Accepted Manuscript

How does it make you feel? a new approach to measuring emotions in food product experience Sara Spinelli, Camilla Masi, Caterina Dinnella, Gian Paolo Zoboli, Erminio Monteleone PII: DOI: Reference: To appear in: S0950-3293(13)00226-7 http://dx.doi.org/10.1016/j.foodqual.2013.11.009 FQAP 2709 Food Quality and Preference

Please cite this article as: Spinelli, S., Masi, C., Dinnella, C., Zoboli, G.P., Monteleone, E., How does it make you feel? a new approach to measuring emotions in food product experience, Food Quality and Preference (2013), doi: http://dx.doi.org/10.1016/j.foodqual.2013.11.009

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

HOW DOES IT MAKE YOU FEEL? A NEW APPROACH TO MEASURING EMOTIONS IN FOOD PRODUCT EXPERIENCE

Authors: Sara Spinelli1*, Camilla Masi2, Caterina Dinnella2, Gian Paolo Zoboli3 & Erminio Monteleone2

SemioSensory - Consulting in Communication, Sensory & Consumer Research, via

Baccheretana 70, 59011, Carmignano, Prato, Italy.

2

University of Florence, Dipartimento di Gestione Sistemi Agrari, Alimentari e Forestali

(GESAAF), via Donizetti 6, 50144, Florence, Italy.

3

Adacta International S.p.a., C.so Vittorio Emanuele, 122, 80121, Naples, Italy.

* corresponding author: Sara Spinelli tel. +39.340.5462245 mail: spinellisara@hotmail.it skype contact: sara.spinelli

Abstract: The objective of this study was to define an approach to describe the emotional profile specific to a product category able to solve some limitations of the current approaches. A multistep approach was applied to measure emotions related to consumer liking for a specific food product category, chocolate and hazelnut spreads, chosen as the first case of application. Three interrelated experiments were conducted: 1) Selection of products that spanned the most relevant sensory variation within the considered product category; 2) Development of a product specific questionnaire (EmoSemio) based on interviews conducted with a modified version of the Repertory Grid Method 1

28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54

and analysed with a semiotic approach. 3) Collection of liking and emotional consumer responses with EmoSemio and with EsSense ProfileTM questionnaires. Both questionnaires used to measure emotional responses produced information that is not captured by only measuring acceptability. However results from ANOVA model applied on emotion ratings highlighted important differences between the two questionnaires. The product specific questionnaire was found to discriminate across the products more effectively, with a higher percentage of discriminating emotions and a higher number of sample groups discriminated by each emotions (LSD99% post hoc test). Different factors contributed to these results: (a) the product-specific and language/culture-specific nature of the questionnaire; (b) a different and clearer way to express emotions in EmoSemio: not using single adjectives but full sentences helps to reduce ambiguity; (c) a reduced length - 23 instead of 39 items. For these reasons, EmoSemio approach seems to be appropriate when the emotional profile of a specific product category is of interest, allowing a fine-grained analysis with relatively modest costs as to the benefits (25 interviews). Further studies are needed to experiment EmoSemio on other product categories, testing its reliability and suitability with different food and also non-food products.

Keywords: Emotion Repertory Grid Method (RGM) Semiotics Sensory Analysis EsSense Profile Language

55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81

1. Introduction

1.1 The role of emotions and other affective phenomena in consumer perception Understanding the motivations that drive consumer choices is one of the major challenges for marketing and consumer studies. Emotions can play a leading role in product experience and the importance of the optimisation of products from both sensory and emotional perspectives has been recently emphasised (Thomson, 2007). Since measurement of the acceptability alone is not a sufficient benchmark for product development and testing (King & Meiselman 2010; Thomson 2007), recently several studies have investigated the relationships between sensory characteristics and emotional responses comparing different product categories (King & Meiselman, 2010; Cardello et al., 2012) or within a specific product category, such as dark chocolate (Thomson, Crocker, & Marketo, 2010), blackcurrant squashes (Ng et al., 2013a) and softeners (Porcherot et al., 2013).

These studies focused on different affective phenomena that do not consist only of emotions; the EsSense ProfileTM includes emotions but also diffuse affect states such as moods, characterized by a relative enduring predominance of certain types of subjective feelings (e.g. loving or affectionate; King & Meiselman, 2010), while the conceptual profile includes a mix of emotions and abstract conceptualisations with emotional connotations (e.g. masculine or sensual; Thomson et al., 2010).

Among psychologists, there is no universal agreement about exactly what an emotion is (see the reviews in Galati, 2002; Frijda, 2008; Scherer, 2005; Frijda & Scherer, 2009; see also the discussion about a working definition of emotion in Mulligan and Scherer, 2012). However, there is now a rather widespread acceptance that they have 3

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108

multiple components, including physiological arousal, motivation, expressive motor behaviour, action tendencies and subjective feeling. Emotions are characterised by a response synchronisation (they prepare appropriate responses to an event that disrupts the flow of behaviour), rapidity of change (they continuously readjust to changing circumstances or evaluations), behavioural impact (they prepare adaptive action tendencies), high intensity and relatively short duration; for these reasons, emotions can be distinguished from other affective phenomena such as preferences, attitudes, moods, affect dispositions and interpersonal stances (Scherer 2005).

However, it is clear that not only emotions, but also other affective phenomena are of interest for consumer sciences. In turn, this allows us to go beyond the problem of the scientific definition of the concept: product perception is mediated not only by the emotion that is elicited by the product at the moment, but also by the preferences and the affective disposition of the subjects, their moods and attitudes and also by the feelings associated with the product in the mind of the consumer. In fact, emotions can be associated with a product by the brand or elicited by specific sensory properties that characterise it (e.g. sweet makes one feel happy, etc.), and especially by odours. A large number of studies have pointed out the tendency of odours to elicit emotions, suggesting a close relationship between olfactory and affective information processing (Chu & Downes, 2002; Herz, 2002; Herz et al., 2004; Herz & Schooler, 2002; Soundry et al., 2011).

1.2 Emotion questionnaires in sensory and consumer studies There are a number of reasons why well-established questionnaires developed in clinical psychology are unsuitable for measuring emotions associated with consumer products. Firstly, such questionnaires are typically focused on negative emotions while, in commercial product experience, positive emotions are predominant (Schifferstein & 4

109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135

Desmet 2010) and, secondly, because many of the terms included in psychological scales are not considered relevant by consumers to describe the emotions elicited by the tested product (King & Meiselman, 2010; Delplanque et al., 2012). Several marketing studies have tried to develop a set of descriptors (a lexicon) that should represent the full range of emotions that consumers most frequently experience in consumption situations (Richins, 1997; Laros & Steenkamp, 2005; Thomson & Crocker, 2013). In recent years, there have been concentrated efforts to develop standardised questionnaires to measure emotions in a product development context. GEOS (Geneva Emotions and Odor Scale) was developed to specifically study emotions associated with odours (Chrea et al., 2009; Porcherot et al., 2010) and recently was applied to actual products (Porcherot et al., 2013). The original GEOS consisted of 36 adjective emotional terms, but the new version of the method, ScentMoveTM (Porcherot et al., 2010 and 2012; Delplanque et al., 2012), consists only of six items, each labelled by a phrase identifying a class and illustrated by three words (nouns and adjectives). The EsSense ProfileTM (originally developed by King & Meiselman, 2010), which has been applied to both food names and food products, employs a list of 39 emotions and mood presented as adjectives, only 3-5 of which are negative. The questionnaire was validated using different food categories for its discriminating power but few validated data are available to evaluate its application in a commercial context within the same product category (Cardello et al., 2012; Ng et al., 2013a and 2013b).

In the last few years, other approaches have appeared aside from the standardised questionnaires, with the aim of developing a reduced list of emotions the most suitable for a specific product category (Ferrarini et al., 2010; Thomson et al., 2010; Ng et al., 2013a). Both of these approaches, standardised and product-specific, boast some advantages but also have some limitations. Standardised questionnaires are 5

136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162

cheaper and easier to use than product-specific ones, but they can be less discriminating because of their general nature (Ng et al., 2013a). Such questionnaires also need to include many items to be certain not to miss important emotional dimensions, but that can have negative consequences on the respondents, such as fatigue or boredom (Ng et al., 2013a; Jaeger & Hedderley, 2013; Jaeger, Cardello e Schutz, 2013) ). For this reason, it is suggested not to use EsSense with more than two samples (King, Meiselman & Carr, 2013). On the other hand, ad hoc questionnaires reveal more about the product specificity and can furnish more finegrained analyses, but they usually require substantial preliminary work including interviews, focus groups, term selections - to develop the questionnaire.

