An investigation on the appropriateness of chocolate to match tea and coffee

Food Research International 63 (2014) 464–476

Contents lists available at ScienceDirect

Food Research International journal homepage: www.elsevier.com/locate/foodres

An investigation on the appropriateness of chocolate to match tea and coffee G. Donadini ⁎, M.D. Fumi Institute of Oenology and Agri-Food Engineering, Università Cattolica del Sacro Cuore, via Emilia Parmense, 84, I-29122 Piacenza, Italy

a r t i c l e

i n f o

Article history: Received 29 December 2013 Received in revised form 3 May 2014 Accepted 29 May 2014 Available online 14 June 2014 Keywords: Chocolate Coffee Tea Flavor match Drivers of liking Pairing liking

a b s t r a c t The aim of this study was to provide some recommendations for selecting a befitting tea and coffee to match chocolate with different cocoa contents. Three coffee samples (chocolate flavored, vanilla flavored and unflavored coffee), four tea samples (black tea, green tea, vanilla flavored tea and citrus flavored tea) and three chocolates (30%, 70%, and 99% cocoa) were hedonically rated by eighty regular chocolate consumers. The beverages were then paired with each chocolate, and the consumers were asked to indicate the hedonic liking of the resulting pairings, and to indicate whether the chocolate or beverage flavor dominated the pairing flavor. This study showed that liking of chocolate and coffee pairings and chocolate and tea pairings significantly varied (p b 0.001) across samples. Consumers preferred pairings with 30% cocoa and 70% cocoa chocolate to pairings with 99% cocoa chocolate. Overall, coffee is significantly (p b 0.001) preferred to tea as a chocolate partner. Chocolate and beverage pairing liking was biased by the liking of the beverage tasted alone, the liking of chocolate tasted alone, beverage type, chocolate type and the level of flavor match between chocolate and tea or coffee in a given pairing. When chocolate and beverage flavor balanced out in a given pairing, chocolate and coffee/tea pairings were significantly preferred by the consumers. A significant decrease in acceptance was observed when beverage or chocolate flavor dominated the flavor of the pairing, much more so when the chocolate flavor dominated the pairing flavor. However consumers don't enjoy any preferred chocolate with any preferred tea or coffee because some flavors match better than do others. Indeed, consumers formulate their hedonic responses taking into consideration what flavors go well together more than they rely exclusively on their hedonic judgments of the chocolates, the teas, and the coffees tasted alone. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Chocolate has a long history of pairings in different cuisines, rituals, and ceremonies (Alberts & Cidell, 2006; Cidell & Alberts, 2006), and various chocolate recipes exist for entrees, desserts, beverages, appetizers, pasta and meat. Despite the use of chocolate in a wide array of foods and the growing interest in the area of taste personalization focusing on the comprehension of factors that drive food and beverage pairings, only one systematic research has been conducted that examined pattern of consumers' preferences for chocolate and beverage pairings (Donadini, Fumi, & Lambri, 2012). The results of this preliminary study showed that: 1) consumers preferred pairings with chocolate containing 30% and 70% cocoa solids over pairings with chocolate containing 99% cocoa solids; 2) balsamic vinegar, coffee, liqueur wine and Port wine were more suitable and versatile

⁎ Corresponding author. Tel.: +39 0523599181; fax: +39 0523599232. E-mail addresses: [email protected] (G. Donadini), [email protected] (M.D. Fumi).

http://dx.doi.org/10.1016/j.foodres.2014.05.038 0963-9969/© 2014 Elsevier Ltd. All rights reserved.

partners of chocolate than were spirits, beers, and other types of wines; 3) chocolate and beverage pairing liking depended on the liking of the beverage tasted alone more than on the liking of chocolate tasted alone; and 4) a drop in acceptance was generally observed when chocolate flavor dominated the flavor of the pairing, whereas a drop or a gain in acceptance was observed when beverage flavor dominated chocolate. This study requires to be complemented by specific pieces of information on flavor interactions that take place when chocolate and beverages are consumed according to a mixed tasting technique and on sensory drivers of consumers' acceptance. The exploration of the relations between sensory qualities of food and consumers' liking appears to be a key point to uncover underlying patterns of consumer preference for food and beverage matches and to select the most befitting beverage to match chocolate and simultaneously satisfy the hedonic demand of consumers. Indeed, previous literature in the field of food pairing showed that for consumers some flavors harmonize better than do others and that consumers formulate their hedonic response to food and beverage matches taking into consideration what flavors go well together more than they rely exclusively on their personal preference for the food or

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

the beverage tasted alone (Donadini & Fumi, 2011; Donadini, Fumi, & Lambri, 2013; Donadini, Spigno, Fumi, & Pastori, 2008). Innovation is strategic to the food industry. Brainstorming activities involving product developers, marketers and consumers are critical for new concept generation as are field tasting surveys with consumers in real environments of consumption in offering valuable feedbacks on product performance (Moskowitz, Porretta, & Silcher, 2005; van Kleef, van Triyp, & Luning, 2005). This is particularly valuable in Italy, home of some of the world's finest chocolate companies and creative artisanal chocolatiers (AIDEPI, 2012) for whom developing new flavors and thinking up new taste combinations are paramount to the fulfillment of consumers' demands. Actually Italians indulge in milk chocolate but also love the sophistication of dark chocolate (Stagni, 2007). Espresso lovers, Italians are used to matching chocolate with coffee in their everyday life. They savor habitually a square of chocolate after dinner before an espresso, melt the chocolate right in the cup, stir their espresso with a chocolate spoon until the tip melts or eat espresso encased in a shell of dark chocolate in the form of praline. Tea culture in turn is not as developed as coffee culture. However a growing proportion of Italian consumers, for the most part females, boosted sales of tea in recent years, especially of green tea, fruit/herbal tea and other tea with low tannin content (Euromonitor International, 2013). Tea is mainly consumed during breakfast or as cold refreshing tea, and pairing tea with particular types of food is still an evolving art far away from being a consolidated habit of consumers. Chocolate producers, restaurateurs, and gourmets are beginning to explore pairing tea with food, chocolate included, in a more systematic way and to offer consumers a range of guidelines warning them that no longer does tea go with just breakfast foods. The current study was designed to explore the hedonic response of consumers to chocolate and tea pairings and chocolate and coffee pairings by conducting tasting sessions with regular chocolate consumers in a natural environment of consumption. To reach this overall goal, a panel of eighty consumers was asked to hedonically rate twenty-one pairings of three coffees, four teas and three chocolates with a range of cocoa and to indicate whether the chocolate or the tea/coffee flavor dominated each pairing. To understand the relationships between the sensory data and consumer liking, expert panelists were asked to explore the sensory properties of the chocolate, the teas, the coffees and the resulting pairings by Descriptive Analysis (DA). The relations between DA data of the twenty-one pairings and the consumers' hedonic responses to these pairings were examined by means of PLS-Regression.

465

2. Materials and methods 2.1. Chocolate, tea and coffee samples and preparation for serving Three chocolates with different cocoa contents (30%, 70%, and 99% cocoa), four teas (black tea, green tea, vanilla flavored tea and citrus flavored tea) and three coffees (chocolate flavored, vanilla flavored and unflavored coffee) were chosen to cover a range of different sensory characteristics. Information on ingredients, manufacturer and chocolate sensory characteristics is summarized in Table 1 and 2. Sample preparation for serving was in typical format extensively described in Donadini et al. (2012).

2.2. Determination of the sensory profile of coffee, tea, chocolate samples and the resulting pairings Sensory profiling of the tea samples, the coffee samples and the resulting pairings were carried out in a test room designed in accordance with ISO standards (ISO, 2007) with Descriptive Analysis (DA; Stone & Sidel, 2004) and a lexicon generated following the guidelines of Lawless and Civille (2013), in the typical format described in Donadini et al. (2012). The descriptors for coffee and tea selected and used in this study are listed in Table 3 and 4 respectively. The descriptors selected and used in DA of chocolate and coffee pairings and in DA of chocolate and tea pairings included basic tastes (sweet, bitter, sour and salty), flavors (burnt, smoky, caramel-like, dried fruit-like; milk-like, licorice-like, tobacco-like, cocoa-like, vanilla-like, spicy), textures (gritty, fatty, sticky, soluble, firm, creamy, mouthcoating, body, and astringent), and two overall terms (flavor persistence, flavor intensity). In addition, specific flavor descriptors were selected for DA of chocolate and coffee pairings (i.e. toasted and roasted) and of chocolate and tea pairings (i.e. straw-like, fruity, cooked vegetable-like, mint-like, herb-like, citrus-like). For a comprehensive description of the lexicon, see Drewnowski and Almiron-Roig (2010); Geel, Kinnear, and De Kock (2005); International Coffee Organization (ICO) (2010); Lee and Chambers (2007); Maetzu et al. (2001); Narain et al. (2004); Navarrini, Cappuccio, Suggi-Liverani, and Illy (2004); Nebesny and Budryn (2006); Lingle (2001); Soh et al. (2008). The panel employed by the authors in DA consisted of eight assessors who had previously participated in the sensory tests described in Donadini et al., 2012.

Table 1 Chocolate, tea and coffee samples selected for the study. Samples

Brand

Producer

Ingredients as labeled a

30% cocoa

Lindt Excellence Extra “fondant lait”

Lindt & Sprüngli , Kilchberg (Ag), Switzerland

70% cocoa 99% cocoa Black tea Green tea Vanilla flavored tea Citrus flavored tea Coffee Chocolate flavored coffee Vanilla flavored coffee

Lindt Excellence 70% “noir intense” Lindt Excellence 99% “noir absolute” Lipton yellow label tea Premia Tè Verde Vanilla Tea AgrumanceTM Tea Lavazza Crema e Gusto Flavored coffee chocolate Flavored coffee vanilla

Lindt & Sprünglia, Kilchberg (Ag), Switzerland Lindt & Sprünglia, Kilchberg (Ag), Switzerland Unilever, London, England. Agorà Network S.c.a.r.l. Milano R. Twining and Company Limited, London, England R. Twining and Company Limited, London, England Luigi Lavazza S.p.A., Torino, Italy Torrefazione MUSETTI, Pontenure, Italy Torrefazione MUSETTI, Pontenure, Italy

a

Distributor: Lindt & Sprüngli S.P.A. I-21056 Induno Olona (VA), Italy.

