Mealiness in Apples: Towards a Multilingual ... - Wiley Online Library

JFS:

Sensory and Nutritive Qualities of Food

Mealiness in Apples: Towards a Multilingual Consumer Vocabulary Z. ANDANI, S.R. JAEGER, I. WAKELING, AND H.J.H. MACFIE

ABSTRACT: Three commercial varieties of dessert apples known to be susceptible to varying degrees of mealiness were evaluated by descriptive sensory profiling. In parallel, consumer testing using the repertory grid technique was conducted with consumers representing 5 different European countries. Generalized Procrustes Analysis of the repertory grid data suggested a cross-cultural consensus with respect to consumer perception of mealiness. In particular, mealiness was associated with the texture of Cox apples. Curved box-plots, an innovative extension of ordinary box-plots, was used to explore the consumers’ use of descriptive attributes. This indicated that although consumers perceived mealiness similarly, they described their perceptions differently. Keywords: repertory grid methodology, cross-cultural differences, generalized Procrustes analysis, sensory analysis

Introduction

T

HE CONDITIONS THAT LEAD TO THE DEVELOPMENT OF


mealy texture in apples and other fruit, as well as the associated physiological and enzymatic changes have received considerable attention in the literature (Ben-Arie and others 1979; Ben-Arie and Sonego 1980; Harker and Sutherland 1993; Harker and others 1997; Lapsley and others 1992; Andani and others, in press). When a fresh apple is eaten, the breaking of cells across their walls result in a characteristic crisp sound (Drake 1963; Vickers and Bourne 1976; Vickers 1981; Christensen and Vickers 1981). As fruit soften, the force required to break cells decreases and the proportion of the cells that separate by cell-to-cell debonding increases. The texture of apples where the majority of cells separate from each while eaten is described as mealy (Lapsley and others 1992; Harker and Sutherland 1993). Sensorially mealiness has been associated with a soft and dry texture (Harker and Hallett 1992). Previous work has indicated that mealy apples, a result of inappropriate sunburn, maturity and storage conditions (Harker and others 1997), are of low quality and have reduced consumer appeal (Szczesniak and Kahn 1971). It has been estimated that up to 30% of apples from northern Europe may be affected by mealiness to some degree (Nicolaï 1998). Thus, a more thorough understanding of the texture deterioration associated with mealiness as perceived and evaluated by consumers is required. In line with the change toward more global food patterns (Cardello 1993; Stone and Sidel 1995; Karahadian 1995; Askegaard and Madsen 1995), consumer perception of mealiness in apples is explored using repertory grid methodology (RGM) (Kelly 1955) among 5 different European consumer groups. For each consumer panel a vocabulary of descriptors is generated and brought together to establish a multilingual vocabulary for consumer perception of apples (see Gains 1994 for a comprehensive introduction to this methodology). However, to be meaningful such an exercise requires that consumers perceive the sensory properties of apples similarly. Establishing whether a cross-cultural consensus with respect to dessert apples exists, must, therefore, be the 1st step in the process of developing a multilingual con-

872 JOURNAL OF FOOD SCIENCE—Vol. 66, No. 6, 2001

sumer vocabulary. In accordance with evidence in the literature, which indicates that sensory perception is largely similar across cultures (Prescott 1998; Prescott and Bell 1995; Druz and Baldwin 1982), a cross-cultural consensus with respect to consumer perception of apples is expected. The idea of sensory vocabularies for international use is not new. However, previous work (Hunter and McEwan 1998; ESN 1996; Mojet and de Jong 1994) has focused on developing common vocabularies for use with trained panels to aid product standardization and identification. In comparison, the present study also explores differences in concept formation and description among consumers. Because the perceptual processes of naive and untrained subjects is known to differ from those of trained sensory judges (Cardello and others 1982), focus is directed toward the latter group who also best represent the typical end-consumer. This consumer oriented approach is also in accord with current business and marketing practises (Griffin and Hauser 1993). The cognitive aspects of concept formation and measurement of sensory stimuli, which has been the subject of intense study (O’Mahony 1990; Ishii and O’Mahony 1990, 1987), clearly indicates that the vocabulary untrained judges use to describe the sensory characteristics of a given food differs widely. This is due to lack of an agreed system of concept categorization and description. Therefore, it is predicted that consumers describe their perception of mealiness in apples differently.

Materials and Methods Samples Dessert apples from 3 commercially available Northern Hemisphere varieties, known to be susceptible to mealiness, namely Schone van Boskoop, Cox’s Orange Pippin and Jonagold were used. Different quality levels were created by storing apples at increased temperature (21 8C, 80% 6 5% R.H.) for zero, 1 or 2 wk in a climate-room (Wilkinson 1965; Vincent 1989). The fruit resulting from these treatment conditions were characterized as ‘fresh’ (1), ‘mid-point’ (2) and ‘mealy’ (3), respectively. Overall, 9 experimental fruit samples

