PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 1 SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS L. PERRIN 1,2, R. SYMONEAUX 1, I. MAITRE 1, C. ASSELIN 2, J. PAGES 3 et F. JOURJON 1 1 : UMT Vinitera - Laboratoire GRAPPE Groupe ESA - 55 rue Rabelais 49 007 Angers France E-mail : l.perrin@groupe-esa.com Tél : 00 33 2 41 23 55 55 2 : Interloire 73 rue Plantagenêt BP 52327 49 023 Angers - France 3 : Laboratoire de mathématiques appliquées Agrocampus Rennes / IRMAR 65 rue de Saint Brieuc CS 84215-35 042 Rennes cedex - France Study presented orally at the 2007 OIV Congress. Abstract In winemaking, the sensorial analysis of wines is continually applied more frequently. To characterize a product, sensorial analysis is often done by a panel of trained judges that complete a conventional profile. For this type of method the judges must follow a long training phase, which can generally even last several months. For wine in particular, professionals are usually not trained together and their performance is not verified, however they are accustomed to wine tastings and describing wines. In contrast to conventional profiles, spontaneous methods, such as the Free Choice Profile or Napping do not need a common preestablished training and give the taster more freedom, especially in descriptor choices. These methods therefore seem suitable to use for a jury of professionals. This study aims to evaluate the potential interest of Free Profile and Napping analyses completed by professionals (winemakers, vineyard managers, technicians) in comparison to conventional profiles completed by a trained panel. These methods were completed on 10 white wines from the Loire Valley. The results show that the free profile and conventional profile are very close in terms of the general characterization. Result interpretation is simpler for the conventional profile but the free profile is more suitable for the professionals. The Napping method is also suitable but it gives a wine representation that is slightly different because of its selective nature: only the most important criteria for the tasters are considered. INTRODUCTION In winemaking the sensorial characterization of wines is continually more frequently applied, in order to determine the impacts of different technical methodologies, to select the most adapted yeast strain (Gerland and Dumont, 2000), or to illustrate the sensorial differences between different terroirs (Fischer, Roth and Christmann, 1999). In the agri-food industry, the reference method used for sensorial characterization is the conventional profile (Lawless and Heymann, 1998). This method makes use of a panel of trained judges. The objective of training is to create a common expertise among all the panel judges and to verify their sensorial performance. The particularity of this approach is that there is no qualitative judgment about the wines. There are no subjective descriptors (harmony, balance, quality ) where the technical experience or the taster s origin could influence the responses and prevent a strictly objective evaluation of the wine characteristics. The training phase for the judges (training and establishment of common vocabulary) generally spans over several months. In the specific case of wine, professionals have not been trained together and their performance has not been verified. They are however accustomed to tasting and describing wines. Each professional has their own expertise and generally their own set of vocabulary terms which, correspond more or less to the notions of the other tasters. Some pay close attention to certain characteristics or qualities of the wines whereas others will focus on other elements. Methods said to be spontaneous such as the Free Choice Profile (Williams and Langron, 1984, Gerland and Dumont, 2000), the Flash Profile (Dairou and Sieffermann 2002, Delarue and Sieffermann 2004), or Napping (Pagès, 2003 and 2005) start from this concept and use the complementary results between the tasters while allowing them to have complete freedom to characterize the wines. There is no need to establish a common mindset beforehand and the taster is given a lot of freedom, particularly in criteria choice. Moreover, the associated statistical evaluations such as the Multiple Factorial Analysis (Escofier and Pagès, 1998) take into account the differences between the individuals and each judge has an equivalent weight in the final configuration.
PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 2 The objective of this study is to evaluate the potential interest of professional descriptions when compared to the classic methodology used for sensorial analysis (Lawless and Heymann, 1998). This study compares Free Profile and Napping analyses completed by professionals (winemakers, vineyard managers, technicians) with conventional analyses completed by a trained panel, using 10 white wines from the Loire Valley. MATERIALS AND METHODS The Products The product selection consisted of ten dry and semi-dry white wines from the Loire Valley, the grape variety for all was Chenin : one Vin de Pays (VDP), three Anjou Blanc (ANJ_A, ANJ_B and ANJ_C), four Saumur blanc (SAU_A, SAU_B, SAU_C and SAU_D) as well as two Savennières (SAV_A and SAV_B). These wines were chosen by professionals and represent the wine type diversity found in these appellations. Data Collection Conventional Profile The conventional profile was completed by a trained panel of the GRAPPE laboratory (ESA), consisting of 17 judges. The descriptors were established by the panel on the product selection under study (the ten wines for this study). A reduction in the number of terms was completed on consensus (Lawless & Heymann, 1998): only the descriptors that the majority of panel judges agreed upon were kept. Once the list was established, each judge gave a score to each wine and for each descriptor in the common list. The judges used linear non structured scales, this being a validated method (Jourjon, Symoneaux, Thibault & Réveillère, 2005). The wine were tasted one at a time and following an order based on Williams Latin Square. Free Profile The Free Profile was completed by twelve professionals from the Loire Valley region (winemakers, vineyard managers, technicians) in one session only. The wines were presented simultaneously. Each judge was instructed to taste all of the products and to formulate descriptors which allowed them to discriminate the products. Therefore, each judge established their own list of descriptors. The product series was then presented to the judges again. In order to balance out order and report effects, the wines were presented one at a time in an order based on Williams Latin square. The judges gave a score to each product for each of the previously established descriptors, on linear non structured scales. Napping Napping is a recent method, adapted and updated by Pagès, 2003 and 2005. The same twelve professionals as before complete the Napping during another session. The wines were presented simultaneously. The judges had to position the products on a paper table cover according to their similarity or dissimilarity. The closer the wines were judged, the closer they were placed to one another. On the other hand, the wines judged as being very different were further apart on the table cover. Every taster based their judgment according to their own criteria. Once the wines were positioned on the table cover and the product codes copied down, the judges were asked to write down next to the wines some terms that they considered characterized that particular wine or group of wines. Data Analysis Each data group was analyzed separately. The conventional profile data was evaluated by a classic method, the Principal Component Analysis (PCA). The data of the Free Profile and Napping were evaluated by Multiple Factor Analysis (MFA) (Escofier & Pagès, 1998). This statistical tool balances out the role of each judge in the analysis and results in an average configuration of the products.
PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 3 RESULTS AND DISCUSSION For each of the tested methods the results permitted for the sensorial characterization of the wines in this study. The multidimensional statistical tools used gave graphical representations or sensorial charts which represented the wines and the characterizing descriptors. Using these tools it was possible to summarize the main differences found between the products. Due to the large number of descriptors for the Free Profile and Napping, (228 and 156 descriptors, respectively), correlation circles are not presented. Nonetheless, the correlation values between each of the descriptors and the axes were studied and applied to identify descriptors associated with the wines found at the extremes. Dimension 2 (21.15%) -6-4 -2 0 2 4 6 ANJ_A ANJ_B SAU_C SAU_B SAV_A VDP SAU_A ANJ_C SAU_D SAV_B Figure 1: Wine representation according to conventional profile (PCA) -5 0 5 10 Dimension 1 (48.19%) ANJ_C SAU_D Dim 2 (16.49 %) -4-2 0 2 SAU_A SAV_A SAU_C ANJ_A VDP SAU_B SAV_B ANJ_B Figure 2: Wine representation according to Free Profile (MFA) -2 0 2 4 6 Dim 1 (28.83 %)
PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 4 Dim 2 (19.15 %) -4-2 0 2 ANJ_C SAU_C VDP SAU_A SAU_B SAV_A ANJ_A ANJ_B SAV_B SAU_D Figure 3 : Wine representation according to Napping (MFA) -2 0 2 4 Dim 1 (40.65 %) The figures corresponding to the three methods tested all clearly placed the SAV_B and SAU_D wines in opposition to the other wines under study. These two wines are close in sensorial terms, they have a more substantial colour, are oaky, lactic and empyreumatic. They also have spicy, exotic fruit and dried fruit notes and are more alcoholic and round. The professionals further added to this description by highlighting the complexity of these wines and their maturity. This complexity can be associated to a notable number of odour and aromatic descriptors used by the expert panel during the sensorial analysis to characterize these wines. The ANJ_B is also a wine that is notably differentiated during the different tastings. It is distinguished from the others by its chemical and medicinal notes, and some professionals identified it as having a defect. For the trained panel sensorial profiles and the professionals Free Profiles, the characterizations of the ANJ_C and SAU_A are also pretty close: these wines are both slightly sparkling with a bit of CO2 gas. The professionals commented that they were lively wines and the trained panel found that they had citrus notes. As for the Napping, these two wines are not very differentiated from the others. The slightly sparkling character of these wines is not evident in the results. However, the results of the Free Profile demonstrate that these professionals are able to identify this aspect. This difference can therefore be attributed to the method and not to a lesser sensibility of the jury. The residual sugars present in the VDP and ANJ_A are perceived by all the tasters and appeared on a third sensorial dimension (chart not shown here), for all the methods. With the Free Profile and the Napping, it is difficult to complete a precise characterization of the products. Whereas, the conventional profiles completed by the panel trained in sensorial analysis permit the use of inferential statistics that allow for a more precise characterization. Hence, using variance analyses (ANOVA) and comparison of means tests completed descriptor by descriptor, it is possible to show the different nuances of wines that are deemed close overall by the other methods. SAV_B is perceived as sweeter and slightly less sparkling than SAU_D. SAU_A is distinguished slightly from ANJ_C by more notable acidity and minerality. ANJ_A and VDP are evaluated as close for their sweetness, white fruit notes, low acidity, low bitterness, non aggressive initial taste and low mineral aroma. The VDP is however distinguished by its roundness, its honey and lactic notes which are not observed in the ANJ_A.
PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 5 CONCLUSION The conventional profile is a laborious method that requires a lot of time, hence prolonging the time needed to complete a study. It is based on the use of objective descriptors, not qualitative, for the characterization. The Free Profile and Napping are much faster methods and are therefore more suitable to the time constraints of professionals. They permit the use of vocabulary that is individual and personal for each taster, without limits. This spontaneity is coupled with the individual opinions during the data analysis so as to not forget any sensorial dimensions. Despite the evident interest of these two methods, they are limited due to a lack of precision in the characterizations. Spontaneous methods are also methods that are quite cumbersome to analyze given the diversity of vocabulary used by the tasters. With the Free Profile, the wines were described using a total of 228 variables and the number of words associated with the Napping was 156. The results show that these methods allow for a global characterization of the wines, but that it is difficult to zoom in and identify specific nuances that differentiate the wines. On the other hand, the conventional profiles can be evaluated using inferential statistical methods and therefore offers a more precise characterization. Finally, the global approach used in Napping allows for each judge to construct their own representation of the products based on the sensorial dimensions that they consider to be important. The fact that the tasters did not put forward the sparkling nature as a distinguishing factor using this method seems to demonstrate that perhaps it is possible to have an underlying hierarchy of sensorial criteria. The Napping could therefore possibly be used to measure the influence of different technical elements (vintage effect, grape variety, producer, etc) upon each other. Given this first trial, it is possible to distinguish two types of approaches requiring two different methods. In a situation where a global characterization of a large number of products based on the identification of main characteristics and the grouping of sensorially similar products is needed, spontaneous methods can be of great interest and are well suited to the professional environment. However when a more fine, precise characterization is needed, notably following an experimental protocol or in research linking chemico-physical and sensorial analyses, a complete conventional profile would be favourable. Keywords: Sensorial Analysis, Free Profile, Napping, Conventional profiles, experts
PERRIN ET AL., SPONTANEOUS METHODS FOR WINE SENSORIAL ANALYSIS, P. 6 Bibliography Dairou, V., & Sieffermann, J.-M. (2002). A comparison of 14 jams characterized by Conventional Profile and a quick original method, the Flash Profile. Journal of Food Science, 67 (2), 826-834. Delarue, J., & Sieffermann, J.-M. (2004). Sensory mapping using Flash profile. Comparison with a conventional descriptive method for the evaluation of the flavour of fruit dairy products. Food Quality and Preference, 15 (4), 383-392. Escofier, B. & Pagès, J. (1998). Analyse factorielles simples et multiples. Dunod. Fischer, U., Roth, D., Christmann, M. (1999). The impact of geographic origin, vintage and wine estate on sensory properties of Vitis vinifera cv. Riesling wines. Food Quality and Preference, 10, 281-288. Gerland C. et Dumont Ann, (2000). Analyse sensorielle descriptive libre choix (F.C.P.) : utilisation pratique. Revue Française d œnologie, 182, 36-39. Jourjon, F., R. Symoneaux, C. Thibault Et M. Réveillère. 2005. Comparaison d'échelles de notation utilisées lors de l'évaluation sensorielle de vins. J. Int. Sci. Vigne Vin. 39:23-29. Lawless, H. T. and Heymann, Hildegarde (1998). Sensory evaluation of food - Principles and practices. Kluwer Academic/Plenum Publishers. New York. pp 362-378. Pagès, J. (2003). Recueil direct de distances sensorielles : application à l'évaluation de dix vins blancs du Val-de-Loire. Sciences des Aliments, 23, 679-688. Pagès, J. (2005). Collection and analysis of perceived product inter-distances using multiple factor analysis: application to the study of 10 white wines from the Loire Valley. Food Quality and Preference, 16 (7), 642-649. Perrin, L., Jourjon, F., Symoneaux, R., Maître, I., Asselin, C., Pagès, J. (2006b) Comparison between a conventional profiling and a free profiling carried out by professionals, using Multiple Factorial Analysis. 8th Sensometrics Meeting Imagine the senses, 2-4 August 2006, Ås, Norway (Communication orale) Perrin, L., Maître, I., Symoneaux, R., Jourjon, F., Asselin, C., Pagès, J. (2006c). Comparison of three sensory methods to explicit Napping : Case of ten wines from Chenin grape variety. A Sense of Diversity Second European Conference on Sensory Consumer Science of Food and Beverage, 26-29 september 2006, The Hague, Netherlands (Poster). Williams, A.A. and Langron, S.P. 1984. The use of free-choice profiling for the evaluation of commercial ports. J. Sci. Food Agric. 35:558-568.