1.3 Language and emotions: background and motivation for the research Apart from the procedure chosen to develop questionnaires - standardised or an ad hoc - all these approaches do not really differ in the final format. Each presents to respondents a previously defined list of adjectives or nouns to select and/or rate in order to describe their emotional experience of the product. Adjectives are commonly recognised as the suited labels to indicate emotions and they are usually preferred to nouns in the studies because they seem to be more easily associated with immediate emotional experience (Plutchik, 1980). However, the emotional lexicon includes also verbs, adverbs, nouns, and interjections (Majid, 2012; Gius, Cozzi, & Spagnotto 1992).

What is not generally considered is that there is not a strict correspondence between emotions and the words used to indicate them. Different phenomena have to be considered, such as the slicing discrepancy, which is due to the fact that emotional structures include more components than those denoted by natural languages. For instance, English, Dutch and Italian do not have different words to indicate different 6

163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188

kinds of fear (Frijda & Zammuner, 1992) and Italian does not have different words to indicate the sadness characterised by rage or the quiet/sadness of boredom (DUrso & Galati, 1990). Often, in absence of a specific word, emotions are expressed using a sentence that paraphrases the meaning.

The problem with many emotional words so is that each has multiple and thus ambiguous meanings, depending on the contexts on the individual experience of each speaker (Kagan, 2007). In each text, in fact, the meaning of a word is selected by the context defined by the other words that surround it in a sentence and by the situation in which the sentence is included: the topic of the text fixes which semantic properties are activated and consequently have to be considered and which of them could potentially be activated, but they were not in that context (Eco, 1979; 1990). Thus, it should be considered that words need a context to be interpreted correctly, that is to say in this case to be interpreted in the way the researcher expects that they should be interpreted.

Questionnaires are particularly sensitive to this problem of ambiguity in wording (Belson, 1981). Jaeger, Cardello and Schutz (2013) pointed out the problem of lack of understanding and misunderstanding in the EsSense emotion list and emphasized the absence of a meaningful context that could help to reduce ambiguity. Presenting emotions organised in groups and not in a unique list has been tried as a way of addressing this problem (ScentMoveTM: Porcherot et al., 2010; Geneva Emotions Wheel: Scherer, 2005). Such a choice can be useful to help the respondent to better understanding the task by clarifying the emotional area indicated. However, the use of complete sentences may provide greater clarity because of the inclusion of a context, in both a linguistic and situational sense. For these reasons, this work is aimed at

189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213

developing a questionnaire where the emotions are presented by using complete sentences and not single words.

1.4 Scope of the present work This work aims to elaborate a protocol in order to develop a product-specific questionnaire able to solve some limitations of the current approaches and where the ambiguity of emotional words is reduced as much as possible. In particular we want to: (a) determine if it is possible to work out an efficient and repeatable procedure to develop a questionnaire specific to a product category based on free consumer responses; (b) verify if complete sentences perform better than adjectives in measuring emotions using a questionnaire. The underlying hypothesis is that by including context, the meaning will be clearer and the risk of ambiguity will be reduced; (c) evaluate if a predetermined questionnaire such as the EsSense ProfileTM developed for and with English speakers can be used with different cultures, in this case with Italian consumers; and (d) study the relationships between emotions and the sensory drivers of liking in the case of a specific product category.

2.

The experimental design

A multistep approach was applied to measure emotions1 elicited by a specific food product category - chocolate and hazelnut spreads. Three inter-related experiments were conducted: EXP. 1) Selection of products that spanned the most relevant sensory variation within the considered product category; EXP. 2) Development of a product specific questionnaire (EmoSemio), based on the identification of emotions that consumers associate with these products.

Note that the term emotion is used in this paper to indicate subjective affective feelings, being aware that they may not be strictly emotions (see 1.1). 8

214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240

EXP. 3) Collection of liking and consumer emotional responses with the EmoSemio and EsSense ProfileTM questionnaires and investigation of the relationships between liking, and sensory and emotional profile within the considered product category.

3.

Experiment 1: Product selection

The aim of this preliminary study was to select products representative of the whole range of sensory variability within a set of chocolate and hazelnut spreads.

3.1 Methods

3.1.1 Participants Subjects were recruited in Florence area (Tuscany, Italy). Fifteen subjects, 6 males and 9 females with an average age of 26, regular chocolate and hazelnut spreads consumers, were selected. Subjects were compensated with a gift token for their participation in the study.

3.1.2 Products Nine products representing the most common chocolate and/or hazelnut spreads available in the Italian market were chosen (codes: B, C, G, L, N, P, R, S, T). A further product, a milk-chocolate cream cheese spread, considered an innovative product, was selected too (code: E).

3.1.3 Procedure Assessors participated to three term generation sessions. In each session, three products were presented. During these sessions, spreads were described according to their appearance, aroma (odour by nose), flavour and mouthfeel characteristics. For each descriptor, the panel reached a consensus on a definition and, when necessary, 9

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267

on a standard. The consensus-building process, managed by the panel leader, ended with the list of attributes reported in Table 1. In order to train subjects to rate the intensity of each attribute, two sessions were run.

The evaluation of the nine chocolate and/or hazelnut spreads were organised in three sessions. In each session all samples were presented and the presentation order was balanced for first order and carry over effects across assessors. Samples were presented in three subsets and each sample was evaluated in triplicate. The innovative sample E, the only one characterized by the yogurt attribute, was always evaluated in separate sessions in all the phases of the descriptive analysis described above.

For each sample, subjects received a 40 cc transparent plastic glass containing 20 g of spread. Following the presentation order, subjects were asked first to smell a sample and rate the aroma attributes. Then they were asked to fill the spoon (5 ml) for a 1/3 with the sample, to take it into their mouth for some seconds, to swallow and to rate the mouthfeel attributes. The procedure was repeated to rate flavour descriptors. Between samples evaluation, assessors were asked to rinse their mouth with purified water, to have some white bread and then to rinse again their mouth with water. The evaluations were performed in individual booths under red light in order to eliminate visual clues. For the evaluation of the appearance attributes, the samples were then re-presented in the same set under white light. Ratings were collected by FIZZ (version 2.40.G; Biosystmes, Courtenon, France).

10

268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
3.3 Results and discussion Results of the PCA computed on descriptive data are summarized in the correlation loading plot reported in figure 1. The first two significant dimensions of the perceptual map accounted for 80% of the variation (PC1: 49% and PC2: 31%). PC1 was positively associated with flavour descriptors bitter, cocoa and hazelnut, while a negative correlation was found for buttery, sweet and milk chocolate flavour 11 3.2.2 Sensory differences among products Sample differences for each attribute were assessed by a two-way, mixed model ANOVA using assessor as random factor, while sample was the fixed factor. Differences and similarities among the nine chocolate and hazelnut spreads resulting from descriptive analysis were studied by means of Principal Component Analysis (PCA) (Unscrambler version 10.1, Camo). A PCA model was computed on panel averages of each significant attribute (p0.05) arising from the ANOVA model. Samples were included as dummy variables (down-weighted in the data matrix) to improve the visual interpretation (Martens & Martens, 2001). The full cross validation was computed to validate the interpretation of the first two components. 3.2.1 Panel Performance Intensity data from the trained panel were analysed by multiblock PCA (Tucker-1) and by pMSE plot (Panel Check software, ver 1.4.0, Nofima, Norway) to assess panel calibration and assessor performance, respectively (Ns et al., 2010). On the basis of the pMSE plots, three out of fifteen subjects were considered unreliable and were taken out from further data analysis. 3.2 Data analysis

295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321

descriptors. Sample L was separated from the rest of the samples along PC1 and showed a positive correlation with bitter and cocoa attributes. PC2 contributed to separate samples N, B and C, associated with vanilla and opposed to product T, associated with the mouthfeel descriptor stickiness. Furthermore, along the second dimension it was possible to note the opposition of sample G, associated with the attribute thickness, to sample P, associated with the appearance descriptors colour and brightness. The relative positions of the samples on the perceptual map resulting from descriptive data were used for the selection of samples to be tested for their emotional profile. Samples L, B, P, G, T spanned the relevant variability of the sensory attributes within the whole sample set used to represent the chocolate/hazelnut spreads available in Italy. Furthermore, sample E was included in the selection because of its sensory diversity and innovativeness in the chocolate spread category.

