Bilingual panels: a tool to evaluate the role of language in descriptive tasks

Cross cultural studies relating sensory profiles established in different countries suggest that part of the observed cultural differences might be due to language. Indeed, participants in cross cultural studies often have different food experiences as well as different languages and so the effects of language and cultures cannot be easily separated. A way to separate these effects is to work with bilingual assessors. Madagascar is a good candidate to perform such a study as two languages coexist in this country: Malagasy and French. Four flash profiles were performed to evaluate six samples of Moringa oleifera leaf powders: two in French and two in Malagasy. The descriptors generated by the four panels were classified in six categories: appearance, touch, odor, flavour, texture and sound; the frequencies of terms of each category were computed for each language condition. Chi-square tests failed to show any difference between the French and Malagasy groups. In both languages the appearance and taste terms were more frequent than the other terms, followed by odor, touch, texture and sound terms. However a qualitative analysis showed that the preciseness of the terms in a given category depended upon the language. Malagasy terminology was richer than the French one for several categories. Multiple factor analyses were then carried out to evaluate whether these qualitative differences in language led to different product maps. Higher between (0.85) than within language RV coefficients (0.78) were observed for the taste / odor descriptors categories. For the other descriptors no effect of language was observed on the positioning of the powders. This effect of language on the odor/taste category can be put in perspective with previous results showing a strong effect of culture on aroma descriptors (e.g. Tu el al., 2010) and suggest that such an influence might be due to language availability effects.