paul: What were you expecting to feel when you saw it? Amy: Nothing. Paul: No sadness? No grief? Amy: Over a stain? Paul: No. Over the loss of your baby. Maybe you wanted to see that stain because you needed to accept it, and all the mixed feelings you have about it.

The interjection oh indicates that Amy is surprised by the fact that the blood stain on Paul’s couch has disappeared. Contrary to examples (39) and (40) examined above, there seems to exist here a correspondence between the lexeme used by the speaker and the emotion experienced by the latter. It is worth noticing, however, that Amy’s first verbal response to the surprising element is not an assertion of surprise but a spontaneous, uncontrolled emotional outburst (oh). This points to the fact that I’m surprised may be uttered to express surprise but only once surprise has been cognitively integrated – as though I’m surprised corresponds to a comment on or an explanation of the initial truly emotional reaction. So although no control strategy is involved in the case of (42), it is still questionable whether I’m surprised should be analysed as expressing surprise. We contend that it is best analysed as an attempt to rationalize surprise, as it is indeed immediately followed by expressions pointing to the discrepancy between the subject’s expectations and reality (I actually thought / I didn’t think).19 It is also relevant that Paul questions 19. Similar findings based on different corpora and genres allow us to claim that this argumentative strategy is not biased towards a therapeutic discourse on emotions. See Tutin (this volume), Celle (to appear).



Expressing and describing surprise 241

Amy’s surprise and reinterprets it as disappointment (I think you’re disappointed). He thus suggests that Amy is denying her true feelings to protect herself. As a therapist, he cannot believe that her miscarriage has not brought about negatively valenced emotions. 5. Conclusion The goal of this paper was to establish what the linguistic status of surprise was, in comparison with other more prototypical emotions, none of which are considered to be reducible to the response to a stimulus (Plantin 2011; Michaelis 2001). Based on our analyses of dialogues taken from the US TV show In Treatment, we have argued that surprise entails an epistemic judgment whereby a specific event is connected to the speaker’s set of expectations. According to our study, surprise can be characterized as the locus of a disconnection between emotive and emotional communication (Marty 1976), identified linguistically as a disconnection between expression and communication (Bally 1926). This means that linguistic constructions express surprise without stating it explicitly. The expression of surprise may either be inherent to a given construction, or stem from the pragmatic use of a topic-comment construction. In our data, emotional reactions of surprise are expressed by the speaker’s intonation in mirative utterances, and by non-sentential elements in constructions. As we have found that surprise lexemes are combined neither with mirative utterances nor with constructions, the second part of our study has concentrated on the former, i.e. surprise (N) and surprised (Adj) in our In Treatment data. Surprise (N) is shown to resist clear-cut categories. It can be classified neither as an epistemic state noun nor as a prototypical psychological noun, as its prototypical sense is actually what we call its source reading. The state reading of the noun is thus identified as a metonymic extension of that prototypical sense, thereby reversing the expected structure of its semantic network, based on those associated with most other psych nouns like sorrow, worry, concern, obsession or disappointment. Similarly, the adjective surprised does not fall into prototypical scenarios, but rather rich and complex ones, pragmatically speaking. A speaker usually does not rely on the phrase I’m surprised to express her/his surprise, but uses it as part of a strategy aimed at playing down negatively valenced emotions like anger or disappointment. The reliance on a more neutral emotion like that of surprise thus becomes a face-saving strategy and/ or an attempt to control one’s emotions. We point out that I’m surprised can only be used to express surprise once that emotion has been cognitively integrated by the speaker. The reliance on surprised (Adj) then becomes a comment on or an explanation of the initial emotional reaction, not the expression of it.

242 Agnès Celle et al.

Thus our study goes beyond the idea that the assertion of surprise is systematically used to hide or deny negative emotions. The dissonance identified between emotive and emotional communication is shown to be more complex in our data. Similarly, even if surprise lexemes do not express emotion, their use in discourse still suggests that surprise is emotionally meaningful.