4. Experiment 2: Development of a product-specific questionnaire (EmoSemio)

A product-specific questionnaire was developed using one-on-one interviews conducted with a modified version of the Repertory Grid Method (RGM). The originality of the method was twofold: (a) the design of interviews and (b) the use of semiotic analysis to interpret gathered information and to develop the questionnaire. We will refer to the approach to develop the product specific questionnaire and consequently to the questionnaire in itself with the name EmoSemio.

4.1 Methods

4.1.1 Products 12

322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348

The six chocolate and hazelnut spreads selected in experiment 1 were used.

4.1.2 Participants Seventy-five Italian mother tongue subjects (38 females and 37 males), all regular consumers of chocolate and hazelnut spreads (at least once a week), were recruited from Bologna and surroundings (Italy) to participate in the study. Subjects were assigned to three consumer panels (1; 2; 3) of 25 subjects each, equally distributed for gender (38 females and 37 males) and age (from 25 to 45 years). Subjects were compensated with a gift token for their participation in the study.

4.1.3 Interviews The objective of the interviews phase was to access and explore the whole gamut of personal constructs associated with chocolate and hazelnut spreads. Several studies suggested that comparative techniques perform better that direct questioning in that task (Reisenzein & Hofmann, 1990; Thomson, 2007). Repertory Grid Methodology (RGM) has proven to be very efficient in collecting information on consumer perception of food products and, recently, it has been applied to develop a questionnaire to measure emotions in commercial contexts (Ng et al., 2013a).

In the present study, a modified version of RGM was used to empirically determine emotions perceived as relevant by individuals in experiencing the products. One-toone interviews were conducted. Interview conditions varied across panels. Subjects from panel 1 were presented with the samples in blind conditions (samples of 20 g were presented in 50 ml transparent glasses). The presentation order was balanced across subjects. Subjects were asked to taste the samples and rank them according to decreasing preference. Then, the interviewer divided the samples into two triads (samples ranked 1-3 and samples ranked 4-6). For each triad, each subject was asked 13

349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

to concentrate on the emotions felt during tasting and to describe how each product makes them feel, compared with the other two. When the respondents were found to be reticent or unclear, they were asked to explain what they meant exactly.

The same procedure was adopted to interview subjects from panel 2, but in this case respondents were provided with an aid, in the form of a list of words associated with emotions. Respondents were asked to choose the words (one or more, if any) in the list that best described how each product made them feel. They were also asked to explain why they choose that word and in which sense they intended the word. In this way, respondents were encouraged to talk about their emotional experiences using the list as a starting point, as a way to break the ice. The aid list included 22 words related to emotions presented in two columns (positive and negative) in alphabetical order. The list was developed selecting words from scientific literature of food studies on emotions, but it also included words derived from general emotion studies. The words are presented in an adjectival form, referred to the subject (e.g. relaxed) or to the product (e.g. amusing) to verify the more effective way in which they should be presented in a questionnaire. The reasons for the introduction of the list were twofold: first, to verify if consumers needed or had advantages in receiving help to start talking about such a private dimension as emotions; second, the list was used to test how ambiguous words commonly included in the questionnaires can be, verifying how consumers interpret these words presented without a specific context (i.e. in a list). The aim was to collect information in order to develop a questionnaire as clear as possible to measure emotional responses.

Subjects from panel 3 received the products in their original packaging. They were asked to rank the products according to their preference without tasting them. Then

14

375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401

they followed the procedure of both panel 1 and 2, describing how each product made them feel, first without the list and then with the help of the list.

The interviewer took note of the consumers answers. Moreover, to help the analysis all the interviews were audio-recorded.

4.2 Interview analysis A semiotic methodology was used to analyse consumer responses in one-to-one interviews to determine which emotions respondents considered relevant to describe their product experience. Semiotics has a long tradition in advertising and communication analysis (Floch, 1990; 1995) and has developed various approaches to research in marketing (see the review in Nth, 2006); it is currently used to study brands, advertisements and consumer styles and recently it was applied in storytelling and consumer food studies (Ares et al., 2011; Piqueras-Fiszman et al., 2011; Smith et al., 2010). The semiotic approach decomposes the texts (in this case, the interviews) in order to deeply investigate their meaning by identifying the semantic units in the text. The words or expressions referring to similar meaning are grouped together and recognised as belonging to the same semantic category. Then, the inter-relationships (e.g. oppositions) between the different semantic categories are investigated (Greimas & Courts, 1979; Hbert, 2007).

4.3 Results and discussion

4.3.1 Interview analysis and semantic category selection As a result of the semiotic analysis, 23 semantic categories from the emotional/affective field were selected to be included in the questionnaire. The rationale behind this selection was that the questionnaire had to be as exhaustive as 15

402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426

possible for the specific product category, and repetitions and possible sources of ambiguity should be avoided. Because semiotics allows a deep analysis of meaning, polysemic words (words that can have different meaning depending from the context) were disambiguated allowing a better understanding of consumers responses. The same word used across the subjects with a different meaning were avoided, and classified in different semantic categories. For instance, the Italian word tranquillo was used both in the sense of calm (positive valence) and in the sense of indifferent, boring (with a less positive valence). Furthermore, thanks to the grouping of words with similar meaning, repetitions in the questionnaire were avoided. Moreover, words currently used to describe an emotion were not considered when subjects used them to metaphorically describe the sensory properties of the products (e.g. daring).

The importance of different semantic categories was studied within each panel and across them. In general, the categories that were found to be more important in describing products were those related to tension reduction/relaxation, satisfaction, feeling full of energy, positive memories and reassurance. The panel that evaluated product packaging used some different categories that were included in the selection (feeling good/generous and sensuality); this was done because the questionnaire was developed to be used with products presented both in blind and afterwards branded conditions2.

The analysis of the interviews conducted with the list (panel 2) gave information that help to develop a questionnaire avoiding words that could be misunderstood by consumers. Moreover, the results obtained with the list and without it were compared.

The comparison of emotional profiles in blind and branded conditions is not discussed in this paper.

16

427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453

Our findings showed that the use of the list added no category to the ones cited by the respondents without any help (but there was an increase of the number of the emotions mentioned); consequently, the list was not indispensable, and motivated respondents can complete the task without difficulties; on the other hand, the list could be a useful tool with particular subjects who have difficulties in talking about such personal aspects of their being, like emotions. For this reason, in further studies interviews could be conducted first without a list (panel 1 procedure) and, immediately after, again with the help of the list, to verify if additional information can be added. This could be a good solution to both avoid influencing the interviewee (because in the first moment the respondent is free to answer what comes to mind) and also to guarantee that important items would not be omitted. This concern was in line with that of Ng et al. (2013a) who suggested a hybrid method to develop the questionnaire, using a list partly defined by the consumer and partly selected from the literature, to avoid the omission of important discriminating emotions in the development of a product-specific questionnaire.

4.3.2 EmoSemio questionnaire development In order to reduce ambiguity within the questionnaire as much as possible, the 23 selected semantic categories were translated into full sentences that could answer the question How does it make you feel? (Table 2). The use of sentences instead of isolated adjectives and nouns was supposed to be advantageous in making clearer the indicated emotional state. In fact, including a linguistic context helps in determining the meaning of a word, thus reducing ambiguity (see 1.3). Using full sentences, in fact, it was possible to specify the emotion for a better understanding by respondents. The semiotic analysis of interviews showed that relax is used by respondents with two meanings. For this reason, the questionnaire included two different sentences where a context helped to clarify the meaning of the emotion we asked to evaluate: 17

454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480

It is an antistress: it calms me, soothes me, reassures me referred to a situation where the product acted as an active agent, able to inspire a passage from a negative state of uneasiness and agitation (a stressful state) to a positive mood characterised by more serenity. The emotion described with this sentence was different from that described in the sentence It makes me feel relaxed, which referred to an emotional state of relaxation and didnt necessarily imply a passage from a negative to a positive state.