Sugar, cocoa butter, powdered whole milk, cocoa paste, anhydrous butter, lactose, powdered skimmed milk, malted barley extract, soy lecithin, flavorings Cocoa paste, sugar, cocoa butter, Bourbon vanilla beans Cocoa paste, cocoa powder, cocoa butter, raw cane sugar Black tea Green tea Black tea, flavorings, vanilla beans Black tea, orange peel, lemon peel, citrus flavorings. Mix (30% Arabica, 70% Robusta) of ground coffee Ground coffee with flavorings Ground coffee with flavorings

466

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

Table 2 Descriptive Analysis (DA) of the three chocolate samples used in this study (Donadini et al., 2012). Sensory descriptors are scored on a 1–9 intensity scale and refer to chocolate qualities perceived in mouth. Legend: b 3 = low to very low; 3–5 = low to medium; b 5 & ≤ 6 = medium to moderately intense; N 6 & ≤7 = moderately intense to intense; N7 = very intense. (n) = by nose.

30% cocoa chocolate

70% cocoa chocolate

99% cocoa chocolate

Low to very low

Low to medium

Medium to moderately intense

Moderately intense to intense

Bitter, sour, salty, dried fruit, dried fruit (n), licorice, licorice (n), spicy, smoky, smoky (n), tobacco, tobacco (n), roasted, roasted (n), cacao, cacao (n), vanilla, vanilla (n), gritty, astringent Sweet, spicy, vanilla, vanilla (n), milk, milk (n), tobacco (n), roasted (n), gritty, sticky, creamy

Caramel, caramel (n), milk (n), firm, crunchy

Flavor persistence, fatty, sticky, soluble, creamy

Sweet, milk, aroma persistence (n), mouthcoating

Salty, smoky, caramel, caramel (n), dried fruit, dried fruit (n), licorice, licorice (n), cocoa (n), tobacco, roasted, fatty, soluble, mouthcoating Sour, salty, caramel (n), licorice (n), sticky, mouthcoating

Sour, smoky (n), firm, crunchy, astringent

Bitter, cocoa, flavor persistence, aroma persistence (n)

Smoky, tobacco (n), roasted (n)

Tobacco, cocoa, cocoa (n) aroma persistence (n), astringent

Sweet, caramel, milk, milk (n), licorice, spicy, roasted, vanilla, vanilla (n), smoky (s), dried fruit, dried fruit (n), gritty, fatty, soluble, creamy

A specific training of the panel was performed in the typical format described in Donadini et al. (2012) for pairing evaluation only because none of the panelists had participated in previous trainings dealing with the description of sensory characteristics of food and beverage combinations. Samples were evaluated in duplicate with replicates tasted in different days for a total of two sessions for tea (four samples per session) and coffee (three samples per session) and fourteen sessions for the resulting pairings (three sessions with three samples each for tea and four sessions with three samples each for coffee). The panelists tasted each pairing according to a sequential and mixed tasting technique (Nygren, Gustafsson, & Johansson, 2003a). Panelists were told to take the portion of chocolate served to them and to let it melt on the tongue for 5 s. They were then instructed to move the chocolate around the mouth so that the chocolate could coat the oral cavity. A sip of the beverage was then taken and moved around the mouth. The sensory attributes of each pairing were evaluated on a 9-point intensity scale after swallowing.

Table 3 Descriptive Analysis (DA) of the coffee samples. Values are means of n = 8 panelists and two replicates. The means followed by different letters within a row are significantly different at p ≤ 0.05 according to Duncan post hoc test. Abbreviations: chocolate coffee = chocolate flavored coffee; vanilla coffee = vanilla flavored coffee. Descriptors Body Bitter⁎ Sweet Sour Astringent Vanilla-like Citrus-like⁎⁎ Chocolate-like Oily Caramel-like Dried fruit Burnt/roasted Tobacco-like Toasted Spicy⁎ Leather-like Smoky Mushroom-like Earth-like⁎⁎⁎ Flavor persistence Flavor intensity ⁎ p = 0.001. ⁎⁎ p = 0.011. ⁎⁎⁎ p = 0.005.

Coffee (4.37 (6.75 (1.12 (1.50 (5.12 (1.12 (1.25 (1.37 (2.50 (1.12 (1.25 (4.25 (2.62 (4.75 (2.37 (2.37 (2.31 (1.12 (1.37 (6.25 (6.62

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

Chocolate coffee a

0.72) 0.45)b 0.35)a 0.52)c 0.62)b 0.34)b 0.45)a 0.50)b 0.52)a 0.34)a 0.45)a 0.50)a 0.60)a 0.45)b 0.50)a 0.50)a 0.48)a 0.35)a 0.50)b 1.00)b 0.50)a

(3.12 (7.25 (1.12 (2.12 (3.75 (1.37 (1.00 (4.00 (2.87 (1.12 (1.25 (2.00 (2.00 (3.37 (1.75 (1.75 (1.37 (1.12 (2.50 (5.75 (5.75

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

c

0.62) 0.45)a 0.35)a 0.62)b 0.68)c 0.50)b 0.00)b 0.52)a 0.34)a 0.34)a 0.45)a 0.52)b 0.52)b 0.50)c 0.45)b 0.45)b 0.50)b 0.35)a 0.73)a 0.45)c 0.68)b

Vanilla coffee (3.62 (6.25 (1.12 (3.62 (5.87 (2.25 (1.00 (1.00 (2.87 (1.25 (1.50 (4.12 (1.75 (5.12 (2.37 (1.37 (1.00 (1.12 (1.37 (6.87 (6.75

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.50)b 1.00)c 0.35)a 0.50)a 0.35)a 0.45)a 0.00)b 0.00)c 0.62)a 0.45)a 0.73)a 0.34)a 0.45)b 0.34)a 0.50)c 0.50)c 0.00)c 0.35)a 0.50)b 0.34)a 0.45)a

Very intense

Bitter, flavor persistence, firm, crunchy.

Sample identification and order of presentation across sessions and assessors was done according to Donadini et al. (2012). The panelists were provided with still mineral water and unsalted breadsticks to cleanse the palate between samples and were required to wait ten minutes before tasting the next sample. 2.3. Consumer hedonic tests 2.3.1. Consumers Consumers were randomly recruited in Milan and its immediate vicinity. Eligibility for selection depended on the following criteria: 1) regular consumers of chocolate; 2) do not have a specific aversion for chocolate with high cocoa content and for un-sugared tea and coffee; 3) do not have any known food allergies or dietary intolerance; 4) spanned a range of age classes and regions of origin; and 5) willingness and availability to participate in the study. As part of the participant recruitment, information on demographics and patterns of chocolate consumptions were collected via a paper & pencil questionnaire in a pre-test meeting as described in Donadini et al., 2012. A range of chocolate tablets were provided as a reward for consumer participation. Eighty consumers (40 females and 40 males) were randomly selected within respondents in order to mirror the demographics and preference patterns of the average Italian chocolate consumer described in Donadini et al. (2012). 2.3.2. Tasting sessions with consumers The hedonic tasting sessions were conducted in a natural environment of consumption. A spacious outdoor veranda for drink and casual meals was chosen. The veranda was lined with plants and with views through the trees of a pre-alpine garden. Tables and chairs were made of durable wood with cotton made Italian country inspired white and red check table linens. Because the maximum number of consumers the veranda could accommodate was 30, the consumers were randomly assigned to four groups of equal size (n = 20) to complete the hedonic tests. Hedonic tests consisted of three stages. In the first stage, each subgroup of consumers were asked to rate how much they liked each chocolate and each beverage tasted alone (two sessions in total, one for coffee and one for tea) on a nine-point hedonic scale (Peryam & Pilgrim, 1957) ranging from 1 (dislike extremely) to 9 (like extremely). In the second stage, chocolate and tea pairings (n = 12; three sessions of four samples each) and chocolate and coffee pairings (n = 9; three sessions of three samples each) were tasted by the consumers according to the sequential and mixed tasting technique illustrated in Section 2.2. Samples were scored for liking on the nine-point hedonic scale previously described.

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

467

Table 4 Descriptive Analysis (DA) of the tea samples. Values are means of n = 8 panelists and two replicates. The means followed by different letters within a row are significantly different at p ≤ .05 according to Duncan post hoc test. Abbreviations: vanilla tea = vanilla flavored tea; citrus tea = citrus flavored tea. Descriptors

Vanilla tea

Straw-like Mint-like Cooked vegetable like⁎ Burnt Spicy Dried fruit like⁎⁎ Tobacco like⁎⁎⁎

(3.25 (1.75 (3.00 (1.12 (1.87 (2.87 (1.37 (1.50 (2.50 (6.50 (1.25 (4.87 (2.12 (3.87 (2.12 (2.25 (6.12 (6.12 (1.50 (1.25 (2.87

Herb-like Caramel-like Vanilla-like Citrus-like Sweet Bitter Fruity Astringent Body⁎⁎⁎⁎ Flavor intensity Flavor lenght Sour Smoky⁎⁎⁎⁎⁎ Salty ⁎ ⁎⁎ ⁎⁎⁎ ⁎⁎⁎⁎ ⁎⁎⁎⁎⁎

p p p p p

= = = = =

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.68)ab 0.68)b 0.52)a 0.34)a 0.31)a 0.34)a 0.50)a 0.89)b 0.73)a 1.03)a 0.45)a 0.34)a 0.34)b 0.34)a 0.62)c 0.45)b 0.34)a 0.62)a 0.52)c 0.45)b 0.34)a

Citrus tea (1.62 (3.37 (1.37 (1.12 (1.00 (1.12 (1.00 (2.37 (1.12 (2.50 (5.50 (3.50 (3.25 (4.00 (4.12 (2.62 (5.12 (5.50 (2.12 (1.37 (2.12

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.50)c 0.72)a 0.50)d 0.34)a 0.00)b 0.34)c 0.00)b 1.02)a 0.34)b 0.52)b 0.73)a 0.73)b 0.86)a 0.52)a 0.62)a 0.50)a 0.34)b 0.52)b 0.34)a 0.50)b 0.62)b

Black tea (2.87 (1.50 (1.87 (1.12 (1.12 (2.12 (1.25 (2.00 (1.25 (1.87 (2.25 (2.75 (1.75 (3.00 (3.37 (2.12 (3.25 (4.25 (1.75 (1.25 (1.62

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.08)b 0.89)b 0.62)c 0.34)a 0.34)b 0.34)b 0.45)ab 0.45)ab 0.45)b 0.34)c 0.45)b 0.45)c 0.45)c 0.52)b 0.50)b 0.34)b 0.45)c 0.45)c 0.44)b 0.45)b 0.50)c

Green tea (3.62 (1.37 (2.37 (1.00 (1.00 (1.12 (1.12 (1.62 (1.12 (1.25 (1.25 (3.25 (2.25 (1.62 (3.37 (2.25 (2.37 (2.37 (1.25 (1.75 (2.62

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.50)a 0.50)b 0.50)b 0.00)a 0.00)b 0.34)c 0.34)ab 0.72)b 0.30)b 0.45)d 0.45)c 0.45)b 0.45)b 0.50)c 0.50)b 0.45)b 0.50)d 0.50)d 0.45)d 0.45)a 0.50)a

0.001. 0.040. 0.015. 0.013. 0.009.