© 2001 Institute of Food Technologists

Multilingual Vocabulary for Apples . . . tural setting and found it to be useful in conducting interviews as it can be carried out in various countries without the use of a common language. In the present study, 25 subjects, aged 18 to 60 years were recruited in each of 4 countries: Britain, Belgium (25 Flemish, 25 French), Spain, and Denmark. Only consumers who ate dessert apples at least once a week and had the requested language as their 1st language were recruited. Consumer perception and evaluation of apples were explored in oneon-one personal interviews. During the interviews sessions, which lasted approximately 1 h, consumers elicited descriptive constructs and rated the intensity of these for each of the 9 sample treatments. The interviews were conducted in the consumers’ native tongue by 1 native interviewer except in the case of the Belgium interviewer who had Flemish as her first language and French as her second language. A set of instructions on how to conduct the RGM interview and the exact wording of the instructions to be given to consumSensory profiling ers, developed and pre-trailed in the UK, was used in all the Descriptive sensory analysis was performed in the UK by countries. To aid construct (attribute) elicitation each consumer was a panel consisting of 12 females between 30 and 60 years old (none were research employees). All panellists were experi- presented with 6 sets of 3 samples (triads), in such a way that enced assessors and 11 of the 12 members had previously each of the 9 samples were presented twice. For each triad been involved in descriptive sensory profiling of apples re- the consumers were instructed to elicit texture and flavor ported by Dalliant-Spinnler and others (1996), Barreiro and constructs. The subjects were presented with the apples in a others (1998) and Jaeger and others (1998). Testing was car- manner similar to that of the trained panel. Dry crackers and ried out in sensory booths equipped with artificial daylight water were consumed between triads. After a short break lighting, air extraction and air conditioning. Apple samples consumers rated each sample for attribute intensity on a 9(at room temperature) were presented peeled, cored and point hedonic box scale, anchored at ‘not at all’ and ‘exquartered on white paper plates with 3-digit random num- tremely.’ Attributes were rated in the order elicited by the bers. It had previously been ascertained (Bhanji and others consumers. The number of constructs used by individual 1997) that the peel had little effect on the perception of tex- subjects ranged between 5 and 35. However, most subjects ture. Samples, which were presented monadically, were pro- used between 10 and 15 constructs, a figure similar to that of filed in 2 replicates. Nine samples were tested in 1 session (5 Gains and Thomson (1990). before a break and 4 after) using a presentation design balanced for order of presentation and first-order carry over Statistical analysis effects (MacFie and others 1989). The samples were profiled The descriptive sensory analysis data from each attribute using 42 attributes, based on a vocabulary previously devel- was subjected to 2-way analysis of variance (ANOVA) using oped to assess Southern Hemisphere eating apples (Dalliant- samples (variety and storage treatment) and assessors as Spinnler and others 1996). The intensity of each attribute was main factors. Averaging across assessors and replicates, the rated on a 100mm unstructured line scale and recorded us- sample mean values for each attribute were input to princiing the sensory computer programme Taste®. Panellists were pal components analysis (PCA) (Krzanowski 1988) of the corinstructed to score attributes in the order of 1st bite texture relation matrix. (2 attributes), texture during chewing (9 attributes) and flaTo examine cross-cultural differences in perception, the vor during chewing (16 attributes). Panellists waited for 15 data from 125 consumers, 25 in each of the 5 different sec before scoring afterswallow characteristics (4 attributes). countries, were explored by Generalized Procrustes AnalyAttributes related to internal appearance (6 attributes) and sis (GPA) (Gower 1975). The significance of the resulting internal odor (5 attributes) were scored last. consensus configuration was examined by permutation tests (Wakeling and others 1992). In summary, GPA takes Consumer testing into account 3 sources of variation inherent in sensory and Information on the consumers’ product perception was perceptual data. Translation of individual subject configuraobtained using repertory grid methodology (RGM). The tions to a common origin removes the effect of panellists method is particularly suited for vocabulary development using different parts of the measurement scale. Another and Piggott and Watson (1992) recommend it to overcome source of variation occurs when subjects interpret the the difficulty in generate sufficient and adequate descriptors same term differently. The step of rotation/reflection inwhen dealing with consumers. It was originally developed by creases the similarity between configurations (Arnold and Kelly (1955) to identify the constructs that people use to Williams 1986). Finally isotropic scaling is used to overcome structure their perceptions of the social world. Olson (1981) the problem of subjects varying in their range of scoring extended the methodology to the food area hereby making it (McEwan and Hallett 1990). One subject may use a very possible for individuals to generate idiosyncratic constructs small range of the scale, while another may use a much for evaluating objects. Since then, the method has success- larger range to express differences between samples on a fully in applied in a number of food related studies (Thom- particular attribute. Note that the combination of RGM son and McEwan 1988; Scriven and others 1989). In particu- with GPA has been successfully used in a number of conlar, Raats and Shepherd (1993) applied RGM in a cross-cul- sumer studies (Thomson and McEwan 1989; McEwan and Vol. 66, No. 6, 2001—JOURNAL OF FOOD SCIENCE

873


varying in accord with a 3 (variety) 3 3 (storage treatments) full factorial design were used. The fruit, which was grown in Belgium, was harvested over a 4-wk period during September and October 1996. For all 3 varieties the fruit that received the ‘mealy’ storage treatment were harvested later than the fruit receiving the ‘midpoint’ treatment, which again was harvested later than the ‘fresh’ fruit. Generally, Cox and Boskoop apples were harvested earlier than Jonagold. Between harvest and initiation of the experimental treatments the fruit was kept in commercial storage. The length of this storage period varied between 9 and 15 wk depending on harvest date and length of the experimental storage treatment. Following the experimental storage treatment, the apples were transported from Belgium to the different countries in which testing took place and stored in cold storage (2-4 8C), from which they were removed between 12 and 24 h before testing commenced.