4.3.3 Relationships among emotions included in the questionnaire The questionnaire was developed to map the semantic space for the subjective experience of emotional responses to chocolate and hazelnut spreads. It included 16 positive and 7 negatives emotions. Negative emotions were included because they resulted relevant for consumers (interview phase), despite the fact that usually they are eliminated from questionnaires on food products on account the phenomenon of the hedonic asymmetry (Schifferstein & Desmet 2010; King & Meiselman 2010). We considered that this result was of interest for product developers as a way of gathering information about a negative performance of a product (e.g. there is a difference in the diagnosis between a low liking score correlated to indifference rather than to disappointment). Fig. 2 is an intuitive visual representation of the relationships among the emotions included in EmoSemio that can help to bring into focus their similarities and differences; thus, it is clear that it is not valid in general, but only for the considered product category (it is the result of interview analysis).

Emotions that share a portion of their meaning are closer than emotions that do not share it or share just a little of it (semantic distance). Emotions that are grouped in the same set share an important feature: for example merry, energetic and amused share dynamism and active feeling, while feeling relaxed, reassured and amusement 18

481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507

share the feature of the reduction of tension. Emotions are opposed for their valence along the horizontal orientation in positive (on the left) and negative emotions (on the right): the arrows indicate these relationships of opposition. Moreover, along the vertical dimension, emotions are opposed for their degree of activation: high activation on the top and lower activation on the bottom for positive emotions, and the contrary in the case of negative emotions.

5. EXPERIMENT 3: COMPARISON OF EMOSEMIO WITH THE ESSENSE PROFILE TM AND

VALIDATION OF THE METHOD

The goal of this experiment was (a) to verify the effectiveness of the developed questionnaire EmoSemio to study emotions in products perception and their relations with sensory drivers of liking and (b) to compare it with the published predetermined questionnaire EsSense ProfileTM.

5.1 Methods

5.1.1 Products The six chocolate and hazelnut spreads selected in experiment 1 were evaluated.

5.1.2 Participants Two hundred and thirty-eight consumers aged from 25 to 45 years participated in the study, equally selected from the area of Florence and Bologna, 50% male and 50% female. As suggested in previous studies (King & Meiselman, 2010), they were all frequent users of chocolate and hazelnut spreads. Participants were compensated with a gift token for their participation in the study.

19

508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534

5.1.3 Experimental design and procedure This experiment consisted of two studies. In the first study, 120 subjects (panel A), equally balanced in terms of city, age and gender, were presented with the EmoSemio questionnaire developed in experiment 2. In the second study 118 subjects (panel B) were presented with the EsSense questionnaire, translated for the occasion in Italian. In both studies, consumers tasted each product presented monadically in blind conditions (labelled with an alphabetic letter) in a balanced order. For each sample, subjects received an 80 cc transparent plastic glass containing 20 g of spread presented at room temperature. Subjects were asked first to smell and then to fill a spoon (5 ml) for a 1/3 with the sample, to take it into their mouths for a few seconds, to swallow and to score their liking with a 9-point scale (anchored from dislike extremely to like extremely). Then consumers were asked to taste the sample again and rate their emotional responses using a specific questionnaire (panel A: EmoSemio; B: EsSense). Both panels used a 5-point scale anchored from not at all to extremely as suggested for the EsSense ProfileTM by King and Meiselman (2010).

Before passing onto the following product, participants were asked to drink some water, to taste some bread (Panbauletto, Barilla) and to drink again in order to clean the palate. Both questionnaires were introduced by the question How does it make you feel?. Emotions were presented in a balanced order across subjects in both experiments. The session lasted on average 20 minutes for the EmoSemio panel and 35 minutes for the EsSense panel.

5.2 Results and discussion

5.2.1

Comparisons between the EsSense ProfileTM and EmoSemio

methodologies 20

535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561
5.2.1.2 The discriminating power of emotions Data relative to each EsSense and EmoSemio emotion were submitted to a two-way ANOVA (factors: products and subjects). A significant sample effect (p< 0.05) was found for 21 items (corresponding to 91.3 %) in EmoSemio, while in EsSense 33 items differed (corresponding to 84.6%). For only 2 out of the 23 emotions included in the EmoSemio questionnaire - guilty and surprise - a significant sample effect was not found (Table 3). The non-discriminating emotions included in EsSense were 21 Liking data were further analysed by means of an Internal Preference Map (IPM). For the purpose, a Principal Component Analysis (PCA) was computed on individual liking scores. Results are summarized in figure 4. Visual inspection of the correlation loading map clearly showed that most of the consumers were allocated on the right side of the first dimension of the IPM, while they were widely spread along the second dimension. For the aim of this paper it is important to note that there were no differences in the position of subjects from both panels (EsSense and EmoSemio) on the map, indicating that the structure of the liking data resulted from the two studies was the same; thus, the results of the two different emotion questionnaires could be compared. 5.2.1.1 Overall liking scores A two-way ANOVA (factors: products and subjects) was carried out on liking scores for the 6 products in all subjects (n = 238). A LSD post hoc test was applied to estimate the differences among products. Significant differences were found in consumers overall liking for the products (p<0.001). G was found to be the most liked product, while T, L and B were moderately liked. P was less liked and E resulted to be the least liked product (fig. 3).

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588

aggressive, eager, guilty, nostalgic, tame and wild (Table 4). Similar results were found in the study on blackcurrant squashes conducted by Ng et al. (2013a), where six emotions were non-discriminating, four of which are also featured in our study (aggressive, guilty, nostalgic, wild). This finding could suggest than these emotions are not suitable to characterise the sensory experience of such products.

A LSD99% post hoc test was used to estimate significant differences among mean scores. The resulting number of sample-groups discriminated by each emotion was different in EmoSemio (up to 5 sample groups) and EsSense (up to 3 sample groups). In particular, some emotions were found to be very discriminating in EmoSemio, e.g. feeling cuddled and calmed down (anti-stress function), as well as the emotions associated to satisfaction, gratification, pleasure, affection, dynamism and tension reduction, all had several distinct subgroups of products (4-5 discriminated sample groups); only one negative emotion (annoyed) discriminated as well as these positive above-mentioned emotions (Table 5).

The two questionnaires (EmoSemio and EsSense ProfileTM) did not include the same emotions. A perfect meaning correspondence was found only in a few cases (happy, merry, satisfied), while some emotions included in both questionnaires shared a part of their meaning even if they did not indicate equivalent concepts: e.g. disgusted indicates a negative emotion of repulsion that usually implies feeling annoyed, but the two expressions have different meanings (Table 5). Nine of the 23 emotions in EmoSemio did not have a correspondent in EsSense: among those, cuddled, gratified, disappointed, amused and sensual were found to be very discriminating. Moreover, EsSense did not only miss several discriminating emotions, but even when it included some correspondent items, they usually discriminated less well than those in EmoSemio (e.g. satisfied, happy, merry, tender, energetic). 22

589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615

On the other hand, some items included in EsSense but not in EmoSemio discriminated among products, even if they had relatively low F-values. These were words that seemed to indicate moods, attitudes, personality traits and interpersonal stances (e.g. free, whole, good, polite, mild, understanding, friendly) more than they did an emotion (see Scherer 2005 for definitions of the different affective phenomena).

One emotion expressed in an adjective form was perceived as inappropriate by respondents, while the description in a discursive form was well understood. While nostalgic (in EsSense) was non-discriminating, the emotion expressed by the sentence I associate it to happy memories of childhood (in EmoSemio) was shown to be very important in characterising the chocolate and hazelnut spread experience, especially in the case of product G. The low F value of nostalgic suggested that participants did not understand this word. Actually, in Italian, this adjective (translated nostalgico; see Scherer 1988) is not frequently used, and the noun is usually preferred (I am filled with nostalgia for instead of I feel nostalgic for). A similar result was found also by Manzocco et al. (2013) in an Italian adaptation of EsSense, supporting the idea that Italian consumers find this adjective inappropriate to describe their emotional responses to food.