Successively the consumers scored each pairing after ingestion of the sample on a verbally-anchored just about right (JAR) scale (Rothman & Parker, 2009; Shepherd, Smith, & Farleigh, 1989; Stone & Sidel, 2004) partially modified as reported by King and Cliff (2005) for the flavor that lingered the most, either the chocolate flavor or the beverage flavor, according to the procedure described in Donadini et al., 2012. Within each session, samples were monadically presented to the consumers in a randomized order of serving. The panelists were provided with still mineral water and unsalted breadsticks to cleanse the palate between samples and were required to wait ten minutes before tasting the next sample. 3. Statistical analysis One-way analysis of variance (ANOVA) and factorial ANOVA were performed according to Meilgaard, Civille, and Carr (2007). One-way ANOVA was performed to detect statistically significant differences (p ≤ 0.05) among the teas tasted alone, the coffees tasted alone, and the resulting pairings for the set of sensory attributes of this study. The data were summarized as means ratings, with rating variability represented as standard deviation. Duncan's multiple range test was performed at p ≤ 0.05 to determine homogeneous mean groups. For reason of wordage Descriptive Analysis tables of the twenty-one pairings are not displayed in this manuscript but can be shared by the authors upon reader request. The same analytic approach was used to analyze consumers' preferences for chocolate, tea and coffee without prior consumption of the chocolate and for the twenty-one pairings. Results are reported in Sections 4.9, 4.10, and 4.11 within brackets as mean ± SEM (standard error of the mean) followed by homogeneous mean groups (italicized letters). Principal Component Analysis (PCA) with Varimax rotation was run on the correlation matrix of (1) the mean intensity ratings of the sensory characteristics of the twelve chocolate and tea pairings and of the nine chocolate and coffee pairings separately to explore sensory differences among pairings; (2) of the mean hedonic intensity scores of the sensory attributes that significantly distinguished among teas (n = 4) and coffees (n = 3) tasted alone and teas and coffee with prior and mixed consumption of chocolate (n = 21) to explore the effect of

prior consumption of chocolate on the sensory qualities of tea and coffee. Sensory variables with communalities b 0.500 were omitted from further analysis. Extracted components were retained if they accounted for more than 1.0 unit of variance (Massart, Vandeginste, Deming, Michotte, & Kaufman, 1988). Results are interpreted through two dimensional bi-plots displaying sensory variables and samples (Gabriel, 1971). Factorial ANOVA tested the effect of chocolate liking (i.e. liking of chocolate samples tested alone by the consumers), beverage liking (i.e. liking of tea and coffee tasted alone by the consumers), beverage type, chocolate type, the level of flavor match between chocolate flavor and tea or coffee flavor in a given pairing and their two-way interactions on the dependent variable ‘overall liking of the pairing’ (p ≤ 0.05). Over all samples, the data of the JAR scale measuring flavor dominance were aggregated into three levels: level of chocolate flavor dominance (chocolate dominance), level of flavor balance (perfect balance), and level of beverage flavor dominance (drink dominance). Duncan's posthoc test was performed to determine homogenous mean groups across levels at p ≤ 0.05. The effect size of significant variables in the ANOVA model was evaluated by omega squared (ω2). A small effect size was assigned when ω2 was between 0.0099 and 0.0588, a medium effect size was assigned when ω2 was between 0.0588 and 0.1379, and a large effect size was assigned when ω2 was greater than 0.1379 (Cohen, 1977). Partial Least Square Regression (PLSR) (Wold, 1995) was used to investigate the relations between the sensory attributes of the pairings, the pairings and consumers' hedonic responses. Two models were produced to predict the liking of chocolate and tea pairings and the liking of chocolate and coffee pairings separately. Sensory data were the explanatory variables (X-matrix) and consumers' hedonic scores were the response data (Y-matrix). Two indicators of fit were considered. R2 is the goodness of fit of the model and represents the proportion of variability in Y accounted for by the model. The Q2 is an index of predictive ability (Wold, Sjöström, & Eriksson, 2001). Standardized coefficients for sensory attributes were considered significant when their variable of importance in projection (VIP) scores were greater than 1.0 (Eriksson, Johansson, KettanehWold, & Wold, 1999).

468

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

All data were analyzed using SPSS version 13.5 (SPSS, Inc., Chicago, IL) and XLSTAT (Addinsoft, New York, NY). 4. Results 4.1. Descriptive Analysis of chocolate Table 2 summarizes the mean intensity ratings of the attributes used to describe the three chocolate samples reported in Donadini et al., 2012. Results are split into sensory characteristics perceived by the nose and in the mouth (i.e. flavors and textural properties). 4.2. Descriptive Analysis of coffee The mean intensity ratings of the attributes for the three coffees are shown in Table 3. Significant differences between the three samples have been observed on seventeen out of twenty-two sensory attributes. Sweet, oily, caramel-like, dried fruit-like, and mushroom-like didn't discriminate significantly among samples. Black coffee had a moderately intense to intense flavor that lasted moderately long to long. It carried a moderately intense to intense bitterness, a medium to moderately intense astringency, a moderately low to medium toasted, roasted and burnt flavor. Vanilla flavored coffee had a moderately intense to intense flavor that lasted moderately long to long. It carried a moderately intense to intense bitterness, a medium to moderately intense astringency, roasted and toasted flavor, a moderately low to medium burnt flavor and a low to moderately low sourness. Chocolate flavored coffee had a medium to moderately intense flavor that lasted moderately to moderately long. It carried an intense to very intense bitterness, a moderately low to medium chocolate flavor, a low to moderately low toasted flavor and astringency. 4.3. Descriptive Analysis of tea The mean intensity ratings of the attributes for the four tea samples are shown in Table 4. Significant differences between the four samples have been observed on twenty out of twenty-one sensory attributes. Burnt didn't discriminate significantly across samples. Green tea was described as having low to moderately low straw-like flavor, sweet notes and astringency. All the other attributes received very low intensity scores in the range 1–3 on the 9-point intensity scale. Black tea was described as having a moderately low to medium flavor length, a low to moderately low overall flavor intensity and astringency. All the other attributes received low intensity scores in the range 1–3 of the 9-point intensity scale. Citrus flavored tea was perceived as having medium to moderately intense flavor intensity and length. Also it carried a medium to moderately intense citrus-like flavor, a moderately low to medium fruitiness and astringency. Furthermore citrus flavored tea carried low to moderately low mint-like, sweet and bitter notes. All the other attributes received low intensity scores in the range 1–3 of the 9-point intensity scale. Vanilla flavored tea had a moderately intense to intense overall flavor that lasted moderately long to long. Vanilla flavored tea carried a moderately intense to intense vanilla-like flavor, a moderately low to medium sweetness and low to moderately low straw-like, cooked vegetable-like flavor and fruitiness. All the other attributes received low intensity scores in the range 1–3 of the 9-point intensity scale. 4.4. Descriptive Analysis of chocolate and coffee pairings Twenty-eight out of twenty-eight descriptors significantly discriminated among chocolate and coffee samples. Sensory differences among chocolate and coffee pairings are visualized by means of PCA bi-plots of Fig. 1.