Multilingual Vocabulary for Apples . . . Thomson 1988; Scriven and others 1989; Gains and Thomson 1990). The GPA analysis was carried out using the GENPROC directed (Arnold and Payne 1989) in the GENSTAT statistical package (Genstat 1993) on a VAX/VMS computer. For each panel, the descriptors used by consumers in the panel were projected onto the consensus space derived from GPA of all 125 consumers. Considering only those descriptors that correlated significantly (r2 . 0.5) with the consensus configuration, a vocabulary of consumer descriptors was established by grouping synonymous descriptors under common headings. This categorization was performed by the persons who had conducted the RGM interviews. A 2nd person, who had not been involved with the interviews, revised the established categories. Disagreements on the categorization of individual terms were resolved by discussion among these 2 persons. Only descriptors used by 3 or more consumers were considered representative of a whole panel and interpreted. Attributes with an apparent hedonic meaning were eliminated, as were words not discriminating between samples. For the remaining terms the variation in the use of descriptors in a category was examined using curved boxplots. While mathematically similar to standard box-andwhiskers plots (Tukey 1977), these graphically illustrate the angular distribution of descriptors used by more than 1 consumer from the same panel. The box itself accounts for the middle 50% of the attributes and the line inside represents the direction of the median attribute vector. Lines or whiskers extend up to 1.5 times the inter-quartile range, or to the left most or right most attribute vector. Attributes outside of the inter-quartile range are represented by an asterisk and can be considered outlying. Further, to aid examination of the relationship between consumer and sensory descriptors, the sensory attributes were projected onto the GPA consensus space (r2 . 0.5) derived from the 125 consumers.

Results Descriptive sensory analysis Analysis of variance showed a significant effect (p , 0.05) of variety and storage treatment for 13 attributes, namely 1st bite juiciness and hardness, crisp, juicy and floury texture, acid, stale and watery flavor, astringent aftertaste and white, green, yellow and juicy internal appearance. This suggested that samples differed between varieties but also with different storage treatment for these attributes. For 8 attributes, namely bitter, pear-drop and plum/cherry flavor, bitter aftertaste, green lines in internal appearance and grassy, damp twigs/stalks and pears internal odor, the variety, storage treatment and the variety by storage treatment interaction were non-significant at a 5% level, suggesting that these attributes did not differ between the apple samples. For the remaining attributes, mainly flavor characteristics, there was no effect of storage treatment, only a difference between varieties. For all attributes the effect of assessors was significant (p , 0.001). These results indicated that the mean scores for each attribute gave a reliable estimate of the sample’s sensory properties and it was acceptable to analyze the averaged data by PCA. PCA of the correlation matrix resulted in a 2-factor solution accounting for 68.7% of the total variation, of which 39.9% was accounted for by the 1st principal component (PC) and 28.8% by the 2nd. In the scores plot, an axis spanned between the lower right hand and upper left hand corner of Figure 1a separated the samples according to storage treatment. Along this axis ‘mealy’ apples generally had higher positive scores than ‘mid-point’ apples, which again had higher positive scores than ‘fresh’ apples. One deviation form this pattern was Cox apples, where ‘mid-point’ Cox (Cox 2) had a higher positive score than ‘mealy’ Cox (Cox 3).


Figure 1—Plot of the first two principal component scores (a) and loadings (b) for the trained panel. For samples, suffixes refer to mealiness condition, where 1= ‘fresh’, 2 = ‘mid-point’ and 3 = ‘mealy’. For attributes, prefixes refer to sensory attribute category, where 1st = ‘first bite texture’, t = ‘texture during chewing’, f = ‘flavor during chewing’, at = ‘afterswallow’, ia = ‘internal appearance’ and io = ‘internal odor’. The dotted axes indicate the separation of the samples.


Multilingual Vocabulary for Apples . . .

Generalized Procrustes Analysis

ample, positioned furthest apart for ‘fresh’ Jonagold, but closest together for ‘mealy’ Jonagold apples. Overall, the results suggested that there were no systematic differences between the different consumer panels with respect to how the samples were perceived. As predicted, the results suggested a cross-cultural consensus with respect to consumer perception of apples.

Vocabularies for apples varying in mealiness For each panel, the variation among consumers with respect to the use of individual terms was explored using curved box-plots. The most important terms pertaining to mealy texture in apples are shown in Figure 3. Like the sen-

Figure 2—GPA consensus plot from repertory grid study showing variations among the five different consumer panels. For samples, suffexes refer to mealiness condition, where 1 = ‘fresh’, 2 = ‘mid-point’ and 3 = ‘mealy’.

Differences in perception of the sensory characteristics of the apple samples were explored by GPA of the 125 consumer matrices. Permutation tests suggested the first 2 dimensions were significant and the corresponding 2-dimensional consensus configuration accounted for 64.7% of the total variation, of which 48.6% was accounted for by the 1st and 16.1% by the 2nd dimension. The separation of samples in this GPA configuration was very similar to that obtained from descriptive sensory analysis, which showed good agreement with previous sensory analysis on these varieties and storage treatments (Andani and MacFie 1998; Jaeger and others 1998; Andani and MacFie, in press). Differences between the 5 consumer panels were explored by examining the relative positions of panels round each sample (Figure 2). From the consensus space based on all 125 consumers the mean panel position for each sample was determined. No one panel differed significantly from the rest, suggesting that consumers in the different countries perceived the samples similarly. For Cox apples the position of panels were overlapping between the 3 samples. Whereas the Spanish, Danish, Flemish, and French consumers perceived the 3 Cox samples similarly, the British panel perceived them as somewhat different. A similar overlap in product positions was observed for the ‘fresh’ and ‘midpoint’ Boskoop apples. There was no consistent pattern to the way consumer panels were positioned round each sam- Figure 3—‘Curved box-plots’ of the ‘mealy’ descriptor used ple. The Danish and Spanish consumer panels were, for ex- by all five of the consumer groups.

