METHODOLOGY FOR ESTABLISHING CORRELATIONS BETWEEN SENSORY AND INSTRUMENTAL ATTRIBUTES OF CRISP PRODUCTS WITH SLIGHT TEXTURE DIFFERENCES

As for any manufactured product, food processing requires efficient qu ality assurance procedures, in particular to cheek the constancy of th eir sensorial properties. The eventual aim of this work is to develop on-line methods for texture assessment of cooked cereal products with low moisture content. This study proposes an ad-hoc methodology for es tablishing correlations between sensorial and instrumental texture mea surements in the case of similar crisp products from industrial origin (small toasted half breads). A classic sensory profiling methodology was used to generate specific texture attributes, as no pre-establishe d list was suitable. The definition of the selected attributes (6) inc orporates a description of the associated test procedure. Simultaneous ly, some simple characteristics (dimensions, hulk density, moisture co ntent) were determined, and mechanical measurements were achieved by c ompression between plates with an Instron Universal Testing machine. T he force at 70% compression was chosen. To take into account the tortu ousness of the curve, two parameters are proposed: the length of the c urve and its specific length (length divided by the area under the cur ve). Analysis of variance shows that texture attributes and mechanical parameters may discriminate between the products. Canonical variate a nalysis of sensory data demonstrates that texture perception is, in th is case, characterized by three dimensions mainly associated with hard ness, sound intensity and texture change during chewing and mixing wit h saliva. Two physical parameters give a good prediction of the first two sensory factorial components: the force at 70% compression and the specific length of the curve (tortuousness), but it was not possible to predict the third sensory dimension from the selected physical para meters. These correlations have been examined using the results genera ted by a new sensory panel working with the previously defined texture attributes.