5.2.2 Relationships among liking, sensory profiles and emotional profiles

5.2.2.1 Emotional profiles: general results Not surprisingly, positive emotions were positively correlated with liking ratings, while negative emotions were negatively correlated with liking. Emotions associated to a negative experience were annoyed, disappointed, sad, bored, neglected and indifferent in EmoSemio, while in EsSense these were disgusted, worried and bored (the three emotions indicated as negative by King & Meiselman, 2010). 23

616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641

While emotions opposed on the valence dimension (positive-negative) were not associated with the same product, emotions opposed on the activation dimension (low-high) can be associated with the same product. This means that emotions characterised by different levels of activation (e.g. emotions associated to the tension reduction and the dynamism areas) were found to be closely connected: the mean values of relaxed, anti-stress and energetic are similar for each product, supporting the idea that a chocolate and hazelnut spread can evoke a sense of well-being that consists both in an energetic push and in a relaxing effect (Table 3). Similar results could also be found in the EsSense ProfileTM between active, energetic, and calm, peaceful, quiet (Table 4).

EmoSemio emotional profiles of the six products are shown in fig. 5. Differences between products are visible: product G is equally associated with happy memories and the area of tension reduction (anti-stress, relaxed) and dynamism (amused and energetic). Product T and L have very similar emotional profiles, at an intermediate level between product G and products B and P. Product E is characterised by an opposite emotional profile: except from surprise and curiosity the product is characterised by negative emotions, primarily disappointed, indifferent and sad. In fig. 6 the EsSense profile for the six products is shown. The most preferred (G) and the least preferred (E) are clearly separate to the others, but the EsSense profiles of T, L, B and P appear more similar than EmoSemio.

5.2.2.1 Comparisons between products with different liking ratings

Our findings show that in the case of products that differ in mean liking scores, emotions provided information that helped to explain this difference more deeply.

24

642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668

Products T and P were characterised by a significantly different sensory profile and a different liking rating (fig. 3). Product T, characterised by a higher hazelnut flavour and a more grainy and thick texture, was significantly more appreciated than product P, characterised by a higher milk-chocolate and vanilla flavour and a more intense colour. The EmoSemio results showed that the two products were evaluated as significantly different for 19 emotions; each positive emotion was evaluated as more intense in product T (except for surprise and curiosity), while some negative emotions were higher in product P: product P was rated more disappointing and annoying, and was more associated with sadness than product T. On the contrary, no significant differences in terms of emotional profile were found, either negative or positive, using the EsSense ProfileTM, with the exception of two emotions: products P was perceived as more disgusting and as something that makes one feel more eager. It must be noted that the Italian word chosen to translate eager (avido) can have a negative connotation more so than the English word and this could be the reason for this association with the product. According to the EsSense findings, then, two products very different for liking and sensory profile were not associated with different positive emotions and their emotional profiles were quite similar (fig. 7).

5.2.2.2 Comparisons between products that do not differ in liking

Some emotions were discovered to be more discriminating than liking, despite the fact that emotional ratings were only made up using a 5-points scale. That means that for this product category, emotional measures were able to provide increased product differentiation compared to the hedonic responses. These results are in line with Ng et al. (2013a). Product G was the most preferred product in the group, while product E, a new product with very distinctive sensory characteristics such as vanilla, yogurt, sour and 25

669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695

bitter flavour was the least preferred (fig. 3). Consequently, emotions associated with the two products were always positive for G and mainly negative for E. In spite of this general result, there was a group of consumers (46 out of the EmoSemio panel and 39 of the EsSense panel) that expressed a blind liking for the product higher than like slightly (6). For these subjects, no significant differences in liking were found between G (the product leader) and E, but the two products differed in the associated emotions. EmoSemio results showed that product G was more associated to happy memories than E, it had the ability to make one feel more generous and it was perceived as something that can have a stronger anti-stress function; on the contrary, product E was evaluated as being more associated with surprise and curiosity (fig. 8). This information could be very important in the development of a brand image associated to sensory characteristics for a new product.

A noticeable difference could be observed comparing the results of EmoSemio with the EsSense ProfileTM. In EsSense, the only emotion that was higher for E than for G was worried (fig. 8). This information seemed misleading because this result made evident that product E was perceived differently from G only for its association with a negative emotion, even if it was liked. This could be due to the fact that the two positive emotions higher in product E in EmoSemio (curiosity and surprise) were not included in the EsSense and worried was the most similar emotion: in fact to be worried implies to be interested in something that evokes a negative emotion (such as anxiety, uncertainty, doubt), while to be curious implies being interested with a positive connotation.

5.3 Discussion This paper detailed the steps needed to develop a questionnaire to measure emotions in a commercial setting based on consumers responses for a specific product category, 26

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721

chocolate and/or hazelnut spreads. The procedure followed, based on interviews conducted according to a modified Repertory Grid technique coupled with semiotic analysis, allowed for the development of a questionnaire, EmoSemio that is product specific, shorter than a standardised one, language and culture specific and presented in a discursive form, by using full sentences rather than single words. The present research directly compared the discriminating ability and the effectiveness of EmoSemio with the sets of terms translated in Italian by the EsSense ProfileTM (King & Meiselman, 2010). Both questionnaires gave information not captured by only measuring acceptability. However, the evidence strongly suggests that the set of sentences proposed by EmoSemio outperformed the list of adjectives used by EsSense. In fact, although EsSense includes more items, it seems that some of these are redundant for Italian speakers (e.g. quiet/calm/peaceful) and that some important items for this specific product category are missing (such as surprise and curiosity, or the emotions with a negative valence). This fact is not surprising and it is likely to be due both to the use of a more focused consumer language related to the specific product category and to the advantages of a questionnaire originally developed in Italian rather than one translated from English (in a test with Italian consumers). However, results from this study suggest that even when the emotions to rate are shared between the questionnaires and expressed with a very similar lexicon, the discriminating power is higher in EmoSemio than in EsSense. It can be reasonably supposed that the use of full sentences can be clearer than isolated words and able to reduce ambiguities, at least in the Italian language (see also DUrso, Cavicchio & Caldognetto, 2005). This evidence is in line with some studies that noted that the emotional lexicon includes words that indicate an emotion only if included in a feeling context and not a being one: for example the word guilty does not refer to an emotion in the sentence Im guilty of something but it indicates an emotional

27

722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748

state in the sentence I feel guilty (Ortony, Clore, & Foss, 1987 and Clore, Ortony, & Foss, 1987). Moreover, our finding suggests that the influence of a long list of items instead of a shorter one on performing the task should be studied carefully (see also Jeager, Cardello, & Schutz, 2013). Decisions regarding the length of the questionnaire should be taken considering the risks/benefits and the objectives of the study.

In agreement with current translation studies, this study suggests that great caution is needed in the use of questionnaires originally developed in other languages, and consequently in cross-cultural studies. In fact, a number of studies showed the difference in the semantic relationships among words in different languages, not only in very different cultures (Fontaine et al., 2013) but also in closer contexts such as between neolatin languages (Galati et al., 2008) or even in the same language when it is spoken in different countries. It is not clear how emotions can be correctly translated into different cultures and languages: an equivalent translation is not always possible and an adaptation could probably be suggested. Moreover, even if a word with a correspondent meaning exists in a language, it can have a different frequency of use (Van Goozen & Frijda, 1993) and this can strongly influence the rating procedure of the emotion. In addition, in some languages, especially in French and Italian, many essentially "correct" translations also have very strong additional connotations, being normally disambiguated by the context (Scherer, 1988).

Comparative studies among affective dimensions associated with odours highlight the importance of culture specific-tools in the investigation of odour-related effects, also because it cannot be excluded that subtle cross-cultural differences exist in the quality of the effects designated by two terms considered correctly translated (Ferdenzi et al. 2011; Kagan 2007). This is the case of nostalgico/nostalgic that we observed in this 28

749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775

study. These findings suggest that great attention should be paid in the way of presenting emotions using language and further research is needed in the translation/adaptation of questionnaires for the application in cross-cultural studies (e.g. to compare consumer emotional responses among different countries; see Thomson & Marketo, 2013; Ferdenzi et al., 2013).

In conclusion, EmoSemio is appropriate when the emotional profile of a specific product category is of interest. With its focus on the product and its attention to reduce ambiguity in the use of language by introducing a context, this method allows a fine-grained analysis with relatively modest costs compared to its benefits. The present study demonstrated that 25 preliminary interviews (each lasting about half an hour) are sufficient to develop an EmoSemio questionnaire. It also highlights the importance of selecting frequent user of the product category of interest, both for interviews and responses collection; moreover, it is extremely important to select products that span the most sensory diversity for the interviews, in order both to facilitate the comparison between products and to be sure not to miss important emotions. Further studies are needed to experiment with EmoSemio on other product categories, testing its reliability and suitability with different food and also non-food products.