References Aikhenvald, A. Y. (2006). Evidentiality. Oxford University Press. Akatsuka, N. (1985). Conditionals and the Epistemic Scale. Language 61 (3): 625–639. Ascone, L. (this volume). The computer-mediated expression of surprise: A corpus analysis of chats by English and Italian native speakers and Italian learners of English. Bally, Ch. (1865–1947. 1926). Le langage et la vie. Paris: Payot. Barcelona, A., ed. (2003). Metaphor and Metonymy at the Crossroads : A Cognitive Perspective. Topics in English Linguistics 30. Berlin: Mouton de Gruyter. Barque, L., Fábregas, A. & Marín, R. (2012). Les noms d’état psychologique et leurs“ objets”: étude d’une alternance sémantique. https://129.242.170.252/munin/bitstream/handle/10037/4603/article.pdf?sequence=3. Blank, A. (2003). Polysemy in the Lexicon and in Discourse. In B. Nerlich, Z. Todd, V. Herman, & D. D. Clarke (Eds.), Polysemy: Flexible Patterns of Meaning in Mind and Language, Trends in Linguistics. Studies and Monographs 142. Berlin, New York: Mouton de Gruyter. 267–296. Blumenthal, P. 2009. Les noms d’émotion: trois systèmes d’ordre. In I. Novakova & A. Tutin (Eds.), Le lexique des émotions. Langues, Gestes, Paroles. Grenoble: ELLUG, Université Stendhal. 41–64. Caffi, Cl., & Janney, R. W. (1994). Toward a Pragmatics of Emotive Communication. Journal of Pragmatics 22 (3–4): 325–373.  doi: 10.1016/0378-2166(94)90115-5. Celle, A. (in press). Surprise in Storytelling. In R. Nita & F. Valetopoulos (Eds.), Expressing emotions in corpora. Peter Lang. Croft, W. (2009). Toward a Social Cognitive Linguistics. In V. Evans & S. Pourcel (Eds.), New Directions in Cognitive Linguistics. Human Cognitive Processing. Amsterdam; Philadelphia: John Benjamins. 395–420. Cruse, D. A. (1986). Lexical Semantics. Cambridge Textbooks in Linguistics. Cambridge [Cambridgeshire]; New York: Cambridge University Press. DeLancey, S. (1997). Mirativity: The Grammatical Marking of Unexpected Information. Linguistic Typology (1). DeLancey, S. (2001). The Mirative and Evidentiality. Journal of Pragmatics 33 (3): 369–382. Goddard, C. (this volume). The complex, language-specific semantics of “surprise”. Goffman, E. (1986). Frame Analysis: An Essay on the Organization of Experience. Boston: Northeastern University Press. Grafmiller, J. (2013). The Semantics of Syntactic Choice, an Analysis of English Emotion Verbs. Ph.D. Dissertation, Stanford University.