Vol. 66, No. 6, 2001—JOURNAL OF FOOD SCIENCE

875


For Jonagold apples, ‘fresh’ and ‘mid-point’ samples were perceived as similar and different from ‘mealy’ samples, which were perceived more like ‘fresh’ Cox samples. Another axis, perpendicular to the storage treatment axis, spanned between the lower right hand and upper left hand of the corner of Figure 1a separated Cox and Jonagold from Boskoop apples. The attribute loadings plot spanned by the first 2 PCs is shown in Figure 1b. This suggested that the axis describing differences between varieties was spanned between red apple, pear-like and sweet flavor with high negative correlations and unripe apple, acid/sour, green apple and cider flavor with high positive correlations. The axis perpendicular to this, related to storage treatment, was spanned between 1st bite hardness and crisp and fibrous texture with high negative correlations and floury, granular and slimy texture, residue in afterswallow, fluffy internal appearance and stale flavor with high positive correlations. As expected, this confirmed that the sensory changes related to the storage treatment strongly affected apple texture. ‘Fresh’ apples were perceived more hard, juicy and crisp than ‘mealy’ apples, which were characterized as old, stale and floury. Cox apples were characterized by yellow internal appearance, cox-like and floral flavor and drying aftertaste. In comparison, Jonagold apples were characterized by white flesh, juicy internal appearance, watery flavor and juicy, hard and crisp 1st bite texture. The texture during chewing was characterized as pulpy and fibrous. The Multilingual Vocabulary for Apples ‘fresh’ and ‘mid-point’ Boskoop samples were characterized by unripe and cooked apple internal odor, green apple, cooked apple, cider and acid/sour flavor and astringent aftertaste. The ‘fresh’ Boskoop samples were also characterized by tough and chewy texture and dense flesh. The ‘mealy’ Boskoop and to a lesser extent ‘mid-point’ and ‘mealy’ Cox , were characterized by stale flavor, off-flavor and slimy texture.

Multilingual Vocabulary for Apples . . . sory panel, consumers also perceived the texture of Cox apples as mealy. In Spain, the 6 consumers using the term ‘harinosa’ were in good agreement with each other. The box-plot was tight suggesting that these consumers agreed on the categorization and labelling of mealy texture sensation. Similar observations pertained to the term ‘melet’ used by 9 Danish consumers and ‘farineuse’ by 8 French-Belgian consumers. In comparison, the box-plot pertaining to the term ‘melig’, which was used by 12 Flemish speaking Belgian consumers, had considerably longer whiskers. This suggested less agreement among consumers in this group on the sensory sensation associated with this term. The box-plot indicated that the term ‘melig’ was associated not only with Cox apples but also the ‘mealy’ Boskoop and Jonagold samples. This was also the case for the use of the terms ‘coarse’ and ‘dry’ by the British consumer panel. Note that the UK consumers differed from the remaining panels in not using a term for mealiness. Instead, this group used the descriptors ‘coarse’, ‘dry’ and ‘spongy’. Although multiple terms were used by all consumer panels, only among the British consumers was a single term not dominant. Accordingly, the proposition that differences between the panels with respect to the description of mealy texture in apples existed was supported. Aside from terms pertaining to mealy texture, all panels used terms to describe sweet and acidic flavor, crisp, juicy, hard/firm and soft texture. Table 1 details the multilingual consumer vocabulary. To gain further insight into the correspondence between consumer terms and the descriptors used by the trained sensory panel, the Danish and British consumer vocabularies were compared to the sensory vocabulary. This was achieved by projecting the descriptor medians from the curved boxplots of the consumer terms and the scratch that the sensory means onto the GPA consensus space (Figure 4). With respect to texture, the British consumers associated soft, coarse and spongy texture with Cox apples. Dryness appeared to be more strongly related to storage treatment. However, the exact position of the descriptor medians must be viewed with some caution owing to the limited number of consumers using individual terms and the angular spread of descriptor positions revealed by the curved box-plots. The Danish consumers associated differences in texture with the storage treatment. Generally, ‘fresh’ apples were perceived hard, crisp and juicy and ‘mealy’ apples as soft, dry, coarse and mealy. However, these descriptors were also considered characteristic of Cox apples. Both panels perceived Cox and Jonagold apples as sweet and Boskoop apples as acidic. The correspondence with the sensory vocabulary was satisfactory, in particular between the British sensory and consumer panels. Note that the absence of odor and appearance terms in the consumer vocabularies was due to the fact that consumers had been asked to elicit flavor and texture attributes only

Table 1—Multilingual consumer vocabulary for dessert apples. a England Acidic 20 Sweet 15 Crispy 11 Juicy 7 Hard/ 7 Firm Soft 6 Bitter 7 Cooking 4 Apple Tasteless 5 Mealy/ Floury Coarse 5 Dry 4 Spongy 5

Denmark Sur Sød Sprod Saftig Hård

Belguim/ Flemish

19 17 9 5 14

Zuur Zoet Krisp Sappig Hard

Blød

9

Zacht

Melet

9

Grov Tør

5 3

20 13 3 5 11

Belgium/ French

Spain

Acide 22 Sucree 14 Croquante 4 Juteuse 10 Ferme 13

Acido 23 Dulce 10 Crujiente 1 Juguso 4 Dura 8

5 Molle 7 Blanda 5 Amere 3 Amargor 4 Stoofappel 4 Pour la 4 Cuisine Fade 13 Insipido 11 Melig 12 Farineuse 8 Harinosa 6 Granuleuse 4 Aspera