Six principal components had eigenvalues N than 1 and accounted for 99.35% of the total variance. PC1 accounted for 42.21% of the variance explained. It was composed of: 1) crunchiness, firmness, grittiness, cocoa-like flavor, astringent, bitter, sour, salty, burnt, and spicy notes with positive loadings N 0.700; and 2) vanilla-like, caramel-like, milklike flavor and sweetness with negative loading b − 0.700. PC2 (15.20% variability) differentiated the samples on smoky and overall flavor intensity with positive loadings N 0.700; and on fattiness and dried fruit-like flavor with negative loadings b −0.700. PC3 (11.37% variability) was composed of toasted and tobacco-like flavor with positive loadings N 0.700 and on perceived level of body and creaminess with negative loading b−0.700. PC4 (11.03% variability) was composed of burnt flavor, mouthcoating, stickiness and flavor persistence with loading N 0.700. As displayed in PCA bi-plot of Fig. 1, PC1 separated 30% cocoa chocolate and coffee pairings from high cocoa chocolate and coffee pairings. Actually 30% cocoa chocolate and coffee pairings resulted sweet, caramel-like, milk-like, dried-fruit-like, and carried a fatty texture. As expected, 30% cocoa chocolate and vanilla flavored coffee pairing was characterized by vanilla-like flavor. 99% cocoa chocolate and coffee pairings and to a lesser extent 70% cocoa chocolate and coffee pairings were characterized by salty, sour, astringent, bitter, roasted, spicy, burnt and cocoa-like notes. Further, they resulted firm, crunchy and gritty during mastication. 70% cocoa chocolate and coffee pairings tended to be richer in smoky notes, poorer in dried fruit-like notes, and more flavor intense overall than 99% cocoa chocolate and coffee pairings. 4.5. Descriptive Analysis of chocolate and tea pairings Thirty out of thirty-two descriptors significantly discriminated among chocolate and tea pairings. Herb-like and burnt flavor did not significantly discriminate among chocolate and tea pairings. Sensory differences among chocolate and tea pairings are visualized by means of PCA bi-plots of Fig. 2. Eight principal components had eigenvalues N 1 and accounted for 99.57% of the total variance. PC1 accounted for 40.26% of the variance explained. It was composed of: 1) firm, crunchy, soluble, gritty, astringent, burnt, tobacco-like, bitter, spicy, sour, and cocoa-like with positive loadings N0.700; and 2) sweet, milk-like, caramel-like, vanilla-like, fatty and creamy with negative loading b − 0.700. PC2 accounted for 15.76% of the variance explained. It consisted of: 1) perceived level of body, citrus-like, fruitylike with loadings N0.700; and 2) straw-like, cooked vegetable-like, and herb-like with loadings b − 0.700. 30% cocoa chocolate and tea pairings are separated from the other chocolate and tea pairings along PC1. They are located in the upper left sector of the bi-plot as being perceived as sweet, milk-like, caramel-like, and vanilla-like. Further they were perceived as being creamy, fatty and easily melting when these samples were manipulated in the mouth during consumption. High cocoa chocolate and tea pairings are located in the right sector of bi-plot of Fig. 2. They are perceived as being bitter, astringent, cocoalike, burnt, roasted, spicy and tobacco-like overall. Further they were perceived as being firm, crunchy and gritty. Pairings with 70% cocoa chocolate and 99% cocoa chocolate and flavored teas resulted fruitier, more bitter, more astringent, and richer in tobacco-like and cocoa-like flavors than unflavored samples. 4.6. Effect of prior and mixed consumption of chocolate on tea flavor The PCA bi-plot in Fig. 3 visually displays differences in sensory qualities between the teas consumed without prior consumption of chocolate and the tea samples consumed in combination with chocolate. Five principal components had eigenvalues N1 and accounted for 83.4% of the total variance. PC1 accounted for 30.2% of the variance explained. It was composed of: 1) astringent, bitter, tobacco-like flavor,

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

469

Fig. 1. Descriptive Analysis results of chocolate & coffee pairings. The bi-plot displays the two principal components (horizontal axis PC1 = 41.22%; vertical axis PC2 = 15.20%) of the PCA model run on the average hedonic intensity scores of the sensory attributes of chocolate and coffee pairings. Dark red squares identify the samples whereas yellow circles identify the sensory attributes with loading ≥ 0.700. Legend: sweetness (swe), caramel-like (crm), milk-like (mlk), dried fruit-like (dfr), vanilla-like (van), flavor intensity (iai), cocoa-like (coa), saltiness (sal), spiciness (spi), burnt (bur), roasted (roa), sour (aci), bitterness (bit), astringency (ast), smoky (smo), fatty (fat); gritty (gry), firm (fir). Cfvan = Vanilla flavored coffee; Cfcho = Chocolate flavored coffee; Cf = Coffee; 30 = 30% cocoa chocolate; 70 = 70% cocoa chocolate; 99 = 99% cocoa chocolate. Pairing labels are coded by adding the identifier of tea sample to the identifier of chocolate sample (i.e. Cf 30 = 30% cocoa chocolate × coffee).

cocoa-like flavor, sour and spicy notes with positive loadings N0.700; 2) saltiness and smoky notes with positive loadings bordering 0.700; 3) vanilla-like flavor with negative loading bordering − 0.700. PC2 (25.1% variability) differentiated the samples on perceived body, flavor length and flavor intensity, sweetness, and milk-like flavor with positive loadings N 0.700; 2) caramel-like flavor and dried fruit flavor with positive loadings bordering 0.700; and 3) herb-like flavor with negative loading b−0.700. PC3 (13.9% variability) differentiated the samples on citrus-like and mint-like flavors with positive loadings N 0.700 and straw-like and cooked vegetable-like flavors with negative loadings b− 0.700. PC4 (8.1% variability) was composed of burnt notes with negative loading b−0.700.

The tea samples tasted without prior consumption of chocolate resulted richer in fruitiness, citrus-like, herb-like and mint-like flavors than samples tasted in combination with chocolate. They were perceived as being less bodied, less flavor intense and less flavor persistent than samples tasted with prior consumption of chocolate. Tea samples tasted with prior consumption of 30% cocoa chocolate were perceived as being richer in sweetness, milk-like, caramel-like, and dried fruit-like flavors. Also they were perceived as being less astringent, bitter, sour and spicy as well as being less rich in cocoa and tobacco-like flavor. The tea samples tasted in combination with 70% cocoa and 99% cocoa were more bitter, saltier, and richer in astringency, sourness,

Fig. 2. Descriptive Analysis results of chocolate & tea pairings. The bi-plot displays the two principal components (horizontal axis PC1 = 40.26%; vertical axis PC2 = 15.76%) of the PCA model run on the average hedonic intensity scores of the sensory attributes of chocolate & tea pairings. Dark red squares identify the samples whereas yellow circles identify the sensory attributes with loading ≥ 0.700. Legend: sweetness (swe), caramel-like (crm), milk-like (mlk), vanilla-like (van), cooked vegetable-like (ckv), cocoa-like (coa), spiciness (spi), tobaccolike (tob), burnt (bur), bitterness (bit), astringency (ast), citrus-like (cit), fruitiness (fru), straw-like (hay), herb-like (her), body (bdy), gritty (gry), firm (fir), cru (crunchy). Van = vanilla flavored tea; Cit = citrus flavored tea; Blk = black tea; Gre = green tea; 30 = 30% cocoa chocolate; 70 = 70% cocoa chocolate; 99 = 99% cocoa chocolate. Pairing labels are coded by adding the identifier of tea sample to the identifier of chocolate sample (i.e. Gre 30 = 30% cocoa chocolate × green tea).

470

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

Fig. 3. Effect of prior and mixed consumption of chocolate on tea flavor. The bi-plot displays the two principal components (horizontal axis PC1 = 30.2%; vertical axis PC2 = 25.12%) of the PCA model run on the average hedonic intensity scores of the sensory attributes of tea samples consumed without prior consumption of chocolate and of the resulting twelve pairings. Dark red squares identify the samples whereas yellow circles identify the sensory attributes. Legend: sweetness (swe), caramel-like (crm), milk-like (mlk), dried fruit-like (dfr), vanilla-like (van), cooked vegetable-like (ckv), flavor persistence (pai), flavor intensity (iai), body (bdy), cocoa-like (coa), saltiness (sal), spiciness (spi), sourness (aci), tobacco-like (tob), burnt (bur), bitterness (bit), astringency (ast), citrus-like (cit), mint-like (min), fruitiness (fru), straw-like (hay), herb-like (her). Van = vanilla flavored tea; Cit = citrus flavored tea; Blk = black tea; Gre = green tea; 30 = 30% cocoa chocolate; 70 = 70% cocoa chocolate; 99 = 99% cocoa chocolate. Pairing labels are coded by adding the identifier of tea sample to the identifier of chocolate sample (i.e. Gre 30 = 30% cocoa chocolate × green tea).

tobacco-like and cocoa-like flavors. Citrus flavored tea and vanilla flavored tea were the richest samples in citrus and vanilla-like flavor. 4.7. Effect of prior and mixed consumption of chocolate on coffee flavor The PCA bi-plot in Fig. 4 visually displays differences in sensory qualities between the coffees consumed without prior consumption of chocolate and the coffees consumed in combination with chocolate. Four principal components had eigenvalues N 1 and accounted for 89.6% of the total variance. PC1 accounted for 33.4.2% of the explained variance. It was composed of: 1) sweetness, caramel-like flavor, vanilla-like flavor, and fatness with positive loadings N0.700; 2) bitterness and astringency with negative loadings b− 0.700. PC2 (23.3% variability) differentiated the samples on spicy, toasted, burnt, roasted, and tobacco-like flavor with positive loadings N0.700. PC3 (17.9% variability) differentiated the samples on saltiness, chocolate-like, dried fruit-like, and smoky flavors with positive loadings N 0.700. PC4 (12.4% variability) was composed of flavor length and flavor intensity with positive loadings N 0.700. Coffee samples tasted with prior consumption of 30% cocoa chocolate were positioned in the lower right quadrant of the biplots of Fig. 4. They were perceived as being fattier and richer in sweetness, milk-like, caramel-like, and vanilla-like flavors. Furthermore pairings

with 30% cocoa chocolate were perceived as being less bitter, astringent, spicy, toasted and burnt than the other samples tested in this study. Black coffee, black coffee paired with 99% cocoa chocolate, chocolate flavored coffee paired with 99% cocoa chocolate and vanilla flavored coffee were positioned in the upper left sector of the biplots. These samples were bitter, astringent, toasted, burnt and spicy as well as tobacco-like and flavor intense. Chocolate flavored coffee and vanilla flavored coffee paired with 70% cocoa chocolate were positioned in the upper right quadrant of the biplots and carried tobacco-like, burnt and spicy notes and resulted flavor intense above all.