When the interest is to compare different product categories to obtain a general overview, a predetermined questionnaire can be used.

Furthermore, this study showed the importance of investigating the emotional profile of food products in relation to identified sensory differences among samples. Individual differences in liking should always be considered to segment consumers in order to explore the relationships between sensory and emotional profiles in depth. 29

776 777 778 779 780 781 782

This information is of great importance in the senso-emotional optimisation of products (Thomson, 2007), because it takes into account the sensory properties that drive liking and the associated emotions.

Acknowledgements This research was supported by funds and a grant Borsa di Studio Annalisa Intermoia from Adacta International.

30

783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806

References Appendix F. Labels describing affective states in five major languages. In K. R. Scherer (Ed.) (1988). Facets of emotion: Recent research (pp. 241-243). Hillsdale, NJ: Erlbaum. [Version revised by the members of the Geneva Emotion Research Group]. Copy retrieved [Date] from http://www.unige.ch/fapse/emotion/resmaterial/resmaterial.html Ares, G., Piqueras-Fiszman, B., Varela, P., Marco, R.M., Lpez, A. M., & Fiszman, S. (2011). Food labels: Do consumers perceive what semiotics want to convey?. Food Quality and Preference, 22, 689698. Belson, W. A. (1981). The Design and Understanding of Survey Questions. London: Gower. Cardello, A. V., Meiselman, H.L., Schutz, H. G., Craig, C., Given, Z., Lesher, L. L., & Eicher, S. (2012). Measuring emotional responses to foods and food names using questionnaires. Food Quality and Preference, 24, 243250. Chrea, C., Grandjean, D., Delplanque, S., Cayeux, I., LeCalve, B., Aymard, L., et al. (2009). Mapping the semantic space for the subjective experience of emotional responses to odors. Chemical Senses, 34, 4962. Chu, S., & Downes, J.J. (2002). Proust nose best: Odors are better cues of autobiographical memory. Memory & Cognition, 30, 51118. Clore, G. L., Orthony, A., & Foss, M. A. (1987). The psychological foundations of the affective lexicon. Journal of Personality and Social Psychology, 53, 751-766. D'Urso, V., Cavicchio, F., & Magno Caldognetto, E. (2005). Le etichette lessicali nelle ricerche sperimentali sulle emozioni: problemi teorici e metodologici. In: F. Albano Leoni (Ed.), La comunicazione bimodale (pp. 122161). Napoli: Liguori.

31

807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830

DUrso, V., & Galati, D. (1990). Lanalisi dello spazio semantico di termini emozionali italiani, Ricerche di psicologia, 14, 2955. Delplanque, S., Chrea, C., Grandjean D., Ferdenzi C., Cayeux I., Porcherot C., Le Calv, B., Sander, D., & Scherer, K.R. (2012). How to map the affective semantic space of scents. Cognition & Emotion. 26(5), 885-898. Eco, U. (1979). Lector in fabula. Milano: Bompiani. Eco, U. (1990). The limits of interpretation. Bloomington: Indiana U.P. Ferdenzi, C., Schirmer, A., Roberts, C., Delplanque, S., Porcherot-Lassallette, C., Cayeux, I., et al. (2011). Affective dimensions of odor perception: A comparison between Swiss, British, and Singaporean populations. Emotion, 11, 11681181. Ferdenzi, F., Delplanque, S., Barbosa, P., Court, K., Guinard, J.X., Guo, T., Roberts, S. C., Schirmer, A., Porcherot, C., Cayeux, I., Sander, D., & Grandjean, D. (2013). Affective semantic space of scents. Towards a universal scale to measure selfreported odor-related feelings. Food Quality and Preference, 30(2), 128138 Ferrarini, R., Carbognin, C., Casarotti, E. M., Nicolis, E., Nencini, A., & Meneghini, A. M. (2010). The emotional response to wine consumption. Food Quality and Preference, 21, 720725. Floch, J.M. (1990). Smiotique, Marketing et Communication. PUF, Paris. Floch, J.M. (1995). Identites Visuelles. PUF, Paris. Fontaine, J. R. J., Scherer, K. R., & Soriano, C. (2013). Components of emotional meaning: A sourcebook. New York and Oxford: Oxford University Press. Frijda, N. H. (2008). The psychologists point of view. In: M. Lewis, J.M. HavilandJones, L. Feldman Barrett (Eds.). Handbook of Emotions (3rd ed.) (pp. 6887). New York: The Guilford Press.

32

831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854

Frijda, N. H., & Scherer, K. R. (2009). Emotion definitions (psychological perspectives). In D. Sander & K. R. Scherer (Eds.). The Oxford companion to emotion and the affective sciences (pp. 142144). Oxford and New York: Oxford University Press. Frijda, N. H., & Zammuner, V. L. (1992). Letichettamento delle proprie emozioni. Giornale Italiano di Psicologia, 19(3), 389-423. Galati, D. (2002). Prospettive sulle emozioni e teorie del soggetto. Torino: Bollati Boringhieri. Galati, D., Sini, B., Tinti, C., & Testa S. (2008). The lexicon of emotion in the neoLatin languages. Social science information, 47(2), 205-220. Gius, E., Cozzi, A., & Spagnotto, D. (1992). Il linguaggio delle emozioni. Studio sullorganizzazione delle parole che comunicano stati emotivi. Giornale italiano di psicologia, 19, 56384. Greimas, A. J., & Courts, J. (Eds.) (1979). Smiotique Dictionnaire Raisonn de la Thorie du Langage. Hachette: Paris. Hbert, L. (2007). Dispositifs pour l'analyse des textes et des images. Limoges: Presses de l'Universit de Limoges. (Tools for Text and Image Analysis. An Introduction to Applied Semiotics. [Ebook] http://www.signosemio.com/documents/Louis-Hebert-Tools-for-Texts-andImages.pdf) Herz, R., Schankler, C., & Beland, S. (2004). Olfaction, emotion and associative learning: effects on motivated behavior. Motivation and Emotion, 28, 363-83. Herz, R. S. (2002). Influence of odors on mood and affective cognition. In: C. Rouby, B. Schaal, D. Dubois, R. Gervais, A. Holley (Eds.). Olfaction, taste, and cognition (pp. 160177). Cambridge: Cambridge University Press.

33

855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877

Herz, R. S., & Schooler, J.W. (2002). A naturalistic study of autobiographical memories evoked by olfactory and visual cues: testing the Proustian hypothesis. American Journal of Psychology, 115, 21-32. Jaeger, S. R., & Hedderley, D. I. (2013). Impact of individual differences in emotional intensity and private body consciousness on EsSense Profile responses. Food Quality and Preference, 27, 5462. Jaeger, S. R., Cardello, A. V., & Schutz, H. G. (2013). Emotion questionnaires: A consumer-centric perspective. Food Quality and Preference, 30, 229241. Kagan, J. (2007). What is emotion? : history, measures, and meanings. New Haven and London: Yale University Press. King, S. C., & Meiselman, H. L. (2010). Development of a method to measure consumer emotions associated with foods. Food Quality and Preference, 21, 168177. King, S. C., Meiselman, H. L., & Carr, B. T. (2010). Measuring emotions associated with foods in consumer testing. Food Quality and Preference, 21, 11141116. King, S. C., Meiselman, H. L., & Carr, B. T. (2013). Measuring emotions associated with foods: Important elements of questionnaire and test design. Food Quality and Preference, 28, 816. Laros, F. J. M., & Steenkamp, J.-B. E. M. (2005). Emotions in consumer behavior: a hierarchical approach, Journal of Business Research, 58, 14371445. Majid, A. (2012). Current Emotion Research in the Language Sciences. Emotion Review, 4, 432-443. Manzocco, L., Rumignani, A., & Lagazio, C. (2013). Emotional response to fruit salads with different visual quality. Food Quality and Preference, 28, 17-22.