3

Droog 3 Loose 5 Structuur

a Based on RGM interviews with 25 consumers in each language. Descriptors correlate significantly (r2 > 0.5) with GPA consensus. Numbers refer to the

total number of subjects using each descriptor. Only descriptors used by 3 or more (> 12%) consuemrs are shown.

tures, there is little evidence to suggest cultural differences in perception of intensity (Lundgren and others 1986), detection thresholds (MacDonald and others 1993) and discriminative ability (Laing and others 1993). While previous work has almost exclusively concentrated on the perception of basic tastes, this study is among the first to suggest that perception is also cross-cultural with respect to ordinary foods. However, a number of factors may have contributed to this finding. Notably, the experimental approach, the inherent sensory differences between the apple varieties and those induced by the storage treatment may have contributed to consumers perceiving the apple samples similarly. The

Discussion Consumer perception of apples


Aside from a few minor differences, perception of the sensory qualities of the apples were similar among consumers. As predicted and in accordance with previous findings, this suggested that sensory perception was, by and large, similar among European consumers. Cross-cultural differences in chemosensory perception have previously been studied with respect to psychophysical responses, detection and discrimination thresholds of sensory qualities. While Figure 4—Projection of sensory vocabulary and consumer taste sensitivity to bitterness is known to vary across cul- descriptors onto GPA consensus map (with r2 . 0.5).


RGM interviews were specifically designed to focus the consumers’ awareness on textural differences. Aiming to minimize differences in flavor, 3 of the 6 triads presented during attribute elicitation were composed of samples from 1 variety only. The storage treatment, which consisted of storing ‘fresh’ apples at room temperature and high humidity for 3 wk, resulted in ‘mealy’ samples that were very different from ‘fresh’ samples and of a nature that would not meet modern quality control standards. This factorial design (3 varieties 3 3 storage treatment conditions) was strongly reflected in the sensory descriptive analysis, where apple samples were separated with respect to variety on the 1st principal component and texture and storage treatment on the 2nd. Although the importance of the above factors in relation to the obtained results remains unclear, it was not unexpected that GPA of the consumer data resulted in a highly significant 2-dimensional consensus configuration similar to that obtained from sensory profiling. Generalizing the present results to food products in general may, therefore, be questionable. Further evidence obtained from a comparison of more different and complex products is required. Interestingly, the 2-dimensional GPA consensus plot indicated that the British consumers may have perceived greater differences between the 3 Cox samples than consumers in the other countries. In a similar vein, a comparison of the British and Danish consumer vocabularies suggested that the axis separating samples according to texture and storage treatment in the Danish vocabulary was spanned between hard, juicy and crisp, which were negatively correlated to mealy, coarse, dry and soft. In comparison, the axis pertaining to texture in the British vocabulary separated Cox from Jonagold and Boskoop apples. It cannot be ruled out that this greater sensitivity among the UK consumers toward Cox apples is a reflection of the strong market position and popularity of Cox apples in the UK (Seaton 1996). This may tentatively point to familiarity exerting an influence on perception and sensory characterization. Exploring the umami taste concept among the US and Japanese subjects, Ishii and O’Mahony (1990) have previously suggested an interaction between familiarity and concept measurement.

Sensory vocabularies for mealy apples As may have been expected, a comparison between the consumer vocabularies and the trained panel sensory vocabulary revealed that the sensory panel characterized the apple samples in somewhat greater detail than the consumers. That the consumers were more simplified in their description of apples than the sensory panellists was particularly evident for texture attributes. For example, the sensory panel associated the texture of Boskoop apples as tough, chewy and dense, characteristics generally not mentioned by the consumers. Also, both the Danish and British consumers perceived crispness and juiciness broadly similar whereas the sensory panel did not. Some confusion of bitter and acidic flavor was also evident among the consumers. For example, the British consumers characterized Boskoop apples as acidic, bitter and cooking apple. However considering that untrained assessors commonly confuse perception of bitter and acidic (Amerine and others 1965; Robinson 1970; McAuliffe and Meiselman 1974), this association was not entirely surprising. Overall, there was good agreement between the trained panel and consumer vocabularies. Both types of vocabularies adequately captured terms characterizing mealy texture and the sensory characteristics are known to drive