4.8. Flavor dominance Flavor dominance data are displayed in Fig. 5 and Fig. 6. Results are reported for chocolate samples across all tea and coffee samples respectively (Fig. 5) and per tea samples and coffee samples across all chocolate samples (Fig. 6). Chocolate flavor dominated tea flavor (Fig. 5). The flavor of 99% cocoa chocolate, the significantly most dominant chocolate in this cocktail of three chocolate samples, and the flavor of 70% cocoa chocolate, the second most dominant chocolate sample, were dominant to intensely dominant over tea flavor. The flavor of 30% cocoa chocolate was

Table 5 Hedonic response of the consumers to chocolate and tea pairings and to chocolate and coffee pairings. Liking scores are reported as means of n = 80 consumers ± standard error of the mean. The means followed by different letters within a column are significantly different at p ≤ 0.05 according to Duncan post hoc test. Abbreviations: co. = cocoa; fl. = flavored. Chocolate and tea pairings 30% co. chocolate 30% co. chocolate 30% co. chocolate 70% co. chocolate 30% co. chocolate 70% co. chocolate 70% co. chocolate 70% co. chocolate 99% co. chocolate 99% co. chocolate 99% co. chocolate 99% co. chocolate

× × × × × × × × × × × ×

black tea vanilla fl. tea citrus fl. tea citrus fl. tea green tea vanilla fl. tea green tea black tea green tea vanilla fl. tea citrus fl. tea black tea

Liking (6.07 (5.70 (5.17 (4.97 (4.86 (4.43 (4.38 (4.00 (3.19 (3.03 (2.63 (2.50

± ± ± ± ± ± ± ± ± ± ± ±

a

0.27) 0.42)ab 0.35)abc 0.41)bcd 0.33)bcd 0.32)cd 0.26)cd 0.33)de 0.32)ef 0.35)f 0.30)f 0.25)f

Chocolate and coffee pairings

Liking

30% co. chocolate 30% co. chocolate 30% co. chocolate 70% co. chocolate 70% co. chocolate 70% co. chocolate 99% co. chocolate 99% co. chocolate 99% co. chocolate

(7.07 (6.17 (5.87 (5.77 (5.69 (5.17 (3.30 (3.27 (2.21

× × × × × × × × ×

black coffee vanilla fl. coffee chocolate fl. coffee chocolate fl. coffee black coffee vanilla fl. coffee chocolate fl. coffee vanilla fl. coffee black coffee

± ± ± ± ± ± ± ± ±

0.19)a 0.32)ab 0.39)b 0.38)b 0.29)b 0.39)b 0.35)c 0.46)c 0.27)d

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

471

Fig. 4. Effect of prior and mixed consumption of chocolate on coffee flavor. The bi-plot displays the two principal components (horizontal axis PC1 = 33.4%; vertical axis PC2 = 23.2%) of the PCA model run on the average hedonic intensity scores of the sensory attributes of tea samples consumed without prior consumption of chocolate and of the resulting nine pairings. Dark red squares identify the samples whereas yellow circles identify the sensory attributes. Legend: sweetness (swe), caramel-like (crm), milk-like (mlk), dried fruit-like (dfr), vanillalike (van), flavor persistence (pai), flavor intensity (iai), body (bdy), cocoa-like (coa), saltiness (sal), spiciness (spi), sourness (aci), tobacco-like (tob), burnt/roasted (bur), bitterness (bit), astringency (ast), toasted (toa), oily (oil). Cfvan = vanilla flavored coffee; Cfcho = chocolate flavored coffee; Cf = coffee; 30 = 30% cocoa chocolate; 70 = 70% cocoa chocolate; 99 = 99% cocoa chocolate. Pairing labels are coded by adding the identifier of tea sample to the identifier of chocolate sample (i.e. Cf 30 = 30% cocoa chocolate × coffee).

slightly dominant to moderately dominant over tea flavor and was perceived as being significantly less dominant than the other samples. Flavored teas were significantly less dominated by chocolate flavor than unflavored teas (p b 0.001) as displayed in Fig. 6. Chocolate dominated to intensely dominated the flavor of green tea and black tea. The flavor of chocolate moderately dominated to dominated the flavor of vanilla flavored tea and citrus flavored tea. Chocolate flavor dominated (p b 0.001) coffee flavor (Fig. 5). The flavor of 99% cocoa chocolate, the significantly most dominant chocolate in this cocktail of three chocolate samples, was dominant over coffee

flavor. The flavor of 70% cocoa chocolate, the second most dominant chocolate sample, was slightly dominant over coffee flavor. The flavor of 30% cocoa chocolate, the significantly least dominant chocolate sample, yet slightly dominant over coffee, was perceived nearly as intense as coffee flavor overall. Overall chocolate samples (Fig. 6), the flavor of flavored coffees tended to be less dominated by chocolate flavor than the flavor of unflavored coffee. The flavor of chocolate was moderately dominant to dominant over the flavor of black coffee and slightly dominant over the flavor of vanilla flavored and chocolate flavored coffees.

Fig. 5. Consumer scores of flavor match level for chocolate and beverage pairings. Data are presented for tea and coffee across chocolate types. Each pair was scored on a Just About Right (JAR) scale verbally anchored at both extremes with “chocolate dominates extremely at −6 and “beverage (coffee or tea) dominates extremely” at +6. Harmonic pairs were considered as chocolate and beverage combinations where neither the chocolate nor the beverage dominated, i.e. pairings that scored closest to zero. The flavor that lingered in the mouth of the consumers after swallowing determined which flavor dominated between chocolate and tea or coffee. Values are means of n = 80 consumers. The different letters on standard error of the mean bars indicate significantly different mean scores across samples according to Duncan post-hoc test at p ≤ 0.05. Abbreviations used: 30% = 30% cocoa chocolate; 70% = 70% cocoa chocolate; 99% = 99% cocoa chocolate.

472

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

Fig. 6. Consumer scores of flavor match level for chocolate and beverage pairings. Data are presented for tea and coffee types across all chocolate samples. Each pair was scored on a Just About Right (JAR) scale verbally anchored at both extremes with “chocolate dominates extremely at −6 and “beverage (coffee or tea) dominates extremely” at +6. Harmonic pairs were considered as chocolate and beverage combinations where neither the chocolate nor the beverage dominated, i.e. pairings that scored closest to zero. The flavor that lingered in the mouth of the consumers after swallowing determined which flavor dominated between chocolate and tea or coffee. Values are means of n = 80 consumers. The different letters on standard error of the mean bars indicate significantly different mean scores across samples according to Duncan post-hoc test at p ≤ 0.05. Abbreviations used: citrus tea = citrus flavored tea; vanilla tea = vanilla flavored tea; vanilla coffee = vanilla flavored coffee; chocolate coffee = chocolate flavored coffee.

4.9. Hedonic response of the consumers to chocolate samples, tea samples and coffee samples Liking varied significantly (p b 0.001) across chocolate samples, tea samples and coffee samples. 30% cocoa chocolate was moderately liked to liked (6.55 ± 0.119; a) by the consumers and significantly preferred to 70% cocoa chocolate (5.76 ± 0.132; b), that was moderately liked, and to 99% cocoa chocolate (2.41 ± 0.191; c) that was disliked to very much disliked by the consumers. Black tea (5.93 ± 0.153; a) was moderately liked by the consumers and significantly preferred (p b 0.001) to vanilla flavored tea (5.07 ± 0.277; b), green tea (4.64 ± 0.196; b) and citrus flavored tea (4.60 ± 0.299b) that were neither liked nor disliked or slightly disliked on average by the consumers. Black coffee was moderately liked by the consumers (6.17 ± 0.200; a) and significantly (p b 0.001) preferred to chocolate flavored coffee (5.27 ± 0.246; b) that was slightly above the neither like nor dislike hedonic threshold, and vanilla flavored coffee (4.70 ± 0.229; c), that was slightly disliked.

4.10. Consumers' hedonic response to chocolate and tea pairings Consumers' average hedonic scores for chocolate and coffee parings significantly (p b 0.001) varied across samples as reported in Table 5. 30% cocoa chocolate and black tea pairing was moderately liked by the consumers. 30% cocoa chocolate and vanilla flavored tea pairing and 30% cocoa chocolate and citrus flavored tea pairing liking ranged from neither liked nor disliked–moderately liked. 70% cocoa chocolate and citrus flavored tea pairing and 30% cocoa chocolate and green tea pairing were moderately disliked and bordered the neither liked nor disliked hedonic threshold. 70% cocoa chocolate and vanilla flavored tea pairing, 70% cocoa chocolate and green tea pairing, 70% cocoa chocolate and black tea pairing ranged from moderately disliked to neither liked nor disliked. 99% cocoa chocolate and green tea pairing and 99% cocoa chocolate and vanilla flavored tea were disliked moderately to disliked.

Finally 99% cocoa chocolate and citrus flavored tea and 99% cocoa chocolate and black tea pairings were disliked very much. 4.11. Hedonic response of the consumers to chocolate and coffee pairings Consumers' average hedonic scores for chocolate and coffee parings significantly (p b 0.001) varied across samples as reported in Table 5. Liking of 30% cocoa chocolate and black coffee was in the range liked–liked very much. This sample was the most appreciated pairing tasted by the consumers. Liking of 30% cocoa chocolate and vanilla flavored coffee, liking of 30% cocoa chocolate and chocolate flavored coffee, liking of 70% cocoa chocolate and chocolate flavored coffee, liking of 70% cocoa chocolate and black coffee, liking of 70% cocoa chocolate and vanilla flavored coffee were in the moderately liked–liked hedonic range. Finally liking of the three pairings with 99% cocoa chocolate was in the range disliked–disliked very much. 4.12. Modeling the consumers hedonic response of the consumers to the pairing Factorial ANOVA model run in this study to predict pairing liking showed a significant (p b 0.001) small effect of: 1) beverage liking (ω2 = 0.035); 2) chocolate types (ω2= 0.022). Over all tea and coffee samples the pairings with 30% cocoa chocolate were significantly preferred (5.86 ± 0.127; a) to the pairings with 70% cocoa chocolate (4.88 ± 0.132; b) and to the pairings with 99% cocoa chocolate (2.84 ± 0.124; c). The model also showed a significant (p b 0.001) small effect of: 1) chocolate liking (ω2 = 0.021); 2) beverage types (ω2 =0.012; see Fig. 7); and 3) flavor dominance (ω2 = 0.010). Harmonically balanced pairing (6.60 ± 0.193; a) were preferred to pairings dominated by tea or coffee flavor (4.91 ± 0.262; b) and to pairings dominated by the chocolate flavor (4.13 ± 0.095; c). Finally, a significant (p b 0.001) small effect of beverage × chocolate liking (ω2 = 0.026) and of beverage × beverage liking (ω2 = 0.022) as well as of chocolate samples × beverage liking and chocolate samples × chocolate liking at p = 0.016 and p = 0.026 (ω2 = 0.008; 0.006 respectively) were observed.