34

878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902

Martens, H., & Martens, M. (2001). Multivariate analysis of quality. An introduction. Chichester (UK): John Wiley & Sons. Mulligan, K., & Scherer, K.R. (2012). Toward a Working Definition of Emotion. Emotion Review, 4, 345-357. Naes, T., Brockhoff, B., & Tomic, O. (2010). Quality control of sensory profile data. In: Statistics for sensory and consumer science (pp. 1138). Chichester (UK): John Wiley & Sons. Ng, M., Chaya, C., & Hort, J. (2013a). Beyond liking: Comparing the measurement of emotional response using EsSense Profile and consumer defined check -all-that-apply methodologies. Food Quality and Preference, 28, 193-205. Ng, M., Chaya, C., & Hort, J. (2013b). The influence of sensory and packaging cues on both liking and emotional, abstract and functional conceptualisations. Food Quality and Preference, 29, pp. 146-156. Nth, W. (2006). Marketing: Semiotics. In K. Brown (ed.) Encyclopedia of Language and Linguistics, 2nd ed. (pp. 504-507). Oxford: Elsevier. Ortony, A., Clore, G. L., & Foss, A. M. (1987). The referential structure of the affective lexicon. Cognitive Science, 11, 341-364. Piqueras-Fiszman, B., Ares, G., & Varela, P. (2011) Semiotics and perception: do labels convey the same messages to older and younger consumers?. Journal of Sensory Studies, 26, 197208. Plutchik, R. (1980). Emotion: A psychoevolutionary synthesis. New York: Harper & Row. Porcherot, C., Delplanque, S., Gaudreau, N., & Cayeux, I. (2013). Seeing, smelling, feeling! Is there an influence of color on subjective affective responses to perfumed fabric softeners?. Food Quality and Preference, 27, 161-169. 35

903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927

Porcherot, C., Delplanque, S., Planchais, A., Gaudreau, N., Accolla, R., & Cayeux, I. (2012). Influence of food odorant names on the verbal measurement of emotions. Food Quality and Preference, 23, 125133. Porcherot, C., Delplanque, S., Raviot-Derrien, S., Calv, B.L., Chrea, C., Gaudreau, N., et al. (2010). How do you feel when you smell this? Optimization of a verbal measurement of odor-elicited emotions. Food Quality and Preference, 21, 938947. Reisenzein, R., & Hofmann, T. (1990). An Investigation of Dimensions of Cognitive Appraisal in Emotion Using the Repertory Grid Technique. Motivations and Emotions, 14(1), 126. Richins, M. L. (1997). Measuring emotions in the consumption experience. Journal of Consumer Research, 24(2), 12746. Scherer, K. R. (2005). What are emotions? And how can they be measured?. Social Science Information. 44(4), 695729. Schifferstein, H. N. J., & Desmet, P. M. A. (2010) Hedonic asymmetry in emotional responses to consumer products. Food Quality and Preference, 21, 11001104. Smith, V., Mgelvang-Hansen, P., & Hyldig, G. (2010) Spin versus fair speak in food labelling: A matter of taste?. Food Quality and Preference. 21, 10161025. Soudry, Y., Lemogne, C., Malinvaud, D., Consoli, S.-M., & Bonfils, P. (2011). Olfactory system and emotion: Common substrates. European Annals of Otorhinolaryngology, Head and Neck diseases, 128, 1823. Thomson, D. M. H. (2007), SensoEmotional optimization of food products and brands. In: MacFie H. (Ed.) Consumer-led and food product development (pp. 281303). Cambridge: Woodhead Publishing Limited. Thomson, D. M. H., Crocker, C. (2013), A data-driven classification of feelings. Food Quality and Preference, 27, 137152. 36

928 929 930 931 932

Thomson, D. M. H., Crocker, C., & Marketo, C. G. (2010). Linking sensory characteristics to emotions: An example using dark chocolate. Food Quality and Preference, 21, 11171125. Van Goozen, S., & Frijda, N. (1993). Emotion words used in six European countries. European Journal of Social Psychology, 23(1), 8995.

37

Tables

Table 1: Descriptors of chocolate and hazelnut spreads (Descriptive Analysis). Category Appearance Descriptor Colour Definition The intensity of the brown colour Standard Pantone (min 171134 TPX, intermediate 181124 TPX, max 191218) -

Brightness

The appearance associated with the amount of light reflected by the product surface Taste associated with sucrose solutions Taste associated with bitter solutions such as caffeine Taste associated with acid agents such as citric acid solutions

Flavour

Sweet Bitter Sour

Hazelnut

Odour associated with roasted 12 g of roasted hazelnut hazelnuts (about 10) pounded in a mortar and presented in an amber glass of 100cc Odour associated with milk chocolate 15 g of crumbled milk chocolate (Novi) in an amber glass of 100cc 10 g of cocoa powder (Perugina) in an amber glass of 100cc 10 g of butter (Mukki) in an amber glass of 100cc 30 l of vanilla essence in 30 ml of water in an amber glass of 100cc 30 g of yogurt (Mukki) in an amber glass of 100cc

Milk chocolate

Cocoa

Odour associated with cocoa powder

Buttery

Odour associated with fresh butter

Vanilla

Odour associated with vanilla

*Yogurt

Odour associated with yogurt

Mouthfeel

Graininess

Degree to which grains or granules are perceived in the mouth (contrary of homogeneous) Tendency of the sample to stick to the palate Tendency of the sample to create a thick and doughy mass (contrary of fluid)

Stickiness Thickness

Table 2: EmoSemio questionnaire for chocolate and/or hazelnut spreads EMOSEMIO QUESTIONNAIRE REAL SENTENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. It is an anti-stress: it calms me, it soothes me, it reassures me It relaxes me and make me feel carefree I associate it with amusement and fun It makes me feel full of energy and reinvigorated It makes me merry It makes me happy It satisfies me It gratifies me, rewards me It makes me feel tender and affectionate It makes me feel cuddled and loved It communicates sensuality, it charms me It communicates security I associate it to happy memories of childhood It makes me feel good and generous It surprises me It makes me curious It makes me feel indifferent It bores me It makes me feel neglected, without any care for me It makes me feel sad It disappoints me It makes me feel guilty It annoys me, it makes me nervous EMOSEMIO QUESTIONNAIRE LABELS Anti-stress Relaxed Amused Energetic Merry Happy Satisfied Gratified Tender Cuddled Sensual Secure Happy memory Generous Surprised Curious Indifferent Bored Neglected Sad Disappointed Guilty Annoyed

In the first column the 23 sentences of EmoSemio questionnaire are reported. For readability for each of them, in the second column, a word used to sum up its meaning in the discussion is reported, but consumers evaluated exclusively the complete sentences presented in column 1.

Table 3: Mean emotion scores using EmoSemio. Emotions Amused Annoyed Anti-stress Bored Cuddled Curious Disappointed Energetic Generous Gratified Guilty Happy Happy memory Indifferent Merry Neglected Relaxed Sad Satisfied Secure Sensual Surprised Tender G 2.98a 1.13d 3.09a 1.28c 2.97a 2.80a 1.23e 3.06
a

T 2.70b 1.30cd 2.83b 1.40bc 2.63b 2.50b 1.36de 2.73

b

L 2.62b 1.40bc 2.71bc 1.55b 2.57b 2.51b 1.48cd 2.58

b

B 2.52b 1.50bc 2.49cd 1.61b 2.31c 2.27bc 1.66bc 2.50

bc

P 2.23c 1.57b 2.35d 1.62b 2.17c 2.29b 1.77b 2.33

c

E 1.93d 2.18a 1.82e 2.11a 1.79d 2.02c 2.32a 1.98

d

p <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.17 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.06 <0.0001

2.90a 3.09a 1.47 3.18a 3.04a 1.39d 3.06a 1.27d 3.04a 1.15d 3.23a 2.65a 2.67a 2.59 2.89a

2.61b 2.65b 1.51 2.70b 2.54b 1.68c 2.77b 1.39cd 2.77b 1.30cd 2.83b 2.43ab 2.34b 2.38 2.60b

2.59b 2.62bc 1.48 2.69b 2.39bc 1.70bc 2.77bc 1.33cd 2.67bc 1.38bc 2.77b 2.42b 2.27bc 2.38 2.40bc

2.33c 2.37cd 1.56 2.53bc 2.36bc 1.93b 2.53cd 1.48bc 2.49cd 1.43bc 2.63b 2.11c 2.04cd 2.18 2.29c

2.16c 2.27d 1.65 2.33c 2.18c 1.93bc 2.37d 1.59b 2.37d 1.52b 2.33c 2.01c 1.99d 2.25 2.23c

1.77d 1.84e 1.59 1.93d 1.63d 2.28a 1.89e 1.83a 1.89e 2.03a 1.86d 1.61d 1.65e 2.29 1.81d

Discriminating emotions across products are indicated in bold. abcde Samples with the same letter code within a row are not significantly different.