consumer preference for apples (Dalliant-Spinnler and others 1996; Jaeger and others 1998; Jaeger and others, in press). Perception of mealy texture was characterized by more than 1 descriptor by all panels. Among all the consumer panels, except the British, a single term was dominant. In this respect, it was particularly interesting to note that these terms (Flemish: melig, Danish: melet, French: farineuse, Spanish: harinosa) all translated into mealy and/or floury in the English language. However, neither the British panel performing the sensory profiling nor the British consumers used the term mealy texture. The British consumers used the 3 descriptors dry, coarse and spongy to characterize this textural sensation. The sensory panel, on the other hand, used the 2 highly correlated attributes granular and floury texture. Usage of the latter rather than mealy for sensory profiling has been noted elsewhere (Harker and others 1997). Overall, the present results suggest that mealiness is an umbrella term covering the textural sensation is associated with floury, coarse, dry and soft texture in apples. Although comparable in aiming to overcome problems associated with the translation of sensory descriptors (Zannoni 1997), the present multilingual product vocabulary differs from previous vocabularies in the literature on 2 factors. Firstly, it was derived using untrained assessors. Differences in perception (Cardello and others 1982) and aspects of information processing of sensory stimuli (Jaeger and others, in press) among trained panellists and consumers mean that a vocabulary derived from consumers rather than sensory judges is more likely to provide a better representation of product characteristics as perceived by those buying and consuming the products. Also, by using consumer terminology in marketing communications problems with respondents misunderstanding what is being communicated may be minimized. For example, the terms floury and granular texture, used by the sensory panel, may not be readily understood by British consumers. It is more likely that the words included in the British consumer vocabulary will. A second point of difference between the present approach and those previously reported is that the multilingual vocabulary has been derived without the need to translate terms from one language to another or impose terms used by one panel on another. These issues have previously been considered problematic (Risvik and others 1992; Hirst and others 1994; Nielsen and Zannoni 1998). Similarly, Drake (1989) encountered a number of problems when developing a polygot list of terms for the textural properties of food. Asking collaborators from 22 countries to translate English terms into their own languages several difficulties arose, for example that some words implied several meanings. From a methodological point of view, the study confirmed earlier reports (Raats and Shepherd 1993) on the usefulness of repertory grid methodology in cross-cultural research. Considering the continued interest in and necessity of cross-cultural research (Tuorila 1998; Douglas and Craig 1997; Sojka and Tanshuaj 1995), there is a need to establish the cross-cultural validity of existing methodologies, as well as developing new tools. The curved box-plots used in the present study is an example of a new tool. They allowed an in-depth examination of the agreement among consumers within the same panel using a common descriptor. Also, they made it possible to identify outlying observations, i.e., consumers using a construct significantly different from the rest. In a cross-cultural context, curved box-plots allow a comparison of the labels consumers from different countries at-


877


Multilingual Vocabulary for Apples . . .

Multilingual Vocabulary for Apples . . . tach to the same sensory sensations. As advocated by Zannoni (1997), this focuses more attention on the perceived reality and less on the verbal representation of this reality.

Conclusions

I

N ACCORDANCE WITH ACCUMULATING EVIDENCE IN THE LIT-

erature, a cross-cultural consensus with respect to consumer perception of the sensory properties of apples was established. In particular, the present study demonstrated how 5 European consumer groups perceived the differences between apples varying in mealiness similarly. Secondly, using repertory grid methodology, a multilingual vocabulary for apples was established. In comparison to other international sensory vocabularies in the literature, this was derived using untrained assessors hereby providing insight into the terms consumers use to characterize eating apples. While less detailed than the vocabulary used by the trained sensory panel, it adequately captured terms pertaining to mealiness. In particular, it was characterized as the sensation associated with floury, coarse, dry and soft texture in apples. The terms known to drive consumer preference for apples were also included.

References


Amerine MA, Pangborn RM, Roessler EB. 1965. Principles of Sensory Evaluation of Food. Academic Press. New York. p 602. Andani Z, MacFie HJH. 1998. Consumer preferences expectations and quality perceptions of dessert apples. Postharvest News and Information. 9(3):39-44. Andani Z, De Smedt V, Nicola B. 1999. Development of mealiness in apples under shelf-life conditions Food Science and Technology Today: In press. p 203204. Andani Z, MacFie HJH. 1999. Consumer Preference in Fruit and Vegetable Quality: An Integrated View. Shewfelt RL, Brÿckner B. Eds. Technomic Press Co. Inc. USA. In press. p 2000, 158-177. Arnold GM, Payne RW. 1989. Procedure GENPROC in Payne RW, Arnold GM. Eds. Genstat Procedure Library Manual: Release 13(2) Oxford: Numerical Algorithms Group. p 76-79. Arnold GM, Williams AA. 1986. The use of Generalized Procrustes Analysis in Statistical Procedures in Food Research Piggott JR. Ed. London. Elsevier Applied Science. p 233-253. Askegaard S, Madsen TK. 1995. European Food Cultures: an Exploratory Analysis of Food Related Preferences and Behaviour in European Regions. MAPP Working Paper Nr 26 The Aarhus School of Business. Aarhus Denmark. Barreiro P, Ortiz C, Ruiz-Altisent M, De Smedt V, Schotte S, Andani Z, Wakeling I, Beyts PK. 1998. Comparison between sensory and instrumental measurements for mealiness assessment in apples A collaborative test. J Texture Studies. 29: 509-525. Ben-Arie R, Sonego L. 1980. Pectolytic enzyme activity involved in woolly breakdown of stored peaches. Phytochem. 19: 2553-2555. Ben-Arie R, Kislev N, Frenkel C. 1979. Ultrastructural changes in the cell walls of ripening apple and pear fruit. Plant Physiol. 64: 197-202. Bhanji Z, Earthy P, Wakeling I, MacFie HJH. 1997. Unpublished data. Mealiness in fruit: consumer perception and means for detection. FAIR project CT95-302 First Progress Report Jan ’96 - Dec ’97. Cardello AV, Maller O, Kapsalis JG, Segars RA, Sawyer FM, Murphy C, Moskowitz HR. 1982. Perception of texture by trained and consumer panelists. J Food Sci. 47(4):1186-1197. Cardello AV. 1993. Cross-cultural sensory testing: a changing tide? Cereal Foods World. 38:699-701. Christensen CM. Vickers ZM. 1981. Relationships of chewing sounds to judgments of food crispness. J Food Sci. 46:574-578. Dalliant-Spinnler B, MacFie HJH, Beyts PK, Hedderley D. 1996. Relationships between perceived sensory properties and major preference directions of 12 varieties of apples from the southern hemisphere. Food Quality and Preference. 7 (2):113-126. Douglas SP, Craig CS. 1997. The changing dynamic of consumer behaviour: implications cross-cultural research. International J Research in Marketing. 14:379-395. Drake BK. 1963. Food crushing sounds: an introductory study. J Food Sci 7:123130. Drake BK. 1989. Sensory textural/rheological properties: A polygot list. J Texture Studies. 20 (1):1-27. Druz LL, Baldwin RE. 1982. Taste thresholds and hedonic responses of. panels representing three nationalities. J Food Sci. 47 (2):561-563, 569. ESN. 1996. A European Sensory and Consumer Study. A Case Study on Coffee. European Sensory Network. McEwan JA. Ed. Campden and Chorleywood. Food Research Assn.. Chipping Campden. Gloustershire UK. Gains N, Thomson DMH. 1990. The relation of repertory grid/generalized procustes analysis solutions to the dimensions of perception: application to Munsell colour stimuli. J Sensory Studies. 5:177-192. Gains N. 1994. The repertory grid approach in Measurement of Food Preferences.