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

473

Fig. 7. Effect of beverage type on chocolate and beverage pairing liking. Values are means of n = 80 consumers. The different letters on standard error of the mean bars indicate significant different mean scores across samples according to Duncan post-hoc test at p ≤ 0.05. Abbreviations used: citrus tea = citrus flavored tea; vanilla tea = vanilla flavored tea; vanilla coffee = vanilla flavored coffee; choc. coffee = chocolate flavored coffee.

4.13. Identification of drivers of liking of chocolate and tea pairings PLS-regression revealed a good fitness (Q2cum = 0.758) for a 3-component model with 98.2% of the hedonic response and 57.7% of the descriptive data explained. Sensory attributes relevant to accounting for the hedonic responses of the consumers included sweetness which drove primarily and positively liking. Caramel-like flavor, milk-like flavor and vanilla-like flavor drove secondarily and positively liking. Astringency, stickiness, mouthcoating, and cocoa-like flavor also accounted for consumers' preferences and they were primarily negatively associated with chocolate and tea pairing liking. Liking is secondarily negatively driven by bitterness, tobacco-like flavor and firmness.

4.14. Identification of drivers of liking of chocolate and coffee pairings PLS-regression revealed a good fitness (Q2cum = 0.861) for a 3-component model with 98.9% of the hedonic response and 75.0% of the descriptive data explained. Sensory attributes relevant to accounting for the hedonic responses of the consumers included sweetness and vanilla-like flavor which drove primarily liking. Fatness, smoked, caramel-like and milk-like flavor drove secondarily liking. Flavor persistence also accounted for consumers' preferences and it was negatively associated with chocolate and coffee pairing liking. Disliking is secondarily driven by bitterness, astringency, sourness and dried fruit-like flavor.

5. Discussion Overall goal of this study was to explore the hedonic response of regular chocolate consumers to a number of chocolate and tea pairings and chocolate and coffee pairings. Little research has been published in this regard, most guidelines being generated in the field of wine and food matches. Only one study (Donadini et al., 2012) reported preliminary results on the consumer perception of chocolate and beverage matches and included one coffee and two tea samples (i.e. black tea and green tea) in a large set of beverages partnered with chocolate for consumer evaluation. In this study three chocolate samples with a range of cocoa content were served to chocolate consumers as partners of four tea samples and of three coffee samples according to an experimental design

providing that all possible pairings were tasted by the consumers in a natural and social environment of consumption. 5.1. Descriptive Analysis of chocolate, coffee and the resulting pairings Descriptive Analysis showed that chocolates (Table 2), coffee samples (Table 3), and tea samples (Table 4) were significantly different with respect to the range of attributes considered in this study. As expected the flavors experienced by the consumers during the combined consumption of foods and beverages of such a great sensory variability spanned a variety of significantly different oral sensations during the consumption of the twenty-one pairings resulting from all possible combinations of beverages with chocolate samples (Figs. 1 and 2). Coffee and tea samples were low to moderately bodied and varied significantly in sweetness, bitterness, astringency, flavor persistency and intensity (Tables 3 and 4). This has logical appeal because coffee and tea samples varying widely in coffee varieties, blends, growing region/conditions, agronomic and manufacturing practices, fermentative historic, polyphenols and their oxidative products, caffeine content, addition of characterizing flavorings or minor ingredients are reported to have significantly different sensory properties (Bhumiratana, Adhikari, & Chambers, 2011; Geel et al., 2005; Obanda, Owuor, & Mang'oka, 2000; Ravichandran, 2004; Sunarharum, Williams, & Smyth, 2014). Black tea and green tea have a very mild flavor. However green tea is richer in straw-like and cooked vegetable-like flavor as well as in sweet, bitter and salty taste than black tea. Black tea is in turn richer in dried fruit-like, vanilla-like, citrus-like and fruity flavor and resulted the most flavorsome and flavor persistent unflavored tea. As expected, flavored teas were perceived as having a more intense and longer lasting flavor than unflavored teas. Actually flavored teas were perceived as being fruitier, sweeter and richer in vanilla-like and citrus-like flavors than unflavored teas. Characterizing flavors of citrus and vanilla in particular were moderately intense to intense and overwhelmed the cocktail of mild sensory qualities of tea for a final unbalanced overall flavor. Coffee samples were perceived as being more flavorsome and flavor persistent than tea samples. They owe their moderately intense to intense flavor to their intense bitterness, to medium astringency, to moderately low to medium burnt, roasted and toasted flavors. Characterizing flavors of chocolate and vanilla of flavored coffees were perceived as being far less intense than characterizing flavors of vanilla and citrus of flavored teas for a more balanced overall flavor of flavored coffees.

474

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

Chocolate sensory characteristics varied largely across samples of this study as described and discussed extensively in Donadini et al., 2012 and summarized in Table 2. This is due to the varying levels of cocoa, fats, minor ingredients and manufacturing techniques employed by the producer (Alberts & Cidell, 2006; Cidell & Alberts, 2006). 99% cocoa chocolate has a long lasting flavor, intense notes of tobacco and licorice, intense roasted and smoky flavors, extremely intense bitterness and overwhelming astringency (Table 2). As the cocoa level decreases, chocolate samples gain higher ratings on fattiness, creaminess, and on a set of descriptors (i.e. caramel-like, dried fruit-like, milk-like and vanilla-like) that contribute to the overall sweetness perception of chocolate. Due to largely varying fat content in the chocolate samples, texture significantly differs from extremely firm and crunchy (99% cocoa chocolate), to firm and crunchy (70% cocoa chocolate), and to easily melting, creamy, fatty and moderately sticky (30% chocolate). These results confirm the large significant differences in aroma, taste, and texture between chocolates varying in cocoa content reported by Thamke, Dürrschmid, and Rohm (2009). 5.2. Consumers' preferences and factors affecting liking of chocolate and coffee pairing and chocolate and tea pairing Consumer hedonic response to chocolates, coffees, teas and the resulting pairings significantly vary across samples. Patterns of preference for 70% and 99% cocoa chocolate are expected considering that consumers usually dislike bitter and astringent foods and dislike chocolates which melt poorly or display very intense sensory properties (Dinella, Recchia, Tuorila, & Monteleone, 2011; Januszewska & Viaene, 2001a; Januszewska & Viaene, 2001b; Lesschaeve & Noble, 2005). Italian consumers indulge in high cocoa chocolate (Stagni, 2007). However 99% cocoa chocolate appears to be so extreme in its flavor and texture perception that 99% cocoa chocolate is disliked much by the consumers. Similar results are reported in Donadini et al., 2012. Overall, unflavored teas and coffees are preferred to flavored samples. Black tea is moderately liked and significantly preferred to vanilla flavored tea, citrus flavored tea and green tea, equally liked and located in the medium range of hedonic liking. Black coffee is moderately liked and significantly preferred to chocolate flavored coffee, which is neither liked nor disliked to moderately liked, and to vanilla flavored coffee, which is slightly disliked and borders the neither like nor dislike hedonic threshold. Similar degree of liking for black tea and black coffee is reported by Donadini et al. (2012). With regard to the preference for the twenty-one pairings, highly significant differences in acceptance are observed across samples. The selection of the type of chocolate, tea and coffee, the liking of chocolate, tea and coffee tested alone, and the level of flavor match between chocolate and tea and chocolate and coffee contribute significantly, yet to different extent, to consumers' hedonic liking of chocolate and tea and coffee combinations. These results replicate and confirm previous findings of studies that explored relevant factors likely to bias patterns of consumers' preference for food and beverage pairings and included in their investigation matches between chocolate and various drinks as well as beer and food combinations (Donadini & Fumi, 2011; Donadini et al., 2008; Donadini et al., 2012; Donadini et al., 2013). In agreement with Donadini et al. (2012), the results of this study showed that the higher the cocoa content of a chocolate is, the less palatable a pairing with tea or coffee. Actually, tea and coffee pairings with 99% cocoa chocolate were disliked moderately to disliked very much whereas tea and coffee pairings with 70% cocoa chocolate were neither liked nor disliked to disliked moderately. By contrast, tea and coffee pairings with 30% chocolate were moderately liked to liked. Beverage type biases the hedonic response of the consumers (p = 0.008) although the effect of beverage type is not as

large as the effect of chocolate type probably because sensory differences between tea and coffee samples are not as large as the difference in flavors between chocolate types (Fig. 5). This appears to be a common finding in literature dealing with the understanding of patterns of preference for food and beverage pairings. According to a previous study of our research team, the selection of food type (i.e. different chocolate types, different dishes of the Italian cuisine, and different cheese types) appears to have a larger effect than the selection of the type of beverage on the formulation of the hedonic judgment of consumers on pairings with a range of beverages including beers of different styles, wines of different styles, mineral water, balsamic vinegar, and spirits (Donadini & Fumi, 2011; Donadini et al., 2008; Donadini et al., 2012; Donadini et al., 2013). Liking of chocolate, tea and coffee tasted alone influence the consumers' hedonic response to the pairings significantly. Overall, the more the consumers like a given chocolate eaten by itself the more the consumers appreciate the pairing with chocolate. Similarly the more the consumers like a given tea or coffee drunk by itself the more the consumers like the pairing with tea or coffee. However a plateauing effect might have occurred since the effect on pairing liking due to the increase in beverage liking tasted alone is not significant for beverage liking ratings that fall in the positive range of the hedonic scale. Similar to personal opinions of maître sommeliers, consumer preference goes to pairings in which food and beverage flavors come together in a perfect harmonic balance. This result corroborates Donadini et al., 2012 who found out that favoring a just about right flavor match is an effective option to increase consumer acceptance for chocolate and beverage pairings. However Donadini et al. (2012) reported that consumers can appreciate and also prefer pairings dominated by the beverage or, less usually, pairing dominated by the chocolate. As confirmed by this study's results, consumers appreciate a pairing dominated by the tea or coffee flavor, or, vice versa, a pairing dominated by chocolate flavor and they seem to base their hedonic judgment of the pairings that don't come in a perfect balance of flavors on the hedonic valence of the dominant flavors in a given match. When bitter, roasted, tobaccolike and astringent notes are dominant flavors in a pairing such as in the case of pairings with high cocoa chocolate, consumers dislike what they taste. When sweet, vanilla-like, caramel-like and milklike notes dominate a given match, consumers like what they taste. Similar patterns of preference were reported by Donadini et al. (2012). The authors found out that consumers enjoy matches between beer and cheese dominated by the flavor of the cheese as far as the cheese is liked and elicits sweet appealing notes in a given cheese and beer match. The complexity of the mechanisms behind the formulation of the hedonic response of consumers to chocolate and tea/coffee pairings is stressed by the evidence that the effect exerted on pairing liking by the number of factors modeled in this study is not linear but it is modulated by a range of interactive effects as reported in Section 4.12. For example, pairing liking was found to depend significantly on liking of chocolate tasted alone. However the effect of chocolate liking on pairing liking is modulated by the chocolate type paired with tea or coffee. Actually consumers don't enjoy any preferred chocolate with any preferred tea or coffee. For consumers some flavors match better than do others and rate a pairing on liking taking into consideration what flavors go well together more than they rely exclusively on their hedonic judgments upon the liking of the chocolate, the tea or the coffee tasted alone. This is congruent with previous literature data on the topic of food and beverage pairing reporting that the best match selection is driven by both individual preferences and food and beverage characteristics (Donadini & Fumi, 2011; Donadini et al., 2013; Harrington, Miszczak, & Ottenbacher, 2008).

