ANALYSIS
Pearson correlation coefficients were calculated between the measures of aroma intensity, acidity, body, and liking. Next, a repeated-measures multivariate analysis of variance (RM-MANOVA) was conducted with wine type (Garnacha, Pinot Noir) and sound condition (silent, 100 Hz, 1000 Hz) as within-participant measures (SPSS, version 23.0, IMB Corp., Armonk, NY, USA). The model included body, acidity, aromatic intensity, and liking as measures (dependent variables). Furthermore, wine familiarity (taken as an average of wine expertise and wine drinking frequency), was introduced as a between-participants factor. Wine familiarity was calculated by taking the average of wine expertise and wine drinking frequency values, and splitting the population into those with low familiarity (n = 21) and high familiarity (n = 29) (see Fig. 3). Follow-up univariate ANOVAs were conducted on dependent variables where there was a significant main effect or interaction effect amongst the independent variables.
RATA intensity data were analysed using two-way Analyses of Variance (ANOVA) (sample as fixed and panellist as random factor), treating the data as continuous data (a non-selected attribute was treated equivalent to ‘not perceived’ and assigned an intensity of 0), with a post-hoc Fisher’s Least Significant Difference (LSD) test and principal component analysis (PCA). Consumer acceptance data were analysed using ANOVA with a post-hoc Tukey’s test. Data were analysed using Senpaq v5.01 (Qi Statistics, Theale, UK) and XLSTAT Version 2016.03.31333 (Addinsoft, New York, NY, USA). All statistical analyses were performed at 5% level of significance.
RESULTS
Significant correlations were found between ratings of aroma intensity, acidity, body, and liking in the main study (see Table 1). Notably, all pairwise correlations were positive. For instance, wine liking was positively correlated with aroma intensity, acidity, and body. Moreover, perceived body was positively correlated with aromatic intensity and acidity.
The mean values of the participants’ wine ratings for both types of wine and all three sound conditions are shown in Chapter 4, Figure 4.2. Overall, the RMMANOVA revealed a significant main effect of wine type [F (4, 45) = 9.21, p < 0.0005, Wilks’ Lambda = 0.55], as well as a significant interaction effect between wine type and sound condition [F (8, 41) = 2.36, p = 0.034, Wilks’ Lambda = 0.68]. However, there was no significant main effect of sound condition [F (8, 41) = 1.70, p = 0.13], nor of wine familiarity [F (4, 45) = 2.38, p = 0.07] on the dependent measures (body, acidity, aromatic intensity, and liking). No significant differences were noted between the responses of the Sydney and Oxford participant groups.
In terms of wine type, follow-up univariate tests revealed that there were significant differences between the two wines in terms of their aroma intensity [F (1, 48) = 24.42, p < 0.0005, η2 = 0.34] and participants’ liking ([F (1, 48) = 6.72, p = 0.013, η2 = 0.12]. Overall, the Pinot Noir was rated as more aromatic than the Garnacha [MPinot (SD) = 5.68 (1.82), MGarnacha (SD) = 4.69 (1.51), p < 0.0005]. In terms of liking, the Pinot Noir was liked significantly more than the Garnacha [MPinot (SD) = 5.38 (1.68), MGarnacha (SD) = 4.90 (1.44), p = 0.013].
Furthermore, there were significant interaction effects between wine type and sound condition for the ratings of aroma intensity [F (2, 96) = 3.48, p = 0.035, η2 = 0.07] and body [F (2, 96) = 3.84, p = 0.025, η2 = 0.074]. For aroma intensity, the interaction was driven by the fact that the Garnacha was rated to be significantly more aromatically intense while participants were listening to the 100 Hz low tone rather than the 1000 Hz higher tone [M100Hz (SD) = 5.14 (1.64), M1000Hz (SD) = 4.38 (1.43), p = 0.007]. No such differences were found for the Pinot Noir. In terms of body, wines were rated as being significantly fuller while listening to the 100 Hz low tone as compared to silence for the Pinot Noir [M100Hz (SD) = 5.54 (1.64), MSilence (SD) = 4.96 (1.59), p = 0.021]. However, the same effect was not observed for the Garnacha, in which there was little change between the different conditions.
While there was no main effect of wine familiarity, we did observe a significant interaction effect between sound condition and wine familiarity when it came to ratings of acidity [F (2, 96) = 3.84, p = 0.025, η2 = 0.074]. As Figure 2 illustrates, this interaction was driven by the difference in acidity ratings in the high frequency (1000 Hz) condition, with those with high wine familiarity perceiving the wines as much more acidic compared to those with low wine familiarity [MLow (SD) = 5.21 (1.57), MHigh (SD) = 6.40 (1.61), p = 0.004].
Non-significant trends suggested that those with high wine familiarity rated the wines to be the most acidic in the 1000 Hz condition out of all sound conditions, whereas those with low wine familiarity rated the wines to be the least acidic in the 1000 Hz condition out of all sound conditions.