MacFie HJH, Thomson DMH. Ed. Blackie Academic and Professional Glasgow. p 51-75. Genstat. 1993. Reference Manual version 53. Clarendon Press Oxford. Gower JC. 1975. Generalized Procustes Analysis. Psychometrika. 40 (1):33-51. Griffin Abbie Hauser JR. 1993. The voice of the customer Marketing Science. 12 (1):1-27. Harker FR. Hallett IC. 1992. Physiological changes associated with development of apple fruit during cool storage. HortScience. 27 (12):1291-1294. Harker FR, Sutherland PW. 1993. Physiological changes associated with fruit ripening and the development of mealy texture during storage of nectarines. Postharvest Biological Technol. 2:269-277. Harker FR, Redgwell RJ, Hallett IC, Murray SH, Carter G. 1997. Texture of fresh fruit. Horticultural Reviews. 20:121-224. Hirst DJ, Muir DD, Naes T. 1994. Definition of the sensory properties of hard cheese: a collaborative study between Scottish and Norwegian panels. International Dairy Journal. 4:743-761. Hunter E, Anthony, McEwan JA. 1998. Evaluation of an international ring trial for sensory profiling of hard cheese. Food Quality and Preference. 9 (5):343355. Ishii R, O’Mahony M. 1987. Defining a taste by a simple standard: aspects of salty and umami tastes. J Food Sci. 52 (5):1405-1409. Ishii R, O’Mahony M. 1990. Group taste concept measurement: verbal and physical definition of the umami taste concept for Japanese and Americans. J Sensory Studies. 4 (4):215-227. Jaeger SR, Andani Z, Wakeling IN, Macfie HJH. 1998. Consumer preference for fresh and aged apples: a cross-cultural comparison. Food Quality and Preference. 9(5):355-366. Jaeger SR, Wakeling IN, Macfie HJH. 1999. Behavioural extensions to preference mapping: the role of synthesis. Food Quality and Preference: In press. p 2000, 349-359. Karahadian C. 1995. Impact of global markets on sensory testing programs. Food Technol. 47(2):77-79. Kelly GA. 1955. The Psychology of Personal Constructs. Norton. New York. Krzanowski WJ. 1988. Principles of Multivariate Analysis: A Users Perspective. Oxford Statistical Series 3. Oxford Signs Publishers. Oxford. Laing DG, Prescott J, Bell GA, Gillmore R, Allen S, Best DJ, Yoshida M, Yamazaki K, Ishii-Mathews R. 1993. A cross-cultural study of taste discrimination with Australians and Japanese. Chemical Senses. 18 (2):161-168. Lapsley KG, Escher FE, Hoen E. 1992. The cellular structure of selected apple varieties. Food Structure. 11:339-349. Lundgren B, Pangborn RM, Daget N, Yoshida M, Laing DG, McBride RL, Griffiths N, Hyvonen L, Sauvageot F, Paulus K, Barylko-Pikienla N. 1986. An interlaboratory study of firmness aroma and taste of pectin gels. Lebensmittel Wissenschaft und Technologie. 19 (1):66-76. MacDonald BA, Watts BM, Fitzpatrick DW. 1993. Canadian and Peruvian gustatory response to bitterness and salt in instant potato purée. Ecology of Food and Nutrition. 30 (2):127-136. MacFie HJH, Bratchell N, Greenhoff H, Vallis LV. 1989. Designs to balance the effects of order of presentation and first order carryover effects in hall tests. J Sensory Studies. 4:129-148. McAuliffe WK, Meiselman HL. 1974. The roles of practice and correction in the categorization of sour and bitter taste qualities. Perception and Psychophysics. 16:242-244. McEwan JA, Hallet EM. 1990. A guide to the use and Interpretation of Generalized Procrustes Analysis. Technical Manual Nr 30. Campden and Chorleywood. Food Research Assn.. McEwan JA, Thomson DMH. 1989. The repertory grid method and preference mapping in market research: a case study on chocolate confectionary. Food Quality and Preference. 1(2):59-68. Mojet J, de Jong S. 1994. The sensory wheel of virgin olive oil. Grasas y Aceites. 45:42-47. Nicolai B. 1998. Mealiness in fruit. F-FE 278/98 FLAIR-Flow Europe Project Bulletin http://expinterspeednet/flair/ffe27898htm. Nielsen RG, Zannoni M. 1998. Progress in developing an international protocol for sensory profiling of hard cheese. International J Dairy Technol. 51 (2):5764. Olsen JC. 1981. The importance of cognitive processes and existing knowledge structures for understanding food acceptance. Criteria of Food Acceptance. Solms J, Hall RL. Ed. Zurich: Forster Verlag AG. 69-81. O’Mahony M. 1990. Cognitive aspects of difference testing and descriptive analysis: criterion variation and concept formation. Psychological Basis of Sensory Evaluation. McBride RL, MacFie HJH. Ed. Elsevier Applied Science. London. p 117-139. Piggott JR, Watson MW. 1992. A comparison of free-choice profiling and the repertory grid method in the profiling of cider. J Sensory Studies. 7 (2):133-145. Prescott J, Bell G. 1995. Cross-cultural determinants of food acceptability: recent research on sensory perceptions and preferences. Trends in Food Science and Technol. 6 (6):201-205. Prescott J, Bell GA, Gillmore R, Yoshida M, Korac S, Allen S, Yamazaki K. 1997. Cross-cultural comparisons of Japanese and Australian responses to manipulations of sweetness in foods. Food Quality and Preference. 8 (1):45-55. Prescott J, Bell GA, Gillmore R, Yoshida M, O’Sullivan M, Korac S, Allen S, Yamazaki K. 1998. Cross-cultural comparisons of Japanese and Australian responses to manipulations of sourness, saltiness, and bitterness in foods. Food Quality and Preference. 9 (1/2):53-66. Prescott J, Laing D, Bell G, Yoshida M, Gillmore R, Allen S, Yamazaki K, Ishii R. 1992. Hedonic responses to taste solutions: A cross-cultural study of Japanese and Australians. Chemical Senses. 17 (6):801-809. Prescott J. 1998. Comparisons of taste perceptions and preferences of Japanese and Australian consumers: overview and implications for cross-cultural sensory research. Food Quality and Preference. 9 (6):393-402.