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

5.3. Effect of prior chocolate consumption on coffee and tea sensory qualities To complement the basic results of this study, a trained panel of expert sensory analysts investigated the flavor interactions that occur when chocolate and coffee/tea are consumed together according to a sequential and mixed tasting technique. This field is a large research area which remains regrettably under-investigated in real foods and is usually limited to single tastant or binary and tertiary taste mixture models (Green, Lim, Osterhoff, Blacher, & Nachtagl, 2010). The results of this study indicated that when coffee and tea are tasted with prior consumption of chocolate, significant variations in the sensory profile of the beverages are observed. The resulting modifications were medium to large in size and depended on the effect of the chocolate type and the beverage type tasted by the consumers. Actually tea carries a light flavor and is modified largely by chocolate while coffee, more flavorsome and texturized than tea, is only slightly modified. Modifications of the sensory properties of the beverage induced by prior and mixed consumption of food are reported in the literature. Overall, the literature showed a reduction of the sensory properties of wine and beer matched with cheese (Bastian, Collins, & Johnson, 2010; Donadini & Fumi, 2011; Donadini et al., 2013; Madrigal-Galan & Heymann, 2006; Nygren, Gustafsson, Haglund, Johansson, & Noble, 2001; Nygren, Gustafsson, & Johansson, 2002; Nygren, Gustafsson, & Johansson, 2003b; Nygren et al., 2003a). However, some evidence exists that prior consumption of food increases the intensity of specific descriptors of the beverages (Donadini et al., 2013; Madrigal-Galan & Heymann, 2006), although flavor enhancement is less frequently reported than suppressive effects. In this study, tea tasted with prior consumption of chocolate loses characteristic light flavors of fruit, mint, citrus, straw and herbs that are largely overpowered by chocolate. These characteristics are hardly perceived during mixed consumption of chocolate. Bitterness of tea is decreased by 30% cocoa chocolate sweetness that, at medium to high intensity is reported to be generally suppressive of other basic tastes (Stevens, 1995; Stevens & Traverzo, 1997). Astringency of tea is broadly reduced by fat rich 30% cocoa chocolate and this result corroborates outcomes of previous studies which stressed the importance of food fat content in modulating the mouthfeel perception of beverages with respect to the astringent perception (Bastian et al., 2010; Madrigal-Galan & Heymann, 2006; Nygren et al., 2001). Furthermore citrus-like flavor and vanilla-like flavor that are medium to moderately intense and moderately intense to intense respectively in flavored teas tasted without prior chocolate consumption are largely reduced by 70% and 99% cocoa chocolate, the more so the higher the cocoa content of chocolate. Similarly 70% cocoa chocolate and 99% cocoa chocolate overpower the bouquet of mild tea flavors that are hardly perceived in any combination with these two chocolates. As a consequence tea acquires flavors and basic tastes of these two chocolates including tobacco-like, smoky, burnt, spicy, cocoa-like, sour, and bitter notes that were hardly perceived in tea tasted alone. Tea acquires also milk-like, caramel-like, dried fruit-like, and vanilla-like flavors of 30% and 70% cocoa chocolate. Sweetness, moderately intense in tea without prior 30% cocoa chocolate, increases in intensity and becomes extremely intense in all of the four tea samples paired with this chocolate. As a result of the overwhelming chocolate flavor, tea becomes longer lasting, more flavor intense and fuller bodied. Changes in coffee flavors were for the most part the result of prior and mixed consumption of 30% cocoa chocolate which elicits sensory qualities of milk, caramel and a sweet taste hardly perceived in coffee tasted alone. Furthermore sweetness and fatness of 30% chocolate modulate the flavor of coffee by reducing the perception of bitterness, acidity, spiciness, tobacco, toasted, and burnt/roasted flavor.

475

This has a logical appeal as it is reasonable to expect that chocolate sweetness would suppress bitterness and acidity of coffee while chocolate fatness would suppress coffee astringency and reduce the access of flavor active molecules to sensory receptors of the mouth (Keast & Breslin, 2003; Peyrot des Gachons et al., 2012; Schifferstein & Frijters, 1991). 70% cocoa chocolate and 99% cocoa chocolate were perceived as being neither sweet nor fatty and therefore didn't decrease the perception of bitterness, astringency and acidity of coffee. Rather they enhanced the perception of characteristic notes they share with coffee tasted alone, the more so the higher the content in cocoa of the chocolate. Bitterness, astringency, burnt, tobacco-like, and sour notes of coffee tend to be enhanced especially by 99% cocoa chocolate, for an overall longer lasting slightly more intense flavor of the resulting pairings. The current authors' findings are in agreement with previous research that has found that enhancement is usually the prevailing phenomenon when tasting mixtures in which the tastes are similar, whereas suppression is more frequent when the foods are dissimilar (Keast & Breslin, 2003). Other sensory qualities such as toasted flavor tend to be reduced or remain unvaried depending on the specific coffee type paired with chocolate. 5.4. Identification of drivers of liking for chocolate and coffee pairings and for chocolate and tea pairings This study showed that when a given pairing is perceived as bitter, sour, astringent, rich in cocoa-like and tobacco-like flavors by reason of a match with high cocoa chocolate, the consumers dislike the pairing. When the pairing is perceived as sweet, milk-like, vanilla-like and caramel-like by reason of a match with low cocoa chocolate, the consumers appreciate what they consume. Furthermore, when chocolate adheres to and extensively coats the oral cavity or melts poorly so that the chocolate breaks down in small firm fragments during mastication, the consumers dislike the resulting pairing with tea. It is therefore paramount to select a chocolate with the proper cocoa content to control the excess of bitterness, cocoa flavor, sourness and astringency that are elicited in a given pairing by high to very high cocoa chocolates. Of similar high profitability to respond to the consumers' hedonic demands is the selection of low-cocoa chocolates that decreases bitterness, astringency, sourness and elicits pleasant flavor of milk, vanilla and sweet taste in a given pairing with coffee or tea. Furthermore the fat content of the chocolate appears to bias consumer preference for the ability of chocolate fats to modulate the melting dynamics that take place in the mouth during chocolate consumption. High fat chocolates appear to respond better to the consumers' demand in this regard as they melt more easily than low-fat chocolates and diminish simultaneously unappealing astringent sensations. They would be preferred as coffee or tea partner for these reasons in agreement with previous results reporting on the effect of the perceived dynamic contrast in consumer preferences in the food domain (Hyde & Witherly, 1993; Januszewska & Viaene, 2001a). 6. Conclusions This study showed that consumers' hedonic response to chocolate and tea/coffee pairings significantly varied across samples. Overall, coffee is an ideal partner of 30% cocoa and is moderately liked in combination with 70% cocoa chocolate. Tea on the other hand is liked in combination with 30% cocoa chocolate only. Combinations with 99% cocoa chocolate are disliked by the consumers. Pairing liking is biased by beverage liking, chocolate liking, chocolate types, drink types, and the level of flavor match between chocolate and coffee or tea in a given pairing. Also, two-way interactions between drink liking, chocolate liking, chocolate types, and drink types