Expressing and describing surprise 243

Hilpert, M. (2006). Keeping an Eye on the Data: Metonymies and Their Patterns. In A. Stefanowitsch & S. T. Gries (Eds.), Corpus-Based Approaches to Metaphor and Metonymy. Trends in Linguistics. Studies and Monographs 171. Berlin: M. de Gruyter. 123–151. Hoad, T. F. (1993). The Concise Oxford Dictionary of English Etymology. Oxford Paperback Reference. Oxford, New York: Oxford University Press. Hochschild, A. (1979). Emotion Work, Feeling Rules, and Social Structures. American Journal of Sociology, 85/3, 551–575. Kaplan, D. (1999). What Is Meaning? Explorations in the Theory of Meaning as Use. Ms. Draft 1. UCLA. Kövecses, Z. (2000). Metaphor and Emotion: Language, Culture, and Body in Human Feeling. Studies in Emotion and Social Interaction. Second Series. Cambridge/New York: Cambridge University Press. Kövecses, Z. (this volume). Surprise as a conceptual category. Krawczak, K. & Glynn, D. (this volume). Operationalizing mirativity: A usage-based quantitative study of constructional construal in English. Jugnet, A. & L’Hôte, E. (to appear). Looking at ‘unexpectedness’. A corpus-based cognitive analysis of surprise & wonder. In N. Depraz & A. Celle (Eds.), Surprise at the intersection of Phenomenology and Linguistics. Amsterdam: John Benjamins [Consciousness and Emotion Series]. Langacker, R. W. (1987). Foundations of Cognitive Grammar: Theoretical Prerequisites. Stanford University Press. Langacker, R. W. (2004). Metonymy in Grammar. Journal of Foreign Languages, 6.2–24. Lefeuvre, F. (1999). Les ‘marqueurs de prédication’ dans la phrase averbale en français. Verbum XXI, 4 (XXI, 4): 429–438. Legallois, D. & François, J. (2012). Définition et illustration de la notion d’expressivité en linguistique. In N. Le Querler, F. Neveu, & E. Roussel (Eds.), Relations, connexions, dépendances: hommage au Professeur Claude Guimier. Rennes: Presses universitaires de Rennes. 197–221. Levin, B. (1993). English Verb Classes and Alternations: A Preliminary Investigation. Chicago: University of Chicago Press. Löbner, S. (2002). Understanding Semantics. Understanding Language Series. London: Arnold. Marty, A. (1874–1914. 1976). Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie. Hildesheim, New York: Olms. Mathieu, Y. Y. (2000). Les verbes de sentiment: de l’analyse linguistique au traitement automatique. Sciences Du Langage. Paris: CNRS éditions. Michaelis, L. (2001). Exclamative Constructions. In M. Haspelmath, E. König, W. Österreicher & W. Raible (Eds.), Language Typology and Language Universals: An International Handbook. Berlin, New York: Mouton de Gruyter. 1038–1050. Morzycki, M. (2009). Degree Modification of Gradable Nouns: Size Adjectives and Adnominal Degree Morphemes. Natural Language Semantics 17 (2): 175–203. Novakova, I., Goossens, V. & Melnikova, E. (2012). Associations sémantiques et syntaxiques spécifiques. Sur l’exemple du lexique émotionnel des champs de surprise et de déception. In F. Neveu, V. Muni Toke, & P. Blumenthal (Eds.), 3e Congrès mondial de linguistique française: Lyon, 4–7 juillet 2012. Les Ulis, France: EDP Sciences: 1017–1029. Ortony, A., Clore, G. L. & Collins, A. (1988). The Cognitive Structure of Emotions. Cambridge ; New York: Cambridge University Press. Peterson, T. (this volume). Grammatical evidentiality and the unprepared mind.