The RATA analysis revealed that the Garnacha and Pinot Noir wines were perceived as similar in most attributes. Relevant to the main study, this included comparable perceptions of acidity, body and viscosity in both wines. However, there were six significant differences (p < 0.05) discovered. The Garnacha was rated higher in chocolate aroma, chocolate and dark fruit flavours, and astringency, while the Pinot Noir was rated higher in red fruit and savoury notes. Liking was similar across all six wines in the RATA study, with the mean liking rating of the Garnacha 5.25 and the Pinot Noir 5.12 on a 9-point hedonic scale. A further visual examination of a biplot of the Principal Component Analysis, which explained 82.08% of the variation of the data in the sensory space in the first two components (data not shown), of all the six wines in the RATA study showed the Garnacha positioned in the sensory space towards the left of the plot, which contained attributes such as confectionary and jammy flavours, while the Pinot Noir was situated to the right, closer to savoury and earthy/dusty flavours.
TECHICAL DISCUSSION
It is proposed that a difference in the complexity and wine chemistry between the Pinot Noir and Garnacha could have resulted in the two wines being differently rated for body and aromatic intensity the 100 Hz tone. As greater viscosity tends to be associated with a lessened perception of intensity of flavours and volatile components in model solutions, depending on the compounds,[1] that there were significant pairwise correlations between ratings of aroma intensity, acidity, body, and liking, could suggest that people might be transferring high intensity across attributes. This could also be due to learned associations that link high levels of attributes, as demonstrated by novice wine drinkers associating fuller bodied wines with greater flavor intensity and vice versa.[2] However, no research examining these specific taste–aroma interactions in wine has been published to date. Furthermore, it should be noted that as the area of taste-aroma interactions is a complex one,[3] and study into cross-modal correspondences between sound and wine is still at a nascent stage, more research is required into how the different attributes of both wine characters, and wine and sound interact.
When the data from the participants were segmented into groups, it was revealed that the more experienced tasters performed in a far more consistent manner in their judgments than did the beginners. This could suggest that the methodology of the experiment was better suited to those with more wine experience. It could be hypothesised that experts might be more used to conceptualising levels of the attributes selected in the study, rating these on a scale. This was perhaps more of a challenge to novices (see discussion of rating scales by Meilgaard et al., 2006).[4] Regular wine drinkers could also be more confident in their interactions with wine, given that it is a regular part of their lifestyle, in contrast with those for whom wine constituted an occasional drink. It should be noted that some less experienced tasters among participants voiced concern to the invigilator with regards to their accurate use of the scale. It could be that any changes needed to be more overt for them to be rated more decisively.
Issues with rating on scales could be overcome, and the results clarified, by running an additional study using a different methodology in order to further explore possible correlations between bass and body through, for example, a matching exercise. In this case, the participants would be given three wines of varying body (light, medium, and full), but with their other main parameters (i.e., acidity, tannins) as similar as possible, to match with three tones of varying pitch (low, medium, and high). In this simple matching exercise any correlations between pitch and body should become more apparent.
When interpreting the data, it is also worth considering the relationship between pitch and body in light of current discussions regarding the relative versus absolute nature of the cross-modal correspondences.[5] Past research suggests that most cross-modal correspondences involving the metathetic pitch dimension are relative.[6] The completely randomised presentation of the tones with the two wines in this study could mean that for some participants the same, rather than a contrasting, tone was presented sequentially. The first tone presented might also have been difficult to categorize as being either high or low. Brunetti et al.’s (2018) investigation highlighted the flexibility and sequential influence of stimuli presentation in relation to the absolute versus relative nature of pitch-size correspondences.[7] From this, it could be deduced that the randomisation in this study could have lessened the correspondences discovered. To overcome these possible effects, future experiments could expose participants to the range of tones used before the start of the experiment. Using a tone higher in frequency than 1000 Hz might presumably also assist in greater differentiation between the tones by participants.
REFERENCES
Belkin, K., R. Martin, S. E. Kemp, and A. N. Gilbert. ‘Auditory Pitch as a Perceptual Analogue to Odor Quality’. Psychological Science 8, no. 4 (1997): 340–42. https://doi.org/10.1111/j.1467-9280.1997.tb00450.x.
Ben-Artzi, Elisheva, and Lawrence E Marks. ‘Visual-Auditory Interaction in Speeded Classification: Role of Stimulus Difference’. Perception & Psychophysics 57, no. 8 (1995): 1151–62.
Brunetti, Riccardo, Allegra Indraccolo, Claudia Del Gatto, Charles Spence, and Valerio Santangelo. ‘Are Crossmodal Correspondences Relative or Absolute? Sequential Effects on Speeded Classification’. Attention, Perception, & Psychophysics 80, no. 2 (2018): 527–34.
Chiou, Rocco, and Anina N Rich. ‘Cross-Modality Correspondence between Pitch and Spatial Location Modulates Attentional Orienting’. Perception 41, no. 3 (2012): 339–53.