Multilingual Vocabulary for Apples . . . Vincent JFV. 1989. Relationships between density and stiffness of apple flesh. J the Science of Food and Agriculture. 47:443-462. Wakeling IN, Raats MM, MacFie HJH. 1992. A new significance test for consensus in Generalized Procrustes Analysis. J Sensory Studies. 7 (2):91-96. Wilkinson BG. 1965. Some effects of storage under different conditions of humidity on the physical properties of apples. J Horticultural Science. 40:58-65. Zannoni M. 1997. Approaches to translation problems of sensory descriptors. J Sensory Studies. 12 (3):239-253 Ms. 19991060 The authors would like to thank colleagues at the Institute of Food Research (UK), the Biotechnological Institute (Denmark), the Catholic Univ. of Leuven (Belgium) and the Instituto de Agroquimica y Tecnologia de Alimentos (Spain) for assisting with the consumer trials. Pam Beyts (Sensory Dimensions, Reading) is thanked for the use of her trained panel and other facilities. The work was supported by the EU FLAIR project CT95-302 (Z. Andani, I.N. Wakeling and H.J.H. MacFie) and the Danish Research Academy (S.R. Jaeger). Helpful comments from Roger Harker are gratefully acknowledged.

Authors were with the Institute of Food Research, Reading Laboratory, Department of Consumer Science, U.K. Author Andani is now with Sensory Dimensions, Science and Technology Centre, Earley Gate, Whiteknights Road, Reading, RG6 6BZ, U.K. Author Jaeger is now with the Dept. of Agricultural and Food Economics, The Univ. of Reading, Reading, U.K. Author Wakeling is with QI Statistics, Norfolk, U.K. Author MacFie is of 43 Manor Road, Keynsham, Bristol, U.K. Direct inquiries to author Andani (E-mail: [email protected]).


879


Raats MM, Shepherd R. 1993. The use and appropriateness of milk in the diet: a cross-country evaluation. Ecology of Food and Nutrition. 30 (3):253-273. Risvik E, Colwill JS, McEwan JA, Lyon DH. 1992. Multivariate analysis of conventional profiling: a comparison of a British and Norwegian trained panel. J Sensory Studies. 7:97-118. Robinson JO. 1970. The misuse of taste names by untrained observers. British J Psychology. 61:375-378. Scriven FM, Gains N, Green SR, Thomson DMH. 1989. A contextual evaluation of alcoholic beverages using the repertory grid method. International J Food Sci and Technol. 24:173-182. Seaton L. 1996. Apples A UK Market Perspective Fresh Product Report 1. Lockwood Press Ltd. London. Sojka JZ, Tansuhaj PS. 1995. Cross-cultural consumer research: a 20-year review. Advances in Consumer Research. 22:461-474. Stone H, Sidel JL. 1995. Strategic applications for sensory evaluation in a global market. Food Technol. 47(2):80-89. Szezesniak AS, Kahn EL. 1971. Consumer awareness of and attitudes to food texture I Adults. J Texture Studies. 2:280-295. Thomson DMH, McEwan JA. 1988. An application of the repertory grid method to investigate consumer perceptions of foods. Appetite. 10:181-193. Tukey JW. 1977. Exploratory Data Analysis. Addison-Wesley. Reading MA. Tuorila H. 1998. Cross-cultural sensory and consumer research on foods Food Quality and Preference. 9 (6):383. Vickers ZM. 1981. Relationships of chewing sounds to judgments of crispness crunchiness and hardness. J Food Sci. 46:121-124. Vickers ZM, Bourne MC. 1976. A psychoacoustical theory of crispness. J Food Sci. 41:1158-1164.