Table 4: Mean emotion scores using EsSense ProfileTM. Emotions Active Adventurous Affectionate Aggressive Bored Calm Daring Disgusted Eager Energetic Enthusiastic Free Friendly Glad Good Good-natured Guilty Happy Interested Joyful Loving Merry Mild Nostalgic Peaceful Pleasant Pleased Polite Quiet Satisfied Secure Steady Tame G 3.06a 2.53a 3.26a 1.72 1.47d 3.19a 2.49a 1.31
d

T 2.84b 2.35ab 2.76bc 1.61 1.73c 2.96b 2.20b 1.51

cd

L 2.65b 2.14bc 2.64c 1.73 1.79bc 2.88b 2.26b 1.73

bc

B 2.69b 2.2b 2.86b 1.75 1.83bc 2.82b 2.26b 1.70

bc

P 2.67b 2.15bc 2.75bc 1.82 1.96ab 2.87b 2.15b 1.76

b

E 2.40c 1.97c 2.25d 1.89 2.08a 2.57c 2.06b 2.53

a

p <0.0001 <0.0001 <0.0001 0.18 <0.0001 <0.0001 0.0035 <0.0001 0.06 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.17 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.49 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0006 0.26

1.55 3.33a 3.21a 3a 3.31a 3.09a 3.42a 3.25a 1.62 3.27a 3.19a 3.19a 3.16a 3.24a 3.17a 1.88 3.36a 3a 3.36a 3.43a 3.25a 3.44a 3.31a 2.94a 2.21

1.45 2.91b 2.6b 2.75b 3.02b 2.78bc 2.96b 2.95b 1.69 2.77b 2.84b 2.75b 2.73b 2.87b 2.91b 1.81 2.88b 2.56b 2.83b 3.12b 2.97b 2.92b 3.03b 2.67bc 2.04

1.58 2.72b 2.60b 2.62b 2.87b 2.64c 2.89b 2.85b 1.74 2.65b 2.64b 2.63b 2.63b 2.68b 2.70b 1.89 2.86b 2.44b 2.77b 2.85c 2.89b 2.65c 2.91b 2.64bc 2

1.5 2.88b 2.69b 2.73b 3b 2.90ab 2.97b 2.86b 1.74 2.80b 2.70b 2.77b 2.76b 2.82b 2.81b 1.75 2.93b 2.58b 2.9b 3.02bc 2.93b 2.87bc 2.87b 2.59bc 2.03

1.74 2.69b 2.51b 2.69b 2.87b 2.61c 2.81b 2.81b 1.71 2.67b 2.75b 2.64b 2.62b 2.76b 2.77b 1.95 2.82b 2.44b 2.74b 2.94bc 2.81b 2.79bc 2.82b 2.69b 2.08

1.61 2.36c 2.01c 2.22c 2.41c 2.21d 2.41c 2.39c 1.89 2.18c 2.18c 2.22c 2.18c 2.36c 2.42c 1.85 2.44c 2.14c 2.3c 2.57d 2.58c 2.08d 2.47c 2.47c 2

Tender Understanding Warm Whole Wild Worried

3.20a 3.19a 3.36a 3.14a 1.87 1.42c

2.74b 2.81b 2.95b 2.64b 1.87 1.54bc

2.61b 2.69b 2.76b 2.56b 1.8 1.55bc

2.68b 2.81b 2.91b 2.72b 1.81 1.69b

2.72b 2.74b 2.82b 2.58b 1.84 1.68b

2.33c 2.33c 2.32c 2.14c 1.86 1.90a

<0.0001 <0.0001 <0.0001 <0.0001 0.96 <0.0001

Discriminating emotions across products are indicated in bold. abcd Sample with the same letter code within a row are not significantly different.

Table 5: Emotions in EmoSemio and EsSense ProfileTM: discriminating power (F-values from ANOVA models computed for each emotion and number of relative sample groups discriminated on the basis of the post hoc LSD99% test) and valence (r=Pearson Coefficient values resulting from the correlation between emotion and liking ratings).
EmoSemio Emotions F5;719 24.5 n. of discriminated groups 5.00 r Emotions EsSense ProfileTM F5;707 14.84 8.24 6.97 24.19 12.16 23.86 12.52 15.01 13.31 18.65 17.59 14.21 20.86 n. of discriminated groups 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 r

Anti-stress

0.99

Peaceful Quiet Calm

0.94 0.97 0.96 0.95 0.98 -0.98 0.94 0.95 0.93 0.95 0.95 0.94 0.95

Cuddled Satisfied Generous Annoyed Merry

21.65 25.09 24.89 23.48 21.89

5.00 4.00 4.00 4.00 4.00

0.97 0.99 0.98 -0.99 0.98

Satisfied Good-natured Disgusted Merry Joyful Pleasant

Happy

21.35

4.00

0.98

Happy Pleased Glad Enthusiastic

Gratified Disappointed Relaxed

20.54 20.41 19.65

4.00 4.00 4.00

0.97 -0.99 0.99

Peaceful Quiet Calm 14.84 8.24 6.97 11.83 17.01 16.33 15.34 15.2 7.67 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 0.94 0.97 0.96 0.90 0.96 0.91 0.96 0.94 0.95

Tender

18.76

4.00

0.98

Tender Warm Affectionate Loving

Energetic Amused Sensual Secure Sad Bored Indifferent Neglected Curious Surprised Guilty -

18.15 16.89 16.29 21.41 17.39 12.41 11.1 9.42 7.81 2.18 1.54

4.00 4.00 4.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 -

0.98 0.97 0.97 0.99 0.97 -0.99 -0.98 -0.97 -0.96 0.95 0.63 -0.69

Energetic Active Nostalgic Secure Bored Interested Worried Guilty Whole Good Polite Friendly Understanding

Happy memory 21.68

0.88 12.26 7.5

3.00 3.00

-0.03 0.99 -0.95

16.12 6.69 1.55 16.36 15.29 14.71 14.04 12.54

3.00 3.00 3.00 3.00 3.00 3.00 3.00

0.95 -0.97 -0.96 0.93 0.98 0.93 0.98 0.95

Free Mild Adventurous Steady Daring Eager Aggressive Tame Wild

11.36 8.49 5.81 4.43 3.56 2.14 1.52 1.31 0.21

3.00 3.00 3.00 2.00 2.00 -

0.96 0.95 0.93 0.86 0.89 -0.44 -0.8 0.65 0.12

Indicates a word in EsSense that share partial meaning but is not correspondent to the concept expressed in EmoSemio. Non-discriminating emotion are indicated in italic.

Figures (captions): Fig. 1. Sensory differences and similarities among chocolate and hazelnut spread samples (B, C, G, L, N, P, R, S, T): correlation loading plot from Principal Component Analysis computed on panel averages of each significant attribute (p<0.05) arising from the two-way ANOVA mixed model. Samples were included as dummy variables. Outer and inner circles on the map represent 100% and 50% explained variance respectively. Selected samples are circled. Fig. 2. Intuitive map of the semantic relationships between emotions (EmoSemio) Fig. 3. Mean liking scores for chocolate and hazelnut spreads (EmoSemio and EsSense ProfileTM respondents; n=238). abcde Letters indicate significantly different mean scores (p<0.001). LSD99%=0.41. Fig. 4 Internal preference map computed on liking scores (n=238). Respondents to EmoSemio questionnaire are indicated in black, while respondents to EsSense ProfileTM questionnaire in white. Fig. 5 Comparisons of the mean emotional responses to the six products (EmoSemio) Fig. 6 Comparisons of the mean emotional responses to the six products (EsSense ProfileTM) Fig. 7 Differences in emotions between T and P (T-P) with EmoSemio (on the left) and EsSense ProfileTM (on the right). T>P; T<P. * Significant differences for p0.05 Fig. 8 Differences in emotions between G and E (G-E) with EmoSemio (on the left) and EsSense ProfileTM (on the right). G>E; G<E. * Significant differences for p0.05

0,5 L T E G 0 -1 -0,5 0 0,5 1

B -0,5

-1

EsSense

EmoSemio