Crisinel, A. S., and C. Spence. ‘A Fruity Note: Crossmodal Associations between Odors and Musical Notes’. Chemical Senses 37 (2012): 151–58. https://doi.org/10.1093/chemse/bjr085.
Lavie, Nilli. ‘Distracted and Confused?: Selective Attention under Load’. Trends in Cognitive Sciences 9, no. 2 (2005): 75–82. https://doi.org/10.1016/j.tics.2004.12.004.
Meilgaard, M. C., G. V. Civille, and B. T. Carr. Sensory Evaluation Techniques. Boca Raton, UK: CRC Press, 2006.
Murphy, Sandra, Charles Spence, and Polly Dalton. ‘Auditory Perceptual Load: A Review’. Annual Reviews 2017 352 (1 September 2017): 40–48. https://doi.org/10.1016/j.heares.2017.02.005.
Niimi, Jun, Lukas Danner, Luxing Li, Hélène Bossan, and Susan E.P. Bastian. ‘Wine Consumers’ Subjective Responses to Wine Mouthfeel and Understanding of Wine Body’. Food Research International 99 (1 September 2017): 115–22. https://doi.org/10.1016/j.foodres.2017.05.015.
Paravisini, Laurianne, and Elisabeth Guichard. ‘Interactions between Aroma Compounds and Food Matrix’. In Flavour: From Food to Perception, edited by Elisabeth Guichard, Christian Salles, Martine Morzel, and Anne-Marie Le Bon, 208–34. John Wiley & Sons, 2017.
Spence, C., and Q. J. Wang. ‘What Does the Term “Complexity” Mean in the World of Wine?’ International Journal of Gastronomy & Food Science 14 (2018): 45–54.
Spence, Charles. ‘On the Relative Nature of (Pitch-Based) Crossmodal Correspondences’. Multisensory Research 32, no. 3 (2019): 235–65.
Spence, Charles, and Bettina Piqueras-Fiszman. ‘Oral-Somatosensory Contributions to Flavor Perception and the Appreciation of Food and Drink’. In Multisensory Flavor Perception: From Fundamental Neuroscience Through to the Marketplace, 59–79. Duxford, CB: Elsevier, 2016.
Tournier, Carole, Claire Sulmont-Rossé, and Elisabeth Guichard. ‘Flavour Perception: Aroma, Taste and Texture Interactions’. Food 1, no. 2 (2007): 246–57.
Turoman, Nora, Carlos Velasco, Yi-Chuan Chen, Pi-Chun Huang, and Charles Spence. ‘Symmetry and Its Role in the Crossmodal Correspondence between Shape and Taste’. Attention, Perception, & Psychophysics 80, no. 3 (1 April 2018): 738–51. https://doi.org/10.3758/s13414-017-1463-x.
Velasco, Carlos, Andy T. Woods, Ophelia Deroy, and Charles Spence. ‘Hedonic Mediation of the Crossmodal Correspondence between Taste and Shape’. Food Quality and Preference 41 (1 April 2015): 151–58. https://doi.org/10.1016/j.foodqual.2014.11.010.
Wang, Qian Janice, Sheila Wang, and Charles Spence. ‘“Turn Up the Taste”: Assessing the Role of Taste Intensity and Emotion in Mediating Crossmodal Correspondences between Basic Tastes and Pitch’. Chemical Senses 41, no. 4 (2016): 345–56. https://doi.org/10.1093/chemse/bjw007.
[1] Tournier, Sulmont-Rossé, and Guichard, ‘Flavour Perception: Aroma, Taste and Texture Interactions’; Spence and Piqueras-Fiszman, ‘Oral-Somatosensory Contributions to Flavor Perception and the Appreciation of Food and Drink’.
[2] Niimi et al., ‘Wine Consumers’ Subjective Responses to Wine Mouthfeel and Understanding of Wine Body’.
[3] Paravisini and Guichard, ‘Interactions between Aroma Compounds and Food Matrix’; Tournier, Sulmont-Rossé, and Guichard, ‘Flavour Perception: Aroma, Taste and Texture Interactions’.
[4] Meilgaard, Civille, and Carr, Sensory Evaluation Techniques, 55–60.
[5] Brunetti et al., ‘Are Crossmodal Correspondences Relative or Absolute? Sequential Effects on Speeded Classification’; Spence, ‘On the Relative Nature of (Pitch-Based) Crossmodal Correspondences’.
[6] Ben-Artzi and Marks, ‘Visual-Auditory Interaction in Speeded Classification: Role of Stimulus Difference’; Brunetti et al., ‘Are Crossmodal Correspondences Relative or Absolute? Sequential Effects on Speeded Classification’; Chiou and Rich, ‘Cross-Modality Correspondence between Pitch and Spatial Location Modulates Attentional Orienting’.
[7] Brunetti et al., ‘Are Crossmodal Correspondences Relative or Absolute? Sequential Effects on Speeded Classification’.