Coffee is a universally popular product, with very distinct sensory characteristics and a stimulant effect due to the caffeine content. While many people consume more than one coffee per day, it is common to have a favourite coffee during the day, associated with a specific context. Preferring a context to another can provide valuable information about the most favoured aspects of the coffee experience, that may reflect individual physiological differences that affect food/beverages perception. Open-ended questions can be useful to collect this information because they allow subjects to freely describe their experiences. This study aims at: 1) exploring consumers' preferred coffee consumption contexts and their relationship with individual physiological characteristics; 2) developing an approach to facilitate and make more efficient open-ended question analysis, using together semiotic and statistical tools. 134 regular coffee consumers were characterized by the number of fungiform papillae (FP), the PROP taster status and the rate of caffeine metabolism (CMI) and were asked to describe the moment of the day in which they prefer to have coffee. The results showed that the preferred context for having coffee differed according to the physiological variables of CMI, FP and, to a less extent, PROP status. Subjects with lower CMI focussed more on the social features of the experience, while higher CMI found more rewarding the sensory properties of coffee. Individuals with lower FP number found more rewarding the product sensory characteristics than individuals with higher FP number, who were more focussed on the stimulating effect of coffee. Both PROP non tasters and supertasters tended to not prefer a context where the focus was on sensory properties (e.g. tasting) and rather preferred contexts where the focus was on the social aspects related to the coffee consumption. The study showed that open-ended questions can be useful to investigate consumer perception. In addition, combined use of semiotic and statistical tools can give valuable results and can help to optimise the text analysis procedure.