244 Agnès Celle et al.

Plantin, C. (2011). Les bonnes raisons des émotions: principes et méthode pour l’étude du discours émotionné. Vol. 94. Sciences pour la communication. Bern, New York: Peter Lang. Potts, C. (2005). The Logic of Conventional Implicatures. Oxford: Oxford University Press. Pragglejaz Group. (2007). MIP: A Method for Identifying Metaphorically Used Words in Discourse. Metaphor and Symbol 22 (1): 1. Radden, G. & Kövecses, Z. (1999). Towards a Theory of Metonymy. In K.-U. Panther & G. Radden (Eds.), Metonymy in Language and Thought. Human Cognitive Processing, 1387–6724; v. 4. Amsterdam, Philadelphia: John Benjamins. 17–59. Rett, J. (2008). A Degree Account of Exclamatives. In Proceedings of SALT 18: 601–618. http://elanguage.net/journals/salt/article/download/18.601/1921. Rett, J. (2011). Exclamatives, Degrees and Speech Acts. Linguistics and Philosophy 34 (5): 411–442. Rett, J. & Murray, S. (2013). A Semantic Account of Mirative Evidentials. In Proceedings of SALT 23: 453–472. http://elanguage.net/journals/salt/article/view/23.453/0. Ruiz de Mendoza Ibanez, F. J. (2003). The Role of Mappings and Domains in Understanding Metonymy. In A. Barcelona (Ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Topics in English Linguistics 30. Berlin, New York: Mouton de Gruyter. 109–132. Schmid, H.-J. (2000). English Abstract Nouns as Conceptual Shells: From Corpus to Cognition. Vol. 34. Topics in English Linguistics. Berlin, New York: Mouton de Gruyter. Searle, J. (1969). Speech Acts: An Essay in the Philosophy of Language. London: Cambridge U.P. Seto, K.-I. (2003). Metonymic Polysemy and Its Place in Meaning Extension. In B. Nerlich, Z. Todd, V. Herman, & D. D. Clarke (Eds.), Polysemy: Flexible Patterns of Meaning in Mind and Language. Trends in Linguistics. Studies and Monographs 142. Berlin, New York: Mouton de Gruyter. 195–216. Soriano, C., Fontaine, J. R. J. & Scherer, K. R. (this volume). Surprise in the GRID. Tutin, A. (2009). Les émotions sont-elles comptables? In I. Novakova & A. Tutin (Eds.), Le lexique des émotions. Langues, Gestes, Paroles. Grenoble: ELLUG, Université Stendhal. 65–79. Tutin, A. (this volume). Surprise routines in scientific writing: A study of French social science articles. Van de Velde, D. (1998). Cet obscur objet du désir. L’objet des verbes de sentiment. Travaux de Linguistique 35: 67–78. Van Dijk, W. W. & Zeelenberg, M. (2002). What Do We Talk about When We Talk about Disappointment? Distinguishing Outcome-Related Disappointment from Person-Related Disappointment. Cognition & Emotion 16 (6): 787–807.

Corpus data In Treatment American Series, 2008, directed and produced by Rodrigo Garcia,broadcast by HBO. . Corpus of American Soap Operas, created by Mark Davies, Brigham Young University. .

Index A affect  153–154, 156, 170, 192, 202–205, 216–217, 226 Affective Computing  2–3, 87, 197–198 appraisal  3, 124, 173, 176, 179, 185, 187–188, 190, 192, 203–205 arousal  3, 124, 139, 173, 177, 180–181, 184–185, 188–189, 198, 201 assertion  9, 66–68, 71, 84, 138–139, 220–222, 234, 239– 240, 242 C cognition  1–2, 30, 119, 123, 126 cognitive  1–3, 7–8, 10, 15–16, 20–24, 32, 36–37, 39, 41–42, 45, 52, 54–55, 64, 66, 73–81, 84, 86, 123, 125, 134, 148, 173–175, 217, 219, 225, 230–233, 235 concept  2–3, 7–11, 13, 19, 23–24, 27, 29–30, 32, 73, 93, 159, 174–175, 178–179, 182–188, 191, 202, 204, 218 conceptual  2, 7–14, 16, 18–24, 28, 30, 41, 46–47, 84, 91–93, 98–99, 102–103, 107, 109–111, 117, 153, 156, 170, 191, 230, 234 congruence  156, 166, 168, 170 constructions  2–3, 32, 35, 46, 71, 76, 81–83, 85, 91, 93–95, 97–100, 102–103, 105, 108, 111, 115–118, 160, 162, 164, 167–168, 215, 218, 222–225, 228, 241 control  2–3, 7, 18–24, 35, 44, 55, 59, 71, 78, 81–85, 122, 127–128, 186, 234–235, 237, 239–241 D dimension  1–2, 27, 39, 91–92, 95, 98, 102–103, 105, 107–108,