476

G. Donadini, M.D. Fumi / Food Research International 63 (2014) 464–476

significantly bias pairing liking. The selection of a specific chocolate type appears to bias consumer preferences for the pairings more than the selection of a specific flavored or unflavored tea or coffee. The selection of each component of a pairing is more complicated than merely choosing a chocolate and a tea or coffee liked by the consumers. Actually consumers' best liked match selection is driven by both individual preferences for tea, coffee and chocolate tasted alone and chocolate and beverage characteristics. To satisfy the hedonic demand of consumers it is therefore paramount to select a chocolate with the proper cocoa content to control the excess of bitterness, sourness, astringency, and cocoa flavor that are elicited in a giving pairing by chocolate varieties high to very high in cocoa content. Of similar high profitability to respond to the consumers' hedonic demands is the selection of low-cocoa chocolates that decrease bitterness, astringency, and sourness and elicit pleasant flavor of milk and vanilla and a sweet taste in a given pairing with coffee or tea. The findings of this study should aid tea and coffee producers as well as maître chocolatiers and restaurateurs in understanding patterns of consumers' preferences for chocolate and tea/coffee pairings and enable them to engineer optimal strategies to target the demands of the market. Acknowledgments The authors thank Anna Carabelli, Giuseppe Donadini and Angela Sanvito for their help in conducting the hedonic tests with consumers. References AIDEPI (2012). L'industria italiana del dolce e della pasta: andamento economico 2012. Available at. www.aidepi.it/associazione/pubblicazioni/115-andamento-economico2012.html Alberts, H. C., & Cidell, J. L. (2006). Chocolate consumption, manufacturing and quality in Western Europe and the United States. Geography, 91, 218–226. Bastian, S. E. P., Collins, C., & Johnson, T. E. (2010). Understanding consumer preferences for Shiraz wine and Cheddar cheese pairings. Food Quality and Preference, 21, 668–678. Bhumiratana, N., Adhikari, K., & Chambers, E. (2011). Evolution of sensory aroma attributes from coffee beans to brewed coffee. LWT – Food Science and Technology, 44(10), 2185–2192. Cidell, J. L., & Alberts, H. C. (2006). Constructing quality: The multinational histories of chocolate. Geoforum, 37, 999–1007. Cohen, J. (1977). Statistical power analysis for the behavioural sciences (2nd ed.). New York: Academic Press. Dinella, C., Recchia, A., Tuorila, H., & Monteleone, E. (2011). Individual astringency responsiveness affects the acceptance of phenol-rich foods. Appetite, 56, 633–642. Donadini, G., & Fumi, M.D. (2011). Are red beers of the Italian market tantalizing partners of cheese: Consumers speak out. Personal communication, 34th European Brewery Congress Convention, Glasgow, United Kingdom. Donadini, G., Fumi, M.D., & Lambri, M. (2012). The hedonic response to chocolate and beverage pairing: A preliminary study. Food Research International, 48, 703–711. Donadini, G., Fumi, M.D., & Lambri, M. (2013). A preliminary study investigating consumer preference for cheese and beer pairings. Food Quality and Preference, 30(2), 217–228. Donadini, G., Spigno, G., Fumi, M.D., & Pastori, R. (2008). Evaluation of ideal everyday Italian food and beer pairings with regular consumers and food and beverage experts. Journal of the Institute of Brewing, 114, 329–342. Drewnowski, A., & Almiron-Roig, E. (2010). Human perception and preference for fat-rich foods. In J. P. Montmayeour, & J. le Coutre (Eds.), Fat detection: Taste, texture and post ingestive effects. Boca Raton (FL): CRC Press (Chapter 11). Eriksson, L., Johansson, E., Kettaneh-Wold, N., & Wold, S. (1999). Introduction to multi- and megavariate data analysis using projection methods (PCA & PLS). Umeå, Sweden: Umetrics AB (490 pp.). Euromonitor International (2013). Tea in Italy. Available at. http://www.euromonitor. com/tea-in-italy/report (accessed April, 2014) Gabriel, K. R. (1971). The biplots graphical display of matrices with applications to principal component analysis. Biometrika, 18, 453–467. Geel, L., Kinnear, M., & De Kock, H. L. (2005). Relating consumer preferences to sensory attributes of instant coffee. Food Quality and Preference, 16, 237–244. Green, B. G., Lim, J., Osterhoff, F., Blacher, K., & Nachtagl, D. (2010). Taste mixture interactions: Suppression additivity, and the predominance of sweetness. Physiology and Behavior, 101(5), 731–737. Harrington, R. J., Miszczak, D. C., & Ottenbacher, M. C. (2008). The impact of beer type, pizza spiciness and gender on match perceptions. PASOS Revista de Turismo y Patrimonio Cultural, 6, 173–188. Hyde, R. J., & Witherly, S. A. (1993). Dynamic contrast: A sensory contribution to palatability. Appetite, 21(3), 1–16. International Coffee Organization (ICO) (2010). Consumer oriented vocabulary for coffee. http://www.ico.org/vocab.asp (accessed October 14, 2010).

ISO 8589 (2007). Sensory analysis — General guidance for the design of test rooms. Januszewska, R., & Viaene, J. (2001a). Sensory segments in preference for plain chocolate across Belgium and Poland. Food Quality and Preference, 12, 96–107. Januszewska, R., & Viaene, J. (2001b). Acceptance of chocolate by preference cluster mapping across Belgium and Poland. Journal of Euromarketing, 11, 61–86. Keast, R. S. J., & Breslin, P. A. S. (2003). An overview of binary taste–taste interactions. Food Quality and Preference, 14(2), 111–124. King, M., & Cliff, M. (2005). Evaluation of ideal wine and cheese pairs using a deviation from ideal scale with food and wine experts. Journal of Food Quality, 28, 245–256. Lawless, L. J. R., & Civille, G. V. (2013). Developing lexicons: A review. Journal of Sensory Studies, 28, 270–281. Lee, J., & Chambers, D. H. (2007). A lexicon for flavor descriptive analysis of green tea. Journal of Sensory Studies, 22(3), 256–272. Lesschaeve, I., & Noble, A.C. (2005). Polyphenols: Factors influencing their sensory properties and their effects on foods and beverage preferences. American Journal of Clinical Nutrition, 81, 330–335. Lingle, T. R. (2001). The coffee cuppers' handbook (3rd ed.). California, CA: Specialty Coffee Association of America. Madrigal-Galan, B., & Heymann, H. (2006). Sensory effects of consuming cheese prior to evaluating red wine flavour. American Journal of Enology and Viticulture, 57, 12–22. Maetzu, L., Andueza, S., Ibanez, C., De Pena, M. P., Bello, J., & Cid, C. (2001). Multivariate methods for characterization and classification of espresso coffees from different botanical varieties and types of roast by foam, taste, and mouthfeel. Journal of Agricultural Food Chemistry, 49, 4743–4747. Massart, D. L., Vandeginste, B. G. M., Deming, S. N., Michotte, Y., & Kaufman, L. (1988). Chemometrics: A textbook. Amsterdam: Elsevier (488 pp.). Meilgaard, M., Civille, G. V., & Carr, B. T. (2007). Sensory evaluation techniques (4th ed.). Boca Raton: CRC Press. Moskowitz, H., Porretta, S., & Silcher, M. (2005). Concept research in food product design and development. Wiley (597 pp.). Narain, C., Paterson, A., & Reid, E. (2004). Free choice and conventional profiling of commercial black filter coffees to explore consumer perceptions of character. Food Quality and Preference, 15, 31–41. Navarrini, L., Cappuccio, R., Suggi-Liverani, F., & Illy, A. (2004). Espresso coffee beverage: Classification of texture terms. Journal of Sensor Studies, 35, 525–541. Nebesny, E., & Budryn, G. (2006). Evaluation of sensory attributes of coffee brews from robusta coffee roasted under different conditions. European Food Research and Technology, 224, 159–165. Nygren, I. T., Gustafsson, I. B., Haglund, A., Johansson, L., & Noble, A.C. (2001). Flavour changes produced by wine and food interactions: Chardonnay wine and hollandaise sauce. Journal of Sensory Studies, 16, 461–470. Nygren, I. T., Gustafsson, I. B., & Johansson, L. (2002). Perceived flavour changes in white wine after tasting blue mould cheese. Food Service and Technology, 2(4), 163–171. Nygren, I. T., Gustafsson, I. B., & Johansson, L. (2003a). Perceived flavor changes in blue mould cheese after tasting white wine. Food Service Technology, 3(3–4), 143–150. Nygren, I. T., Gustafsson, I. B., & Johansson, L. (2003b). Effects of tasting technique — sequential tasting vs. mixed tasting — on perception of dry white wine and blue mould cheese. Food Service Technology, 3, 61–69. Obanda, M., Owuor, P. O., & Mang'oka, R. (2000). Changes in the chemical and sensory quality parameters of black tea due to variations of fermentation time and temperature. Food Chemistry, 75, 395–404. Peryam, D. R., & Pilgrim, F. J. (1957). Hedonic scale method of measuring food preferences. Food Technology, 9–14. Peyrot des Gachons, C., Mura, E., Speziale, C., Favreau, C. J., Dubreuil, G. F., & Bresli, A. S. (2012). Opponency of astringent and fat sensations. Current Biology, 22(19). Ravichandran, R. (2004). The impact of pruning and time from pruning on quality on quality and aroma constituents of black tea. Food Chemistry, 84(I), 7–11. Rothman, L., & Parker, M. J. (2009). Just about right (JAR) scales: Design, usage, benefits, and risks. 63, MNL: ASTM International (committee E 18 on sensory evaluation). Schifferstein, H. N., & Frijters, J. E. (1991). The effectiveness of different sweeteners in suppressing citric acid sourness. Percept Psychophys, 49, 1–9. Shepherd, R., Smith, K., & Farleigh, C. A. (1989). The relationship between intensity, hedonic and relative-to-ideal ratings. Food Quality and Preference, 1, 75–80. Soh, M. L., Seo-Jin, C., Ok-Hee, L., Hye-Seong, L., Young-Kyung, K., & Kwang-Ok, K. (2008). Development of sample preparation, presentation procedure and sensory descriptive analysis of green tea. Journal of Sensory Studies, 23, 450–467. Stagni, M. (2007). Analisi strategica dell'industria del cioccolato. Collana casi aziendali n.3/ 2007. (pp. 19) (Milano). Stevens, J. C. (1995). Detection of heteroquality taste mixture. Perception and Psychophysics, 57, 18–26. Stevens, J. C., & Traverzo, A. (1997). Detection of a target taste in a complex masker. Chemical Sense, 22, 529–534. Stone, H., & Sidel, J. L. (2004). Sensory evaluation practice (3rd ed.). San Diego: Elsevier Academic Press. Sunarharum, W. B., Williams, D. J., & Smyth, H. E. (2014). Complexity of coffee flavor: A compositional and sensory perspective. Food Research International, 62, 315–325. Thamke, I., Dürrschmid, K., & Rohm, H. (2009). Sensory description of dark chocolate by consumers. LWT - Food Science and Technology, 42, 534–539. van Kleef, E., van Triyp, H. C. M., & Luning, P. (2005). Consumer research in the early stage of new product development. Food Quality and Preference, 16, 321–341. Wold, S. (1995). PLS for multivariate linear modelling. QSAR. In H. van de Waterbeemd (Ed.), Chemometric methods in molecular design methods and principles in medicinal chemistry, Vol. 2. (pp. 195–218). Weinheim, Germany: Verlag Chemie. Wold, S., Sjöström, M., & Eriksson, L. (2001). PLS-regression: a basic tool of chemometrics. Chemometrics and Intelligent Laboratory Systems, 58, 109–130.