111, 114, 117–118, 122, 124, 128, 147–148, 154, 159, 166, 169, 173, 175, 177–178, 180, 184–185, 188–192, 198, 201, 203–204, 216, 220 discrepancy  61, 74–80, 123–124, 190, 235, 239–240 disfluency  4, 126, 236–237 disruption  36, 39, 41, 121, 123, 125, 127–128, 141, 143–144 see also disruptive E emotion  1–4, 7–10, 13–16, 18, 20–24, 28, 30–31, 36, 40, 45, 73, 76, 86, 106, 113–114, 123– 128, 130–132, 135, 139–142, 148, 153–157, 159, 162, 166, 168–170, 173–182, 184–192, 197–206, 209, 215–218, 222, 225–227, 233–235, 237, 239–242 emotional  1, 3–4, 14–15, 18, 20–21, 24, 28, 73, 122–124, 126–127, 132, 135, 139, 154–156, 169, 174–175, 177, 179–180, 182, 186, 188–192, 197, 202–203, 205, 207–208, 215–217, 221–222, 225–226, 229, 233, 235, 237–242 emotive  43, 96, 127, 129, 155, 216–217, 222, 225, 235, 238– 239, 241–242 epistemic  1–3, 23, 92–97, 99–100, 102–107, 109–110, 114, 117–118, 155, 161, 164, 215–216, 219, 227, 229, 233–235, 239, 241 see also epistemicity evaluation  37, 39, 46, 73, 76, 96, 99, 124–125, 129, 155, 177, 203–204, 209, 239

evidentiality  1–2, 51, 53–56, 58, 60, 62–64, 66–69, 70, 72, 81, 84–85, 118, 218–219 see also evidentials exclamation  4, 76, 99–101, 104, 113–114, 129, 133, 135, 139–140, 146, 220, 222 exclamative  77–78, 81–82, 94, 99–101, 104–105, 111, 126–127, 129, 133, 137–139, 222–224, 227 expectations  3, 52, 58–59, 74– 75, 77, 83–84, 92, 100, 103–105, 109–110, 112, 114, 116, 123–125, 127, 131, 141, 143–144, 156, 166, 168, 186, 209, 216, 219, 229, 233, 235–236, 239–241 expression  1–4, 9, 11–12, 15, 29, 35–36, 39, 41, 73, 76, 80, 93, 95–97, 99, 101–105, 110–111, 114, 121–123, 126–131, 133–138, 140–145, 147–149, 153–155, 164–166, 175–177, 179–180, 186–187, 198, 200, 203, 205– 207, 215–218, 220–221, 223, 225, 227, 230, 234, 236–237, 239, 241 expressivity  1, 3, 4, 15, 21, 23, 95, 99, 106, 114, 127, 131, 175–176, 186–188, 190, 192, 216–218, 220–222, 235 see also expressive F feeling  13, 16, 18, 32–35, 39–40, 42, 44, 46, 74, 76, 105, 132, 142, 161, 166, 170, 176, 179–180, 187, 225–226, 229, 233, 236 G gestures  1, 176 GRID  2, 20, 24, 173–174, 176– 180, 182, 184, 186–187, 189–192

246 Expressing and Describing Surprise

I incongruity  91–94, 96–99, 101–105, 107–117, 223–224 intensity  11, 102, 104, 121, 123, 127–130, 140–141, 143, 145– 149, 199–200, 204, 216, 230 interaction  3–4, 25, 95, 111, 117, 124, 127, 137, 154, 173, 197–198, 202, 204–210, 216 interjections  4, 9, 28, 46, 81, 113, 126–128, 131, 133, 137–138, 142, 218 K knowledge  2, 30, 32, 51, 54, 56–69, 72–73, 75, 77–78, 81–82, 84, 103, 116, 135, 143, 156, 168, 187, 223, 232–233 L lexeme  1–4, 8, 10–11, 13, 40, 94, 97, 100, 105, 126, 182–183, 191, 207, 215–218, 222, 225, 233–234, 237, 240–242 lexical  2, 7, 10–13, 17, 23–24, 27, 31, 40, 51, 60, 67, 81, 84–85, 94, 97–98, 100, 121, 154, 157, 159, 170, 178–179, 209, 220–221, 225 lexicon  2–3, 28, 46, 123, 125, 130, 139, 143, 154, 157–158, 169, 174, 190, 204–206, 216–217, 237 M metaphor  2, 7–8, 11–12, 17–23, 126, 187, 204, 209, 231–232, 235 metonymy  2, 7–8, 14–18, 21–23, 126, 187, 230–232 mirativity  1–3, 51–60, 62–64, 67–68, 72–73, 78–86, 91–95, 97–102, 103, 105–111, 114, 116–118, 190, 215, 218–220, 222–225, 241 see also mirative

N Natural Semantic Metalanguage  2, 16, 27–29, 32, 40, 46, 216 see also NSM novelty  3–5, 20, 34–35, 38, 51–52, 54–55, 57–59, 64–65, 68–72, 74, 76, 79–81, 85, 101–102, 119, 121, 124, 127–128, 131–132, 134, 136, 141, 145–146, 173–176, 178, 180–181, 184–185, 187–196, 209–211, 218, 226, 232 see also new P power  3, 98, 116, 173, 177, 180–181, 184–186, 189, 201 pragmatic  3–4, 54, 71, 81–82, 86, 100, 104–105, 108, 110, 190, 216, 221, 223, 225, 234, 241 psychological  2, 9, 11, 18, 20, 57–60, 64, 78, 86, 123, 150, 173, 179, 191, 202–203, 215–217, 222, 236, 241 psychology  2, 8, 30, 46, 73, 98, 157, 174, 176–178, 202 R rhetorical  2–3, 81, 153, 155–156, 169–170 S scenario  2, 18–19, 23, 37, 39, 61, 65–66, 68–69, 74, 80, 124, 153, 167–170, 207, 235, 239 scientific  2, 8, 30, 153–159, 161–171 semantic  2, 16, 27–32, 36, 41, 45–46, 56, 60–61, 81–83, 105, 129, 143, 147, 154–157, 159–160, 162, 164–165, 173–174, 177–178, 180, 182, 184–185, 192, 204–205, 216, 220, 223, 230–232, 241

semantics  27–28, 31, 41, 45, 54, 60, 78, 84–85, 94, 178, 217, 225, 230–232 sentiment analysis  202, 204, 209–210 source (of surprise)  144, 159–160, 164–165, 167–169, 215, 219, 224–232, 238, 241 stimulus  3, 36, 38–39, 99, 122, 124–125, 128, 130, 132, 139, 145, 148, 176, 216, 225–226, 241 U unexpectedness  2, 13, 18–20, 22, 27–28, 31–33, 35–36, 38, 41, 43, 53–54, 58–59, 61, 63, 65, 74–76, 78–79, 81, 83, 99–100, 104, 117, 122, 124–125, 128, 130–133, 140–141, 143–144, 146, 148, 154, 160, 162, 166, 175–176, 185, 190, 215, 218, 231 see also unexpected unpreparedness  51–52, 54, 64, 73, 78, 81, 190, 235 see also unprepared V valence  3, 121, 123, 125, 127, 130, 141, 143, 145–149, 173–175, 177, 180–181, 184–185, 189, 201–202, 208, 215, 237–239, 237–238, 241 see also valenced

Among emotions, surprise has been extensively studied in psychology. In linguistics, surprise, like other emotions, has mainly been studied through the syntactic patterns involving surprise lexemes. However, little has been done so far to correlate the reaction of surprise investigated in psychological approaches and the efects of surprise on language. This cross-disciplinary volume aims to bridge the gap between emotion, cognition and language by bringing together nine contributions on surprise from diferent backgrounds – psychology, human-agent interaction, linguistics. Using diferent methods at diferent levels of analysis, all contributors concur in deining surprise as a cognitive operation and as a component of emotion rather than as a pure emotion. Surprise results from expectations not being met and is therefore related to epistemicity. Linguistically, there does not exist an unequivocal marker of surprise. Surprise may be either described by surprise lexemes, which are often associated with igurative language, or it may be expressed by grammatical and syntactic constructions. Originally published as a special issue of Review of Cognitive Linguistics 13:2 (2015).

ISBN

978 90 272 4280 8

John Benjamins Publishing Company